PageSourceSearch

https://www.sciopen.com/home/news_article_page

html sciopen.com collected 2026-10-01 10:22:21 UTC 91,453 bytes, 763 lines download raw bytes

1<!doctype html>
2<html lang="en">
3  <head>
4    <meta charset="UTF-8" />
5    <title>Featured Articles List - SciOpen</title>
6    <meta name="keywords" content="" />
7    <meta name="description" content="" />
8    <meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1.0, user-scalable=0" />
9    <meta http-equiv="X-UA-Compatible" content="IE=edge,Chrome=1" />
10    
10<script>
11      var _hmt = _hmt || [];
12      (function () {
13        var hm = document.createElement('script');
14        hm.src = 'https://hm.baidu.com/hm.js?459d6fc0106b99052c6f753275cafeea';
15        var s = document.getElementsByTagName('script')[0];
16        s.parentNode.insertBefore(hm, s);
17      })();
18    </script>
vendor: 68 bytes, lines 18-19
18
19    <script async src="https://www.googletagmanager.com/gtag/js?id=
19G-5SWG8SFZJW
vendor: 16 bytes, lines 19-20
19"></script>
20    
20<script>
21      
vendor: 165 bytes, lines 21-29
21window.dataLayer = window.dataLayer || [];
22
23      function gtag() {
24        dataLayer.push(arguments);
25      }
26
27      gtag('js', new Date());
28
29      gtag('config', '
29G-5SWG8SFZJW
vendor: 8 bytes, lines 29-30
29');
30    
30</script>
30
31    <link rel="icon" type="image/x-icon" href="/favicon.ico" />
32    <link rel="stylesheet" href="/assets/bootstrap/css/bootstrap.min.css" />
33    <link rel="stylesheet" href="/assets/bootstrap/css/bootstrap-theme.min.css" />
34    
34<script src="/assets/js/jquery.min.js"></script>
34
35    <link rel="stylesheet" type="text/css" href="/assets/css/main-d930173d972c4e6f30aa8aa72f899105.css" />
36    <link rel="stylesheet" href="/assets/css/home-fdaf1cc12a2c64a09954873444e7494e.css" />
37    <link rel="stylesheet" type="text/css" href="/assets/css/codehim-dropdown-22ed6710231b6e74611f4df430379e32.css" />
38    <link rel="stylesheet" href="/assets/css/zxx-55251ff97a2ba9343684a7e8590b84c9.css" />
39    <link rel="stylesheet" href="/assets/css/lxf-8bbabe0a77189752d03805cedbf8f18f.css" />
40    <link rel="stylesheet" type="text/css" href="/assets/css/yxpage-fd559a07a4aa9e80f468f33499d38ad9.css" />
41    <link rel="stylesheet" type="text/css" href="/assets/css/v4-news-list-4e80e8da3a7e911c59395ee4553b61b8.css" />
42  </head>
43  <body class="v4-body">
44    <div class="v4-news-page v4-news-page-news">
45      <div id="header">
46        <div id="v4-header" class="v4-header" @mouseleave="hoverMenu('')">
47  <link rel="stylesheet" href="/assets/js/element-ui/lib/theme-chalk/index.css" />
48  <div class="header-left">
49    <a href="/home">
50      <img class="header-logo" src="/assets/img/home/logo.png" />
51    </a>
52    <div class="v4-menu">
53      <div class="menu-item" :class="{active: currentMenu === 'Subjects'}" @click="hoverMenu('Subjects','click')">
54        All Journals
55        <span><i class="el-icon-arrow-down"></i></span>
56      </div>
57      <div class="menu-item" @click="goToMenu('v4-webinar')">Webinars</div>
58      <div class="menu-item" @click="goToMenu('v4-news')">News</div>
59      <div class="menu-item" @click="goToMenu('v4-services')">Services</div>
60      <div class="menu-item" :class="{active: currentMenu === 'About Us'}" @click="hoverMenu('About Us','click')">
61        About Us
62        <span><i class="el-icon-arrow-down"></i></span>
63      </div>
64    </div>
65  </div>
66  <div class="header-right">
67    
68    <a href="/home" class="mobile-header-logo">
69      <img src="/assets/img/home/logo.png" />
70    </a>
71    <div class="search-box" v-if="showSearchBtn" @click="showSearch('pc')">
72      <i class="el-icon-search"></i>
73      Search
74    </div>
75    <a class="user-login-item" href="/user/user/login_forward">Login</a>
76    <span class="mobile-hr"> | </span>
77    <a class="user-sign-item" href="/user/user/register_forward">Sign up</a>
78    
79    
80  </div>
81  <div class="mobile-menu" id="menu">
82    <div class="menu-icon right" @click="mobileShowMenu()"></div>
83  </div>
84  <div class="modal fade" id="logoff" tabindex="-1" role="dialog" aria-labelledby="myModalLabel">
85    <div class="modal-dialog" role="document">
86      <div class="modal-content">
87        <div class="modal-header">
88          <button type="button" class="close" data-dismiss="modal" aria-label="Close"><span aria-hidden="true">&times;</span></button>
89          <h4 class="modal-title" id="myModalLabel">Logging off</h4>
90        </div>
91        <div class="modal-body">All data generated by the account on the platform will be deleted after logout.</div>
92        <div class="modal-footer">
93          <button type="button" class="btn btn-default" data-dismiss="modal">Close</button>
94          <button type="button" class="btn btn-primary modal-logoff">Save</button>
95        </div>
96      </div>
97    </div>
98  </div>
99  <div class="v4-header-hover-panel" :class="{'show-panel':currentMenu === 'Subjects'}" v-show="currentMenu==='Subjects'" @mouseover="hoverMenu('Subjects')" @mouseout="hoverMenu('')">
100    <div class="hover-left">
101      <div class="left-title">Browse by Subject</div>
102      <div class="left-desc">
103        <p>Search for peer-reviewed journals with full access.</p>
104      </div>
105      <div class="left-abc">
106        <a href="/home/discipline?tab=&id=&language=" target="_blank">Journals A - Z <i class="el-icon-arrow-right"></i></a>
107      </div>
108    </div>
109    <div class="hover-right subject">
110      <a :href="item.href" target="_blank" class="subject-item" v-for="item in subjectList">
111        <div class="item-desc">{{item.name}}</div>
112        <div class="item-abc">{{item.abc}}</div>
113      </a>
114    </div>
115  </div>
116  <div class="v4-header-hover-panel" :class="{'show-panel':currentMenu === 'About Us'}" v-show="currentMenu==='About Us'" @mouseover="hoverMenu('About Us')" @mouseout="hoverMenu('')">
117    <div class="hover-left">
118      <div class="left-title">About Us</div>
119      <div class="left-desc">
120        <p>Discover the SciOpen Platform and Achieve Your Research Goals with Ease.</p>
121      </div>
122    </div>
123    <div class="hover-right">
124      <div class="about-item">
125        <div class="item-panel" v-if="aboutList.length">
126          <div class="item-title">About Us</div>
127          <div class="item-desc">
128            <a class="desc-item" :href="item.href" target="_blank" v-for="item in aboutList">{{item.title}}</a>
129          </div>
130        </div>
131        <div class="item-panel" v-if="publishList.length">
132          <div class="item-title">Publish with Us</div>
133          <div class="item-desc">
134            <a class="desc-item" :href="item.href" target="_blank" v-for="item in publishList">{{item.title}}</a>
135          </div>
136        </div>
137        <div class="item-panel" v-if="supportList.length">
138          <div class="item-title">Support</div>
139          <div class="item-desc">
140            <a class="desc-item" :href="item.href" target="_blank" v-for="item in supportList">{{item.title}}</a>
141          </div>
142        </div>
143      </div>
144    </div>
145  </div>
146  <div class="v4-header-hover-panel search" :class="{'show-panel':searchShow}" v-show="searchShow" @click.stop>
147    <div class="warp search-wrap">
148      <div class="search-wrap-content">
149        <p class="search-title">Search articles, authors, keywords, DOl and etc.</p>
150        <div class="search-d1" :class="journalId?'journal':''">
151          <el-input v-model="searchForm.keyword" clearable placeholder="Enter Search Terms" class="search-d1-i" @keyup.enter.native="!expandStatus ? goSearch(1) : goSearch(2)"></el-input>
152          <el-select v-if="journalId" v-model="searchType" class="item-s1">
153            <el-option :key="1" label="The Platform" :value="1"></el-option>
154            <el-option :key="2" label="This Journal" :value="2"></el-option>
155          </el-select>
156          <el-button v-if="!expandStatus" class="icon-button" type="primary" icon="el-icon-search" @click="goSearch(1)"></el-button>
157        </div>
158        <div class="search-d2" v-if="expandStatus">
159          <div v-for="(item,index) in searchForm.keywordDTO" class="search-d2-item">
160            <el-select v-model="item.tag" class="item-s1">
161              <el-option :key="1" label="AND" :value="1"></el-option>
162              <el-option :key="2" label="OR" :value="2"></el-option>
163              <el-option :key="3" label="NOT" :value="3"></el-option>
164            </el-select>
165            <el-select v-model="item.type" class="item-s2" @change="item.key='';item.key2=''">
166              <el-option v-if="!journalId" label="Journal Title" value="journal"></el-option>
167              <el-option v-if="!journalId" label="ISSN" value="issn"></el-option>
168              <el-option label="Article Title" value="title"></el-option>
169              <el-option label="Author" value="author"></el-option>
170              <el-option label="Keywords" value="keywords"></el-option>
171              <el-option label="DOI" value="doi"></el-option>
172              <el-option label="Abstract" value="abstracted"></el-option>
173            </el-select>
174            <template v-if="item.type !== 'author'">
175              <el-input v-if="item.type === 'doi'" v-model="item.key" clearable :placeholder="`Search in DOI`" class="item-i1"></el-input>
176              <el-input v-else-if="item.type === 'abstracted'" v-model="item.key" clearable :placeholder="`Search in Abstract`" class="item-i1"></el-input>
177              <el-input v-else-if="item.type === 'keywords'" v-model="item.key" clearable :placeholder="`Search in Keywords`" class="item-i1"></el-input>
178              <el-input v-else-if="item.type === 'title'" v-model="item.key" clearable :placeholder="`Search in Article Title`" class="item-i1"></el-input>
179              <el-input v-else-if="item.type === 'issn'" v-model="item.key" clearable :placeholder="`Search in ISSN`" class="item-i1"></el-input>
180              <el-input v-else-if="item.type === 'journal'" v-model="item.key" clearable :placeholder="`Search in Journal Title`" class="item-i1"></el-input>
181              <el-input v-else v-model="item.key" clearable :placeholder="`Search in ${item.type}`" class="item-i1"></el-input>
182            </template>
183            <template v-if="item.type === 'author'">
184              <el-input v-model="item.key" clearable placeholder="First name" class="item-i2" style="width: 100%"></el-input>
185              <el-input v-model="item.key2" clearable placeholder="Last name" class="item-i3" style="width: 100%"></el-input>
186            </template>
187            <span class="add-btn">
188              <img src="/assets/img/index/search-plus.svg" alt="" @click="addInput(index)" />
189              <img v-if="searchForm.keywordDTO.length > 1" src="/assets/img/index/search-reduce.svg" alt="" @click="delInput(index)" />
190            </span>
191          </div>
192          <div class="search-d2-time">
193            <label class="search-title">Published Date</label>
194            <div class="time-range-panel">
195              <el-date-picker class="time-range-1" v-model="searchForm.startTime" type="date" placeholder="Start Time" :picker-options="startTimeOptions"> </el-date-picker>
196              <div class="time-hr">-</div>
197              <el-date-picker class="time-range-2" v-model="searchForm.endTime" type="date" placeholder="End Time" :picker-options="endTimeOptions"> </el-date-picker>
198            </div>
199          </div>
200          <div class="search-d2-btn" :class="journalId ? 'journal' : ''">
201            <span @click="resetInput">Reset</span>
202            <span @click="goSearch(2)">Search</span>
203          </div>
204        </div>
205        <div class="search-d3">
206          <img src="/assets/img/index/search-more.svg" alt="" :class="expandStatus ? 'img-rote' : 'img-unrote'" @click="expandStatus = !expandStatus" />
207          <span @click="expandStatus = !expandStatus">{{expandStatus?'Exit ':''}}Advanced Search</span>
208        </div>
209      </div>
210    </div>
211  </div>
212  <el-drawer class="menu-drawer" :class="{'show-panel':showMobileMenu}" :visible.sync="showMobileMenu" direction="rtl" :show-close="false" :with-header="false" size="290px">
213    <div class="mobile-menu-panel">
214      <div class="menu-item" @click="clickMenu('Subjects')">
215        All Journals <span class="arrow" :class="{active: currentMenu === 'Subjects'}"><i class="el-icon-arrow-right"></i></span>
216      </div>
217      <collapse-transition>
218        <div class="subject-content" v-show="currentMenu === 'Subjects'">
219          <a :href="item.href" class="subject-item" v-for="item in subjectList"> <div class="item-desc">{{item.name}}</div></a>
220          <a class="subject-item" href="/home/discipline?tab=&id=&language="
221            ><div class="item-desc">Journals A - Z <i class="el-icon-arrow-right"></i></div
222          ></a>
223        </div>
224      </collapse-transition>
225      <div class="menu-item" @click="goToMenu('v4-webinar')">Webinars</div>
226      <div class="menu-item" @click="goToMenu('v4-news')">News</div>
227      <div class="menu-item" @click="goToMenu('v4-services')">Services</div>
228      <div class="menu-item" @click="clickMenu('About Us')">
229        About Us <span class="arrow" :class="{active:currentMenu === 'About Us'}"><i class="el-icon-arrow-right"></i></span>
230      </div>
231      <collapse-transition>
232        <div class="about-content" v-show="currentMenu === 'About Us'">
233          <div class="about-item">
234            <div class="item-panel" v-if="aboutList.length">
235              <div class="item-title">About Us</div>
236              <div class="item-desc">
237                <a class="desc-item" :href="item.href" target="_blank" v-for="item in aboutList">{{item.title}}</a>
238              </div>
239            </div>
240            <div class="item-panel" v-if="publishList.length">
241              <div class="item-title">Publish with Us</div>
242              <div class="item-desc">
243                <a class="desc-item" :href="item.href" target="_blank" v-for="item in publishList">{{item.title}}</a>
244              </div>
245            </div>
246            <div class="item-panel" v-if="supportList.length">
247              <div class="item-title">Support</div>
248              <div class="item-desc">
249                <a class="desc-item" :href="item.href" target="_blank" v-for="item in supportList">{{item.title}}</a>
250              </div>
251            </div>
252          </div>
253        </div>
254      </collapse-transition>
255      <div class="menu-item" v-if="showSearchBtn" @click="showSearch(&#39;mobile&#39;)">Advanced Search</div>
256    </div>
257  </el-drawer>
258  
258<script>
259    var user = null;
260    var journalId = "";
261    var issn = "";
262  </script>
262
263  
263<script src="/assets/js/vue.min.js"></script>
263
264  
264<script src="/assets/js/axios.min.js"></script>
264
265  
265<script src="/assets/js/element-ui/lib/index.js"></script>
265
266  
266<script src="/assets/js/element-ui/lib/umd/locale/en.js"></script>
266
267  
267<script src="/assets/js/collapse-transition.js"></script>
267
268  
268<script src="/assets/js/dayjs/dayjs.min.js"></script>
268
269  
269<script src="/assets/js/v4-header.js"></script>
269
270</div>
271      </div>
272      <div class="bread-menu">
273        <span><a href="/home">Home</a></span>
274        <span> <i class="glyphicon glyphicon-menu-right"></i> </span>
275        <span> Featured Articles List </span>
276      </div>
277      <div class="v4-news-cc">
278        <div class="v4-news-container">
279          <div class="v4-news-item">
280            <div class="v4-news-content v4-news-content-article">
281              <div class="v4-news-cl">
282                <div class="v4-news-qikan">
283                  <a href="/journal/3135-3169">iFuture</a>
284                </div>
285                <h2 class="v4-news-title">
286                  <a href="/home/news_article_detail?id=2105179976597270530">Solving the increasing complexity of AI problem-solving one bite at a time</a>
287                </h2>
288                <div class="v4-news-dsc" style="height: 60px">
289                  <a href="/home/news_article_detail?id=2105179976597270530">“How do you eat an elephant? One bite at a time,” goes the proverb that dates back to the early 1920s. Turns out the same idea may apply to advancing artificial intelligence (AI), according to a team from the Institute for Interdisciplinary Information Sciences (IIIS) at Tsinghua University.
290The researchers developed a framework they call the “calculus of intelligence,” or COIN, capable of breaking down the increasingly complex problems agentic AI — the systems expected to operate with minimal human oversight to do things like monitor for cybersecurity breaches, write code and much more — is tasked with solving. They published their approach on July 17 in iFuture.
291“The calculus of intelligence is a mathematical framework for decomposing a complex task into smaller, well-defined subtasks that are simple enough to solve, and then combining their solutions into a coherent whole,” said Yang Yuan, associate professor and corresponding author on the paper. “Just as classical calculus can calculate the area under a complicated curve by dividing it into tiny pieces and adding them up, this framework provides a way to build and understand complex systems through the step-by-step composition of numerous simple components.”
292The framework translates splices up the overarching AI goal into smaller tasks within a logic system called a Grothendieck topos, which Yuan described as a mathematical rulebook for facilitating the coordination of building something as complex as a large airplane.
293“It must be divided into many local components — such as the wings, engines and control systems — which are designed by different teams,” Yuan said, explaining that these limited perspectives are the local views, with each team only seeing the information relevant to its own task rather than every detail of the entire airplane, while still retaining the shared information needed to connect its component to the rest of the system. “This allows different teams to work independently while ensuring that the components they produce match at their shared boundaries.”
294The Grothendieck topos expresses the structure of local design, shared interfaces and global coordination in mathematical language. It specifies what each local component can see, what requirements it must meet and how different components must agree where they overlap. According to Yuan, COIN takes the next step by providing the rules for decomposing tasks and reliably recomposing local solutions that comply with the larger system’s rulebook.
295“Intelligence resides in structure, and structure can be decomposed, learned and recompose,” Yuan said, noting that complexity does not mean incomprehensibility. “Many systems that appear overwhelmingly complex may simply be waiting for the right decomposition. A good structure can turn a difficult global problem into a collection of clearly bounded local problems that can be analyzed step by step. Through appropriate structural decomposition, many complex tasks can become easier to understand, execute and verify.”
296This structure can also lend itself to making extremely large systems with limited intelligence, Yuan pointed out.
297“The truly transformative future may not be an infinitely powerful individual intelligence — how some may imagine AI — but rather humans using many limited intelligences to construct systems far beyond the scale that any single person or model could independently understand or complete,” Yuan said.&nbsp;
298According to Yuan, COIN is a step in that direction.
299“The broader goal is to develop a common mathematical language for intelligence: One that can describe what models learn, how complex tasks are decomposed and how many limited intelligences can be organized into larger systems,” Yuan said.
300Andrew Chi-Chih Yao, professor and dean of IIIS, co-authored this paper. Yuan and Yao are also affiliated with the Shanghai QiZhi Institute.
301DOI Link:
302https://doi.org/10.26599/IF.2026.9710001
303About iFuture
304iFuture is a premier open-access journal published by Tsinghua University Press on the SciOpen platform, with academic support​ from the Institute for Interdisciplinary Information Sciences at Tsinghua University. Led by Turing Award Laureate Prof. Andrew Chi-Chih Yao​ as Editor-in-Chief, the journal is the core component of the AI Open Alliance. Its core mission is to break through AI’s theoretical bottlenecks and foundational infrastructure.</a>
305                </div>
306                <div class="v4-news-date">
307                  <span>Information Sciences</span>
308                </div>
309              </div>
310              <div class="v4-news-img v4-news-img-l">
311                <a href="/home/news_article_detail?id=2105179976597270530">
312                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-09-30/1156a44e-e4be-4040-b2b2-f17c00694a68.jpg?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=afob99p3V%2BvbwpK7D1KzfraMGvo%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
313                </a>
314              </div>
315            </div>
316          </div>
317          <div class="v4-news-item">
318            <div class="v4-news-content v4-news-content-article">
319              <div class="v4-news-cl">
320                <div class="v4-news-qikan">
321                  <a href="/journal/2097-5023">Communications in Transportation Research</a>
322                </div>
323                <h2 class="v4-news-title">
324                  <a href="/home/news_article_detail?id=2102593305624211457">Can Large Language Models Capture Human Risk Preferences?</a>
325                </h2>
326                <div class="v4-news-dsc" style="height: 60px">
327                  <a href="/home/news_article_detail?id=2102593305624211457">Large language models (LLMs) are increasingly used as agents to simulate human behavior, yet their fidelity in complex decision-making under uncertainty remains insufficiently understood. To address this gap, we develop a comparative framework that benchmarks LLM-simulated risk preferences against empirical human behavior. Using demographic profiles from surveys conducted in Sydney, Hong Kong, and Nanjing, we construct role-playing prompts and evaluate three LLMs on abstract lottery-choice tasks. We adopt the classical Constant Relative Risk Aversion (CRRA) framework as a domain-neutral “standard ruler” to compare risk attitudes. The analysis yields three main findings. First, off-the-shelf LLMs do not exhibit a universal risk profile: the two GPT models are more risk-averse than human benchmarks, whereas Gemini is more risk-seeking. Second, prompt language systematically affects simulated risk attitudes, with English-to-Chinese switching inducing a more conservative shift in most cases. Third, LLMs do not reliably reproduce the empirical heterogeneity of human risk preferences, tending either to generate overly concentrated distributions or unrealistically large dispersion. Taken together, these findings show that off-the-shelf LLMs remain vulnerable to model-family-specific miscalibration, language-sensitive distortions, and failures in distributional fidelity. Rigorous empirical calibration is therefore necessary before off-the-shelf LLMs can be reliably deployed in computational social science and choice modeling.
328The team published their study in Communications in Transportation Research (https://doi.org/10.26599/COMMTR.2026.9640025).
329Our findings highlight an important limitation of using off-the-shelf LLMs as tools for behavioral prediction. Since different model families exhibit different baseline calibration biases, the choice of model can materially affect the inferred pattern of public risk preferences. In practice, one model family may overstate conservatism, whereas another may overstate willingness to accept risk. Without empirical calibration, such biases can distort inference and lead researchers to draw policy conclusions that do not accurately reflect observed human behavior.
330This limitation is particularly consequential in transportation research, where risk perception is central to decision-making under uncertainty. Travel behavior frequently involves probabilistic trade-offs, including route choice under unreliable travel times, mode switching during service disruptions, and the adoption of emerging mobility technologies under safety and performance uncertainty. If the underlying risk preference parameters are systematically miscalibrated, demand forecasts, welfare evaluation, and policy design may all be bi
330ased. For example, using uncalibrated LLM-generated data to infer willingness to adopt safety-critical systems such as autonomous vehicles or low-altitude mobility services could yield either overly conservative or overly optimistic projections, depending on the model family used.
331The multilingual results add a further layer of caution. In linguistically diverse settings, prompt language is not a neutral implementation choice: it can systematically perturb the behavioral calibration of the model. This is especially relevant for transportation systems serving multilingual populations, where researchers may be tempted to use native-language prompting as a straightforward way to improve realism. Our results suggest that such an assumption is unwarranted unless the model has first been validated against human benchmarks in the relevant linguistic context.
332&nbsp;
333Overall, the implications of this study are methodological as much as substantive. Off-the-shelf LLMs should not be treated as direct substitutes for human respondents in risk-sensitive behavioral applications. Instead, they should be regarded as tools whose outputs require domain-specific and language-sensitive calibration. Rigorous validation against human ground truth remains a necessary prerequisite for deploying these models in transportation and other cross-cultural social science settings.
334&nbsp;
335DOI Link:
336https://doi.org/10.26599/COMMTR.2026.9640025
337&nbsp;
338About Communications in Transportation Research
339Communications in Transportation Research was launched in 2021, with academic support provided by Tsinghua University and China Intelligent Transportation Systems Association. The Editors-in-Chief are Professor Xiaobo Qu, a member of the Academia Europaea from Tsinghua University, and Professor Xiaopeng (Shaw) Li from University of Wisconsin–Madison. The journal mainly publishes high-quality, original research and review articles that are of significant importance to emerging transportation systems, aiming to serve as an international platform for showcasing and exchanging innovative achievements in transportation and related fields, fostering academic exchange and development between China and the global community.
340It has been indexed in SCIE, SSCI, Ei Compendex, Scopus, CSTPCD, CSCD, OAJ, DOAJ, TRID and other databases. It was selected as Q1 Top Journal in the Engineering and Technology category of the Chinese Academy of Sciences (CAS) Journal Ranking List. In 2022, it was selected as a High-Starting-Point new journal project of the “China Science and Technology Journal Excellence Action Plan”. In 2024, it was selected as the Support the Development Project of “High-Level International Scientific and Technological Journals”. The same year, it was also chosen as an English Journal Tier Project of the “China Science and Technology Journal Excellence Action Plan Phase Ⅱ”. In 2024, it received the first impact factor (2023 IF) of 12.5, ranking Top1 (1/58, Q1) among all journals in "TRANSPORTATION" category. In 2026, its 2025 IF was announced as 12.7, maintaining the Top1 position (1/66, Q1) in the same category.
341From Volume 6 (2026), Communications in Transportation Research will be published by Tsinghua University Press on the SciOpen platform with the official journal website at https://www.sciopen.com/journal/2097-5023. We kindly request that all new manuscript submissions be made through the journal’s submission system at https://mc03.manuscriptcentral.com/commtr. For any submission-related inquiries, please contact the Editorial Office at [email protected].</a>
342                </div>
343                <div class="v4-news-date">
344                  <span>Physical Sciences and Engineering</span>
345                </div>
346              </div>
347              <div class="v4-news-img v4-news-img-l">
348                <a href="/home/news_article_detail?id=2102593305624211457">
349                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-09-23/acb462a8-0725-4334-adcf-bc1a6a71b69b.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=55l%2FqvY%2Bs6JLFgxZ%2BOVuVNS%2FZbY%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVV
349hwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
350                </a>
351              </div>
352            </div>
353          </div>
354          <div class="v4-news-item">
355            <div class="v4-news-content v4-news-content-article">
356              <div class="v4-news-cl">
357                <div class="v4-news-qikan">
358                  <a href="/journal/3006-9203">Energy and Climate Management</a>
359                </div>
360                <h2 class="v4-news-title">
361                  <a href="/home/news_article_detail?id=2101870036583477250">Can China maintain power supply security while phasing down coal?</a>
362                </h2>
363                <div class="v4-news-dsc" style="height: 60px">
364                  <a href="/home/news_article_detail?id=2101870036583477250">China’s renewable power expansion is reshaping the country’s electricity system and raising a question: how can China maintain power supply security while reducing its reliance on coal?
365Wind and solar power are variable by nature, and their rising share changes the role of conventional power plants. As noted in the paper, variable renewable energy reached 18.2% of China’s power generation in 2024 and is expected to grow to 65%-70% by 2060. As this transition continues, coal plants may operate fewer hours and face weaker revenues, while the system requires more flexibility to balance fluctuating renewable output.
366A research team led by Tsinghua University examined this challenge using causal loop diagrams to trace feedback mechanisms among renewable deployment, storage investment, market price signals, capacity remuneration, and coal retirement decisions.
367The article was published in Energy and Climate Management on May 25, 2026.
368According to the study, China’s current coal capacity payment mechanism may help stabilize coal plant revenues, but it also has potential drawbacks. Because the mechanism mainly supports incumbent coal and gas capacity, it may exclude emerging flexible resources such as battery storage, demand-side response, distributed resources, and virtual power plants. The paper also points out that administratively determined payments may create risks of overpayment, weaker innovation incentives, and delayed retirement of inefficient coal assets.
369The study contrasts this approach with the United Kingdom’s capacity market, which uses competitive auctions to procure reliable capacity from multiple technologies. The UK experience suggests that market-based capacity mechanisms can help reveal capacity prices and support a more diverse resource mix. At the same time, the paper notes a design challenge: if capacity markets treat all resources similarly without considering duration, short-duration batteries may be favored even when longer-duration resources are needed during prolonged system stress events.
370As noted in the paper, battery energy storage systems can absorb surplus renewable electricity and discharge it when demand rises, helping reduce curtailment and improve system flexibility. However, China’s earlier energy storage mandates produced mixed effects. Although these policies encouraged physical deployment, the average utilization rate of covered storage capacity was only 9% in 2023. One reason highlighted by the study is that storage assets often lacked sufficient market access and revenue opportunities. In the first half of 2023, thermal generators received 91.4% of ancillary service market revenues.
371The paper suggests that China could deepen spot market reforms, adjust overly restrictive price limits, and expand competitive ancillary service markets so storage can earn revenues from arbitrage, frequency regulation, capacity mechanisms, and other services. Rather than requiring each renewable project to self-balance with its own storage, the study points to the value of treating flexibility as a system-wide resource.
372The paper also discusses long-duration energy storage as an important option for maintaining supply security. It notes that China may require more than 700 GW of long-duration storage c
372apacity in a fully decarbonized power system. Strategic reserves are presented as another possible transitional tool, under which some retiring coal units could be kept outside the regular electricity market and called upon only during system stress events.
373The study suggests piloting capacity markets, creating broader revenue channels for storage, supporting long-duration storage, strengthening carbon pricing through an emissions cap and price floor, and using strategic reserves where older coal fleets may still be relevant for reliability.
374The research was primarily conducted by Ying Zhou (first author), Jian&nbsp;Han (corresponding author) and Da Zhang (corresponding author).
375This work was supported by the National Natural Science Foundation of China (Nos. 72401160 and 72504282), the International Science and Technology Cooperation Project (No. 20253000014), the China Three Gorges Corporation Research Project (No. 202303160), and the Sichuan Province Science and Technology Innovation Cooperation Project for Hong Kong, Macao and Taiwan (No. 2025YFHZ0226).
376&nbsp;
377DOI Link:
378https://doi.org/10.26599/ECM.2026.9400033
379&nbsp;
380About the Authors:
381Ying Zhou conducted her postdoctoral research at the Institute of Energy, Environment and Economy, Tsinghua University, China. She is currently a faculty member in the School of Management, China University of Mining and Technology-Beijing. Her research focuses on energy and power system decarbonization, renewable electricity markets, electricity market design, and climate policy evaluation.
382Jian Han conducted his postdoctoral research at the Institute of Energy, Environment and Economy, Tsinghua University, China. He is focusing on energy economics, electricity markets, and agent-based modeling.
383Da Zhang is Assistant Director of the Institute of Energy, Environment and Economy at Tsinghua University, Tenured Associate Professor, and Ph.D. Supervisor. He also serves as Deputy Director of the Tsinghua-Three Gorges Climate and Low-Carbon Center. His research focuses on energy economics, climate policy, low-carbon energy transition, integrated assessment modeling, and electricity market design.</a>
384                </div>
385                <div class="v4-news-date">
386                  <span>Humanities and Social Sciences</span>
387                </div>
388              </div>
389              <div class="v4-news-img v4-news-img-l">
390                <a href="/home/news_article_detail?id=2101870036583477250">
391                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-09-21/32340b97-1715-4cb7-9325-492411f15068.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=PDFu%2FMNY7PZMXWGspTqSdj3wexw%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
392                </a>
393              </div>
394            </div>
395          </div>
396          <div class="v4-news-item">
397            <div class="v4-news-content v4-news-content-article">
398              <div class="v4-news-cl">
399                <div class="v4-news-qikan">
400                  <a href="/journal/2097-5740">Environmental Chemistry and Safety</a>
401                </div>
402                <h2 class="v4-news-title">
403                  <a href="/home/news_article_detail?id=2097581545729060866">Research team combines machine learning with environmental technology to achieve healthier soil</a>
404                </h2>
405                <div class="v4-news-dsc" style="height: 60px">
406                  <a href="/home/news_article_detail?id=2097581545729060866">Scientists have environmental recycling processes that convert farm and forestry waste and polluted soil into useful energy and other useful materials. However, it has been very challenging to precisely control these processes. A research team is using machine learning to regulate the environmental processes, so they are more effective and predictable. Their work provides a technical path for achieving precise, intelligent, and sustainable remediation of polluted soil.
407Their research was published in the journal Environmental Chemistry and Safety on July 1, 2026.
408Co-pyrolysis technology is an environmental recycling process that heats waste products from farms and forests, along with polluted soil, in an oxygen-limited environment. This heating reaction converts the waste products and polluted soil into fuel and other useful materials.
409The challenge that scientists have faced with the co-pyrolysis technology lies with the differences in the raw materials. Because each batch of raw materials behaves a little differently, it is difficult for scientists to accurately control the co-pyrolysis process.
410The research team from Panzhihua University, Tsinghua University, and Harbin Institute of Technology has developed a theoretical framework that combines machine learning with the co-pyrolysis technology to provide a precise and sustainable solution to turn polluted soil to clean soil.
411The world produces over 2 billion tons of farm and forest waste every year. Most of this waste is not used in any way. At the same time, more than one-third of the world’s farmlands cannot be used for farming because the soil is polluted with toxic chemicals. Much of the plant waste is dumped in landfills or burned, which causes air pollution. Cleaning the polluted soil requires methods that harm the soil’s health and create new pollution. These two environmental problems are connected to each other.
412The co-pyrolysis technology offers a solution for farm and forest waste and polluted soil. This process “bakes” the plant waste and polluted soil under oxygen limited conditions. &nbsp;The plant waste and polluted soil react better together than either of them would work on their own. The reaction that occurs cracks and destroys the organic pollutants and traps the heavy metals so they can no longer harm the environment. Finally, the process turns the waste and soil into a high-performance charcoal-like substance called biochar.
413While the co-pyrolysis technology offers a promising solution to these environmental problems, the process has been hard to control. Each batch of plant waste and polluted soil is different, making it hard to predict and control the exact repair effects they can achieve.
414The advances in machine learning technology in recent years give scientists the tools they need to better manage the co-pyrolysis process. Algorithms, such as deep learning, neural networks, and reinforcement learning, help scie
414ntists better understand the complex data related to the co-pyrolysis process. Machine learning tools let them to fix the polluted soil at different sizes, ranging from the molecular level all the way to the larger ecosystem.
415Using machine learning with the co-pyrolysis methods allows scientists to predict the connections across entire process. “Compared to traditional experience-oriented process development models, the core advantage of machine learning driven co-pyrolysis closed-loop design lies in its ability to model and predict high-dimensional nonlinear relationships across the entire chain of raw materials, processes, products, and ecological responses,” said Yuanchuan Ren, Panzhihua University, China.
416Looking to the future, the research team suggests that the co-pyrolysis closed-loop system that recycles waste could have potential for use beyond Earth. “The technical framework and engineering methods of the co-pyrolysis closed-loop system are expanding beyond the scope of Earth's environmental remediation, gradually extending to the utilization of in-situ resources in extraterrestrial celestial bodies and the construction of interstellar human settlements,” said Ren.
417The research team includes Yuanchuan Ren, Yuhang Lin, Xuejun Zhu, Hongbo Han, Shiyong Zhao, Renjie Huang, Tingfeng Su, Yan Guo, Fenghui Wu, Qiang Niu, Dandan Chen from Panzhihua University; Cheng Wang from Tsinghua University; and Nanqi Ren from Harbin Institute of Technology.
418This research is funded by Sichuan Science and Technology Program, the Panzhihua Key Laboratory of Chemical Resource Utilization Open Science Project, Panzhihua Association for Science and Technology Youth Science and Technology Talent Support Project, the Key Laboratory of Dry-hot Valley Characteristic Bio-Resources Development at University of Sichuan Province, and The College Students' Innovation and Entrepreneurship Training Program.
419DOI Link:
420https://doi.org/10.26599/ECS.2026.9600049
421&nbsp;</a>
422                </div>
423                <div class="v4-news-date">
424                  <span>Physical Sciences and Engineering</span>
425                </div>
426              </div>
427              <div class="v4-news-img v4-news-img-l">
428                <a href="/home/news_article_detail?id=2097581545729060866">
429                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-09-09/8f6eb499-4801-4929-9980-df9ca06157b1.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=wb%2FUSZEfaB8r3zhTsefGE3bmJcQ%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
430                </a>
431              </div>
432            </div>
433          </div>
434          <div class="v4-news-item">
435            <div class="v4-news-content v4-news-content-article">
436              <div class="v4-news-cl">
437                <div class="v4-news-qikan">
438                  <a href="/journal/2957-9821">Polyoxometalates</a>
439                </div>
440                <h2 class="v4-news-title">
441                  <a href="/home/news_article_detail?id=2095758555664117762">Spinning up a better understanding of molecules</a>
442                </h2>
443                <div class="v4-news-dsc" style="height: 60px">
444                  <a href="/home/news_article_detail?id=2095758555664117762">One hundred years after the technique was recognized with the Nobel Prize in Chemistry, analytical ultracentrifugation (AUC) is still making waves in the scientific world. Theodor Svedberg developed the approach in the 1920s, using it to characterize tiny gold particles. Now, AUC helps researchers determine specific properties of individual molecules clustered together, such as those used in pharmaceuticals or industrial 
444chemicals.&nbsp; &nbsp;&nbsp;
445A team led by researchers at the University Akron recently published a review of the technology, cataloging use cases focused on how AUC can help solve challenges in characterizing molecular clusters. These clusters comprise clumps of proteins, sugars and other particles that scientists need to understand in deep detail to predict interactions, stability and more of the many diverse solutions encountered across a bevy of areas. They published their work on May 27 in Polyoxometalates.
446“Analytic ultracentrifugation is a powerful and information-rich technique for characterizing the molecular weight, size, shape, dispersity and association behavior of species in their native solution environment,” said corresponding author Tianbo Liu, professor in the Department of Polymer Science at the University of Akron. “With recent instrumentation and data analysis software, AUC enables more accurate and reliable characterization of diverse species in solution systems, including biomacromolecules, like proteins and carbohydrates; colloids, such as nanoparticle suspensions; surfactants assemblies such as micelle; and polymers, including synthetic polymers in solution and polymer-based nanoparticles in dispersion.”
447AUC works by spinning solution samples up to 60,000 rotations per minute, forcing the contents to sediment, or to disperse into concentrated gradients based on size and weight. The approach also integrates optical analytics to monitor and characterize the sedimentation of each molecule, taking AUC a step beyond traditional centrifugation, according to Liu.
448“This overcomes a key limitation of techniques such as light and X-ray scattering, which typically measure the collective properties of solute mixtures,” Liu said, explaining that the other techniques work by assessing how light or X-rays at a sample scatter.
449He noted that AUC also lacks the need for stationary phases or calibration standards like other techniques, such as chromatography, enabling direct, absolute determination of molecular properties. Another benefit of AUC is that the measurements are performed in solution, meaning surface interactions and system perturbations — and the artifacts they may result in — are minimized.
450To demonstrate the wide array of information AUC can glean, the researchers highlighted several examples.
451The first pointed to how AUC reveals the more complicated nature of hydration shells, or the water encapsulating various molecules. Rather than a homogenous layer of oxygen and hydrogen atoms, the water shifts in physical properties and behaviors through the shell, which also varies in thickness.
452The second example focused on determining the distance between components in molecular cluster solutions. According to the researchers, measuring intermolecular distances is key to understanding how charged macroions self-assemble in dilute solution and how changes in these distances correlate with transitions between different macroscopic phases.
453The third example detailed the interactions of molecular clusters with amino acids — the compounds that make up proteins. Understanding how amino acids interact with the surfaces of molecular clusters can help inform understanding of how the clusters will interact with biomolecules. Liu noted that these structures are governed by weak, noncovalent interactions between molecules. Although these forces are difficult to measure directly, AUC can reveal their effects by detecting subtle changes in molecular size, weight and shape in solution.
454“AUC is a powerful technique for investigating complex solution systems,” Liu said. “Owing to their well-defined and uniform size, shape and mass, molecular clusters are particularly well suited for AUC
455measurements.&nbsp;… AUC offers the distinct advantage of resolving different species prior to analysis, enabling direct determination of their individual concentrations.”&nbsp;
456According to the researchers, there is a drawback, however.
457“The technique requires substantial expertise, experience and instrumentation resources,” Liu said. “As a result, the barrier to entry for new users can be relatively high, with a need for extensive training covering both experimental operation and data analysis.”
458Other contributors include Ruixin Li, Xiaohan Xu, Kexing Xiao and Bahareh Afsari, all with the University Akron; and Lake N. Paul, with BioAnalysis, LLC.
459The University of Akron supported this research.
460DOI Link:
461https://doi.org/10.26599/POM.2026.9140132
462&nbsp;About Polyoxometalates
463Polyoxometalates (ISSN 2957-9821) is a peer-reviewed (single-blind), open-access and interdisciplinary journal, sponsored by Tsinghua University. Polyoxometalates publishes original high-quality research papers and significant review articles that focus on cutting-edge advancements in Polyoxometalates, and clusters of metals, metal oxides and chalcogenides. Rapid review to ensure quick publication is a key feature of Polyoxometalates. The journal is indexed by ESCI (IF 2025 = 10.4, Top 3), Scopus (CiteScore 2025 = 17.6, Top 3), Ei Compendex, CAS, and DOAJ. For details about Polyoxometalates, please visit: https://www.sciopen.com/journal/2957-9821.</a>
464                </div>
465                <div class="v4-news-date">
466                  <span>Physical Sciences and Engineering</span>
467                </div>
468              </div>
469              <div class="v4-news-img v4-news-img-l">
470                <a href="/home/news_article_detail?id=2095758555664117762">
471                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-09-04/9876b98e-beda-4627-b68e-18ce4c35c903.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=bTGaK%2F0Ln26AmPlCyrkUHZ4lwZM%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
472                </a>
473              </div>
474            </div>
475          </div>
476          <div class="v4-news-item">
477            <div class="v4-news-content v4-news-content-article">
478              <div class="v4-news-cl">
479                <div class="v4-news-qikan">
480                  <a href="/journal/3005-3315">Energy Materials and Devices</a>
481                </div>
482                <h2 class="v4-news-title">
483                  <a href="/home/news_article_detail?id=2090703870787874818">ZnAl-layered double hydroxides template-induced formation of ZnO/ZnSe heterostructures on the surface of coal-tar-pitch derived carbon for high-efficiency sodium storage</a>
484                </h2>
485                <div class="v4-news-dsc" style="height: 60px">
486                  <a href="/home/news_article_detail?id=2090703870787874818">Sodium-ion batteries are considered a promising technology for large-scale energy storage because sodium is abundant and widely distributed. A major challenge, however, is finding anode materials that combine high capacity, fast reaction kinetics and long cycling stability while remaining practical and low cost.
487A research team from Taiyuan University of Technology and Taiyuan University of Science and Technology has reported a strategy that addresses this challenge by using coal tar pitch, an industrial 
487carbon-rich byproduct, as the carbon source for a new composite anode.
488Their work was published in Energy Materials and Devices on June 11, 2026.
489The team used a zinc-aluminum layered double hydroxide (ZnAl-LDH) as a template to induce the formation of ZnO/ZnSe heterostructures embedded in hierarchical porous carbon. During synthesis, the LDH template serves two functions: it guides the local structure of the active material and helps generate pores in the carbon framework. This produces intimately connected ZnO/ZnSe-carbon interfaces rather than simply mixing active particles with carbon.
490The heterostructure is important because ZnO and ZnSe can work together at their interface. The interfacial contact promotes charge transfer, while the mixed oxygen and selenium anion environment, the nanoscale dispersion of ZnO/ZnSe and the conductive carbon network help improve reaction kinetics and reduce the mechanical strain that usually occurs during repeated sodium insertion and extraction.
491&nbsp;“Our goal was to turn an inexpensive industrial byproduct into a functional carbon host and then use interface engineering to overcome the kinetic limitations of metal selenide anodes,” said Jian Wang, the corresponding authors of this paper, professor in the College of Materials Science and Engineering at Taiyuan University of Technology. “The LDH template allowed us to build the active heterostructure and the porous carbon architecture in one integrated design.”
492Electrochemical tests showed that the composite anode achieved a reversible capacity of 637.5 mAh g<sup>−1</sup> at 100 mA g<sup>−1</sup>. Under high-rate cycling, it retained 259.7 mAh g<sup>−1</sup> after 1000 cycles at 5 A g<sup>−1</sup>. Kinetic analysis further showed that capacitive storage dominated the response, contributing 93.3% of the total charge storage at 1.2 mV s<sup>−1</sup>. This behavior explains the material’s strong rate capability. The team also assembled a full cell using an Na<sub>3</sub>V<sub>2</sub>(PO<sub>4</sub>)<sub>3</sub> cathode to evaluate practical potential. The full cell retained 147.5 mAh g<sup>−1</sup> after 100 cycles, indicating that the material design is not limited to half-cell testing.
493“This study shows that template-directed heterostructure engineering can be an effective route for developing advanced sodium-ion battery anodes from low-cost carbon resources,” the Dr. Wang said. “The next step is to further optimize electrode formulation and evaluate the material under more practical cell conditions.”
494Other contributors include the Yiming Liu, the professor in the College of Environmental Science and Engineering at Taiyuan University of Technology and deputy dean of School of Chemical Engineering and Technology at Taiyuan University of Science and Technology; Yibo Zhao also from the College of Environmental Science and Engineering at Taiyuan University of Technology and in Taiyuan, China. Peihua Li, Haochen Xie, Yalong Wang and Wanggang Zhang from the College of Materials Science and Engineering at Taiyuan University of Technology; Rufeng Tian and Xiaohong Li from the College of Chemistry and Chemical Engineering at Taiyuan University of Technology.
495This work was supported by the National Natural Science Foundation of China (Grant Nos. U25B20110, 22075197 and 22278290), the Shanxi Provincial Central Guidance Fund for Local Science and Technology Development Projects (Grant No. YDZJSX2024D022), and the Key Research and Development (R&amp;D) Projects of Shanxi Province (Grant No. 202102040201003).
496DOI Link:
497https://doi.org/10.26599/EMD.2026.9370093</a>
498                </div>
499                <div class="v4-news-date">
500                  <span>Physical Sciences and Engineering</span>
501                </div>
502              </div>
503              <div class="v4-news-img v4-news-img-l">
504                <a href="/home/news_article_detail?id=2090703870787874818">
505                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-08-21/309484f3-85c8-447c-8eba-242b7f5f83ba.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=be1VS5GakvOp%2BCeV%2BK6owrJDRvM%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVV
505hwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
506                </a>
507              </div>
508            </div>
509          </div>
510          <div class="v4-news-item">
511            <div class="v4-news-content v4-news-content-article">
512              <div class="v4-news-cl">
513                <div class="v4-news-qikan">
514                  <a href="/journal/2226-4108">Journal of Advanced Ceramics</a>
515                </div>
516                <h2 class="v4-news-title">
517                  <a href="/home/news_article_detail?id=2090702636160937986">Insight into the synergistic effect of rare-earth elements on the CMAS corrosion behavior in (RE1/4Tm1/4Yb1/4Lu1/4)2Si2O7 (RE = Gd, Ho and Sc) materials at 1300 °C</a>
518                </h2>
519                <div class="v4-news-dsc" style="height: 60px">
520                  <a href="/home/news_article_detail?id=2090702636160937986">Rare-earth disilicates RE<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> have been widely acknowledged as state-of-the-art environmental barrier coating material in commercial applications due to its superior thermochemical stability and high temperature water vapor corrosion resistance. However, a primary limitation of RE<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> EBCs for high-performance aero-engine applications-particularly at sustained operating temperatures exceeding 1300 °C is their inadequate resistance to CMAS deposits. The advent of high-entropy or multicomponent design in materials engineering has brought new inspiration to address this bottleneck, and its effectiveness has been proven in recent research. In fact, all these benefits originate from the introduction of various rare-earth ions and their respective advantages in coupling. However, the correlation mechanism between the efficacy of rare-earth components and the final corrosion resistance is still unclear, and an exact guideline for component elements selection is still absent.
521Recently, a joint team of Professor Sun Luchao’s group from Institute of Metal Research, Chinese Academy of Sciences, and Professor Wang Jingyang’s group from Liaoning Academy of Materials confirmed the possibility of achieving tunable CMAS corrosion resistance in rare-earth disilicates through the synergistic effect of multiple rare-earth components through three novel multicomponent rare-earth disilicate EBC materials, and provided a criterion for designing multicomponent rare-earth silicates with enhanced CMAS corrosion resistance.
522This team published their work in Journal of Advanced Ceramics on July 28, 2026.
523“The findings in this work provide valuable insights for in-depth understanding of the synergistic effects among rare-earth components and the correlation mechanism of multicomponent rare-earth disilicates.” Said Jingyang Wang, Vice President of Liaoning Academy of Materials (LAM) and the director of Institute of Coating Technology for Hydrogen Gas Turbines in LAM (China).
524“In this study, three multicomponent (RE<sub>1/4</sub>Tm<sub>1/4</sub>Yb<sub>1/4</sub>Lu<sub>1/4</sub>)<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> (RE = Gd, Ho and Sc) materials were designed and exposed to CMAS at 1300 °C for durations of 1, 4, and 50 h. Mechanistic analysis reveals that the performance divergence primarily stems from distinct corrosion mechanisms: (Gd<sub>1/4</sub>Tm<sub>1/4</sub>Yb<sub>1/4</sub>Lu<sub>1/4</sub>)<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> and (Ho<sub>1/4</sub>Tm<sub>1/4</sub>Yb<sub>1/4</sub>Lu<sub>1/4</sub>)<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> predominantly undergo dissolution-reprecipitation processes, whereas (Sc<sub>1/4</sub>Tm<sub>1/4</sub>Yb<sub>1/4</sub>Lu<sub>1/4</sub>)<sub>2</sub>Si<sub>2</sub>O<sub>7</sub> is dominated by intergranular corrosion penetration. Such mechanistic transition is attributable to compositional tuning of rare-earth elements (Gd and Ho to Sc) in the silicates.” Said Luchao Sun, a professor from Institute of Metal Research, Chinese Academy of Sciences (China).
525“In this study, elements such as Gd and Ho were proven to reduce the Ca/Si ratios by accelerating the precipitation of apatite phase during reaction. This process not only diminishes calcium in CMAS to reduce its corrosion aggressiveness but also retards CMAS diffusion through the formation of a dense product layer, collectively enhancing the corrosion resistance of disilicates. Thus, an optimal stoichiometric ratio between active (e.g., Gd and Ho) and inert (e.g., Yb and Lu) elements is essential to synergistically activate the precipitation for corrosion mitigation and the intrinsic resistance enhancement, thereby maximizing CMAS corrosion resistance in disilicate systems.” Said Luchao Sun.
526About Author
527Luchao Sun is currently a professor of advanced ceramics and composites division, Institute of Metal Research, Chinese Academy of Sciences. His main research interests cover theoretical and experimental investigations on advanced ceramics and composites for harsh environment applications and advanced materials for thermal/environmental barrier coatings.
528Jingyang Wang is currently the Vice President of Liaoning Academy of Materials (LAM) and the director of Institute of Coating Technology for Hydrogen Gas Turbines in LAM. His research covers fundamental research and engineering applications of structural ceramics, composite materials and high-temperature protective coatings for extreme service environments.
529Ziyu Wang is currently a doctor candidate in Shenyang National Laboratory for Materials Science, Institute of Matel Research. His research focuses on the composition design, preparation, and performance optimization of the environmental barrier coatings for aero-engines.
530Funding
531This work was supported by the National Natural Science Foundation of China (U21A2063); LiaoNing Revitalization Talents Program (XLYC2203090); International Partnership Program of the Chinese Academy of Sciences (172GJHZ2022094FN).
532About&nbsp;Journal of Advanced Ceramics
533Journal of Advanced Ceramics&nbsp;(JAC) is an international academic journal that presents the state-of-the-art results of theoretical and experimental studies on the processing, structure, and properties of advanced ceramics and ceramic-based composites. JAC is Fully Open Access, monthly published by Tsinghua University Press, and exclusively available via&nbsp;SciOpen. JAC’s 2024 IF is 16.6, ranking in Top 1 (1/34, Q1) among all journals in “Materials Science, Ceramics” category, and its 2024 CiteScore is 25.9 (5/130) in Scopus database. ResearchGate homepage:&nbsp;https://www.researchgate.net/journal/Journal-of-Advanced-Ceramics-2227-8508</a>
534                </div>
535                <div class="v4-news-date">
536                  <span>Ceramics</span>
537                </div>
538              </div>
539              <div class="v4-news-img v4-news-img-l">
540                <a href="/home/news_article_detail?id=2090702636160937986">
541                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-08-21/af4cce66-9baa-4485-812a-4db9bbfc1d33.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=1XX6VsecZqmQ6GMt5HT%2FTRlAn4Y%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
542                </a>
543              </div>
544            </div>
545          </div>
546          <div class="v4-news-item">
547            <div class="v4-news-content v4-news-content-article">
548              <div class="v4-news-cl">
549                <div class="v4-news-qikan">
550                  <a href="/journal/3008-1203">Ocean</a>
551                </div>
552                <h2 class="v4-news-title">
553                  <a href="/home/news_article_detail?id=2087376531194671106">Predicting underwater landslides to protect vital infrastructure</a>
554                </h2>
555                <div class="v4-news-dsc" style="height: 60px">
556                  <a href="/home/news_article_detail?id=2087376531194671106">Underwater landslides are a destructive force, putting vital submarine infrastructure at serious risk and can even generate damaging tsunamis. By analyzing research completed from 2000 to 2025, scientists tracked this phenomenon to better understand what triggers the landslides, how they move and generate tsunamis, and the dangers they pose.
557After reviewing the data, researchers determined three recommendations for submarine landslide research going forward. The results of this review were published in Ocean on 6 March.
558“Given the rapid growth of global marine development, studying the mechanisms of triggering, movement, and disaster impacts behind submarine landslides has become increasingly urgent. More than 25% of global oil and gas production currently originates offshore, with projections indicating substantial growth in marine energy, including oil, gas, and wind energy, activities by 2040”, said Prof. Fawu Wang, a researcher at Tongji University in Shanghai, China.
559Marine environments are complex, and landslides are caused by interacting factors. They are most often caused by earthquakes, but longer-term processes like rapid sedimentation, material like magma and mud settling in the ocean, and erosion also contribute to submarine landslides. Shaking caused by earthquakes increases sliding force, reduces the strength of the soil, and starts liquefaction, which is when loosely packed sediments weaken. Even even small earthquakes can trigger large submarine landslides.
560Hydrodynamic forces are a more easily monitored cause of submarine landslides. “Waves, tides, bottom currents, and internal waves can trigger submarine landslides by increasing bottom shear stress, decreasing the shear strength of sediments, and causing dynamic changes in pore water pressure. These processes interact with the sediment structure, compromising its stability and potentially resulting in slope failure, particularly in regions with steep slopes or loose unconsolidated sediments”, said Prof. Wang. While scientists have been able to monitor the seafloor for the conditions leading to landslides, more research is needed to make the models more accurate.
561There have been multiple studies to better understand how submarine landslides move using different techniques. They have used experimental devices and numerical simulations and physical models, but both methods are limited by the complexity of marine conditions. Researchers suggested that future research should look at modeling frameworks that incorporate machine learning, observation, and numerical models.
562Finally, the review focused on research into how submarine landslides can cause a tsunami. Most tsunamis are caused by earthquakes, and those caused by submarine landslides are usually smaller in strength and scale. Even so, these tsunamis can be destructive, affecting coastal regions and engineering infrastructure that is right off the coast. Though these tsunamis are generally weaker, the initial wave can often be much higher than an earthquake-generated tsunami and cause significant damage.
563Researchers also focused on specific offshore infrastructure, which is only increasing because of renewable energy developments. “Submarine landslides can result in catastrophic consequences, including pipeline suspension and rupture, platform overturning, and erosion of wind turbine foundations. As human activities extend into deeper seas, interactions between submarine landslides and marine engineering infrastructure are becoming more frequent, highlighting the urgent need to systematically uncover the disaster mechanisms associated with these events”, said Prof. Wang.
564Looking ahead, researchers developed three conclusions that give guidance for future research into submarine landslides. First, they suggested that future research focus on understanding how different triggering factors interact and establishing metrics for predicting dangerous landslides. Second, they suggested improving numerical simulations to better reflect actual underwater environments. And third, they recommended improving modeling to incorporate machine learning and probabilistic analysis, along with incorporating coastal vulnerability for better tsunami risk assessment.
565Other contributors include Youqian Feng and Ye Chen of Tongji University and Kongming Yan of University of Cambridge.
566The Fundamental Research Funds of China for the Central Universities and the Interdisciplinary Collaborative Research Demonstration Project at Tongji University supported this research.
567&nbsp;
568DOI Link:
569https://doi.org/10.26599/OCEAN.2025.9470014
570About Ocean
571Ocean is an international peer-reviewed journal that offers open access and serves as a multidisciplinary platform for the state-of-the-art research and practice in the domains of ocean science, technology, and engineering. The journal is dedicated to publishing articles, reviews and perspectives in these areas, with the goal of promptly disseminating and promoting theoretical, numerical, site-based, and experimental advancements in the context of global sustainability.</a>
572                </div>
573                <div class="v4-news-date">
574                  <span>Physical Sciences and Engineering</span>
575                </div>
576              </div>
577              <div class="v4-news-img v4-news-img-l">
578                <a href="/home/news_article_detail?id=2087376531194671106">
579                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-08-12/57aae987-328d-406d-9fb3-97f50356d532.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=mNobyg3CUJY9YNqQtqjqrb1jiss%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
580                </a>
581              </div>
582            </div>
583          </div>
584          <div class="v4-news-item">
585            <div class="v4-news-content v4-news-content-article">
586              <div class="v4-news-cl">
587                <div class="v4-news-qikan">
588                  <a href="/journal/2096-0433">Computational Visual Media</a>
589                </div>
590                <h2 class="v4-news-title">
591                  <a href="/home/news_article_detail?id=2082711880624308226">Self-supervised 3D AI data training alleviates model learning bottleneck</a>
592                </h2>
593                <div class="v4-news-dsc" style="height: 60px">
594                  <a href="/home/news_article_detail?id=2082711880624308226">Artificial intelligence (AI) is vastly superior to human analysis in many ways—particularly in processing, analyzing and synthesizing massive quantities of information at incredibly fast speeds—but AI model performance is only as good as the data used to train it.
595Much like humans, AI models need to be taught to perform specific tasks, and oftentimes, the more quality data used to train a model, the more proficient the model will be at recognizing patterns or processing data, for example.
596Currently, one of the biggest bottlenecks for AI model implementation is the training phase, where models are taught to perform specific tasks using data. Prior to training, large amounts of data often require humans to annotate or structure data into specific formats that algorithms require. As models get larger, the amount of labor and time required to transform or appropriately label data can grow exponentially, delaying model training and deployment.
597To address this issue, a group of researchers from Fudan University and Nanyang Technological University wrote a comprehensive literature review outlining how AI models can supervise their own learning using point cloud data, or raw, three-dimensional (3D) spatial coordinates that correlate to the surface of an object, greatly reducing or eliminating the amount of human labor required to annotate large datasets for model training.
598The team published their review on April 22 in the journal Computational Visual Media, published by Tsinghua University Press.
599“Just as self-supervised learning has already revolutionized how AI understands natural language and 2D images, it is now becoming the key to unlocking 3D data. 3D point clouds are essential for technologies like self-driving cars and robotics, but training AI to understand them traditionally required humans to manually label millions of 3D coordinates, creating a great bottleneck. This technology is attracting immense attention because it allows the AI to learn generic, powerful 3D representations directly from raw, unlabeled data, paving the way for the emergence of true 3D foundation models,” said Ben Fei, research fellow at the Chinese University of Hong Kong and first author of the review paper.
600One popular self-supervised AI learning approach uses deep neural networks (DNNs) and assigns various pretext tasks for the network to 
600solve. The pretext task serves as a temporary, artificially created problem to train an AI model on unlabeled data. This task forces the model to understand the fundamental structure and patterns of the data, which can later be applied to downstream real-world applications.
601Several different pretext tasks have been proposed for self-supervised learning, including point cloud reconstruction, or the conversion of unstructured, 3D data into a usable digital 3D model; contrastive learning, which teaches models to understand data through comparison; and multi-modal learning, which integrates many different types of data into a single unified model.&nbsp; Pretext tasks share two common properties: 1) The visual features of point clouds must be captured by DNNs to solve the pretext task, and 2) the supervisory signal is generated from the data itself, which results in self-supervision, by exploiting its structure.
602“The main takeaway is that effective 3D representation learning is closely tied to choosing the right pre-training tasks that leverage the data itself rather than human labels. For instance, these self-supervised approaches allow a single model to learn rich, generic geometric features from the data's own structure. This comprehensive understanding serves as a powerful prior that can be easily transferred to various applications, from indoor robotics to outdoor autonomous driving. Our paper provides a comprehensive, unified roadmap that categorizes these learning schemes, showing how to systematically build and adapt these frameworks toward capable 3D foundation models,” said Fei.
603Despite these advances, hurdles still exist for self-supervised 3D model training.
604“Our next step is to overcome the challenges unique to 3D representation learning, including optimizing the massive computing power required and establishing higher-quality, standardized 3D pre-training datasets. Our ultimate goal is to break down the barriers between different modalities. By successfully scaling up 3D foundation models and aligning them with large language models and 2D vision, we hope to provide the crucial spatial intelligence needed to achieve true Artificial General Intelligence,” said Fei.
605Jingyi Xu, Yixuan Li, Weidong Yang, Qingyuan Zhou, Liwen Liu and Tianyue Luo from the School of Computer Science at Fudan University in Shanghai, China; and Ying He from the College of Computing and Data Science at Nanyang Technological University in Singapore, Singapore also contributed to this research.
606This research was supported by the JC STEM Lab of AI for Science and Engineering, funded by The Hong Kong Jockey Club Charities Trust, the MTR Research Funding (MRF) Scheme (CHU-24003), the Research Grants Council of Hong Kong (CUHK14213224) and the Ministry of Education, Singapore, Academic Research Fund Grant (RT19/22).
607&nbsp;</a>
608                </div>
609                <div class="v4-news-date">
610                  <span>Computer Science</span>
611                </div>
612              </div>
613              <div class="v4-news-img v4-news-img-l">
614                <a href="/home/news_article_detail?id=2082711880624308226">
615                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-07-30/129dc233-c991-4d05-8115-a58b644f59d8.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=Oy5%2BPjdrhi0%2FtsbbGld5lLKWnms%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
616                </a>
617              </div>
618            </div>
619          </div>
620          <div class="v4-news-item">
621            <div class="v4-news-content v4-news-content-article">
622              <div class="v4-news-cl">
623                <div class="v4-news-qikan">
624                  <a href="/journal/2223-7690">Friction</a>
625                </div>
626                <h2 class="v4-news-title">
627                  <a href="/home/news_article_detail?id=2080481331704823810">The multifaceted science of where rubber meets road</a>
628                </h2>
629                <div class="v4-news-dsc" style="height: 60px">
630                  <a href="/home/news_article_detail?id=2080481331704823810">Increasing pavement skid resistance on a London highway by 50% reduced the total number of accidents over four years by 45%, according to the United Kingdom Trans
630port and Road Research Laboratory. The United States National Transportation Safety Board and Federal Highway Administration stated that poor pavement skid resistance is a major cause of driving accidents. These are just two sources cited by an international team of researchers who recently reviewed more than 3,000 studies on tire-road friction and found that while tire-road friction theory and simulations have been well developed across the research scales — rubber-pavement scale, the tire-road scale and the vehicle scale — the scales themselves remain largely unconnected.
631According to the researchers, who published their work on Feb. 10 in Friction, that disconnect is a major hindrance in advancing the field, and ultimately, road safety for all.&nbsp;
632“The concept of pavement skid resistance pertains to the capacity of road surfaces to provide adequate friction during diverse vehicular operations, including braking, accelerating and cornering,” said co-corresponding author Yuchuan Du, professor, Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, China. “It is critical for preventing skidding incidents and ensuring driving safety, serving as a significant indicator for road performance evaluation and maintenance decisions.”
633Tire-road friction has been studied for more than half a century, Du said, explaining that early studies focused on improving the materials and structure of the tire. However, researchers later learned that simply improving tires did not reduce the proportion of skidding accidents, so more researchers in the field focused on the skid resistance of the road itself.
634“From the perspective of road researchers, the primary concern is the role that the pavement itself plays within the tire-rubber friction system, which is referred to as the pavement skid resistance,” Du said. “This is an inherent characteristic of the pavement within the tire-road friction system, and understanding the tire-road friction system is fundamental for evaluating the skid resistance of asphalt pavements.”
635From their analysis of the field’s scientific literature, the researchers found that studies typically fell into one of four research paradigms: experimental science, theoretical science, computational science and data science. They also found that the research can be categorized into the rubber-pavement scale, the tire-road scale and the vehicle scale. Across the paradigms and scales, the team found that experimental studies were limited by specific variables that have made reproducibility difficult, and that computational studies were limited by their ability to integrate friction mechanisms, making them difficult to interpret properly.
636“The synergistic development of the four research paradigms can promote and advance the understanding and application of tire-road friction mechanisms,” Du said.
637The researchers developed a multiscale architecture for tire-road friction research to clarify the different concepts and boundaries as they feed into two coefficients: rubber-pavement friction and tire-pavement adhesion. Both feed into determining the skidding risk of the vehicle, along with other variables. The two coefficients are determined by pavement factors, such as texture and aggregate; rubber-related material properties, like density; environmental factors, including temperature and humidity; tire-related physical conditions, such as pattern and pressure; and vehicle-related operation conditions, like load and velocity.
638“With the accumulation of data, data-based research has gradually become the primary research paradigm for evaluating pavement skid resistance,” Du said. “However, the integration of data with theoretical research remains a question that requires further exploration. Leveraging theory to enhance the predictive capabilities of data models and utilizing data mining to reveal the underlying mechanisms and patterns of pavement skid resistance are areas that warrant deeper investigation.”
639Collaborators include Zihang Weng and co-corresponding author Zhen Leng, Department of Civil and Environmental Engineering, Hong Kong Polytechnical University, China; Chenglong Liu and Difei Wu, Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, China;
639 and Bryan T. Adey, Institute of Construction and Infrastructure Management, Swiss Federal Institute of Technology, Switzerland. Weng is also affiliated with Tongji University and Hong Kong Polytechnic University Shenzhen Research Institute.
640The National Natural Science Foundation of China, the Shanghai Science and Technology Innovation Action Plan and the Carbon Neutrality Funding Scheme of PolyU supported this research.</a>
641                </div>
642                <div class="v4-news-date">
643                  <span>Engineering</span>
644                </div>
645              </div>
646              <div class="v4-news-img v4-news-img-l">
647                <a href="/home/news_article_detail?id=2080481331704823810">
648                  <img src="https://wqketang.cn-beijing.oss.aliyuncs.com/image/journal_prod/2026-07-24/8ca411f3-f144-4335-b24d-6eda85732fa7.png?Expires=1791022940&amp;OSSAccessKeyId=STS.NYPeUykizZdednLmX6Au2rHW9&amp;Signature=zUFLLEA6Eyzv53MndB7JTCzV3cU%3D&amp;security-token=CAISywJ1q6Ft5B2yfSjIr5rlLu%2FNhrZb7aaOZkj9iVhjTfoenY382zz2IHtKenhsBOsbtfk1mG5W5%2FgZlqJ9SptIAEfJa9d99MyAYfAChNaT1fau5Jko1beHewHKeTOZsebWZ%2BLmNqC%2FHt6md1HDkAJq3LL%2Bbk%2FMdle5MJqP%2B%2FUFB5ZtKWveVzddA8pMLQZPsdITMWCrVcygKRn3mGHdfiEK00he8TouufTinpHMskGA1Aell7Mvyt6vcsT%2BXa5FJ4xiVtq55utye5fa3TRYgxowr%2Fwo0v0YpGya5YzHXwcPskvdKZbo78UqLQlla6w%2BGqFJqvPxr%2Fp8t%2Fx5fWJKAezhVgs8cVM8JOjIqKOscIsihoG24WjtQD1LXoQ6vH1ZZ5%2FAzVZaVWUEMpNqcRxTMldXaQqhSZT8m3upKXzOJeX7ls9%2Fv%2FV7p7%2FZlbLUTzDnGoABLLQFVtFve1ioQOqb%2FOGBeb9SSwgtqZu3sWVVhwdr6NfJWppHwu5K9TB5JO6HA2aY%2BETBPLjZWGrKTrXmYBeFTbCxNjbeY6TQTrV77TjQCRZpjI9EVZaYolgSY69CW0GFIjU4%2Fnr82U7vdhIFZ6ZrJemhOIfXvn%2BDA6%2FhISalB00gAA%3D%3D" alt="" />
649                </a>
650              </div>
651            </div>
652          </div>
653        </div>
654        <div class="zxx-list-page clearfix">
655          <div class="zxx-list-page-left"><span>Total 188</span></div>
656          <div class="zxx-list-page-right" id="yx-page"></div>
657        </div>
658        
659      </div>
660      <div id="footer">
661        <div id="v4-footer" class="v4-footer">
662  <div class="footer-panel">
663    <div class="panel-item item-1">
664      <a href="/home"><img class="footer-logo" src="/assets/img/home/logo.png" alt="SciOpen"/></a>
665      <div class="footer-share">
666        <a class="share-icon" target="_blank" href="https://twitter.com/SciOpenTUP"><img src="/assets/svg/twitter.svg" alt="twitter"/></a>
667        <a class="share-icon" target="_blank" href="https://www.linkedin.com/company/91320577/admin/feed/posts/"><img src="/assets/svg/linkedin.svg" alt="linkedin"/></a>
668        <a class="share-icon" target="_blank" href="https://www.facebook.com/SciOpentup/"><img src="/assets/svg/facebook.svg" alt="facebook"/></a>
669        <a class="share-icon" target="_blank" href="https://bsky.app/profile/sciopentup.bsky.social"><img src="/assets/svg/blueSky.svg" alt="bsky"/></a>
670        <a class="share-icon wechat">
671          <img src="/assets/svg/weixin.svg" alt="wechat"/>
672          <div class="wechat-qrcode">
673            <img src="/assets/img/wechat.jpeg" alt="code"/>
674          </div>
675        </a>
676      </div>
677    </div>
678    <div class="panel-item item-2" v-if="aboutList.length">
679      <div class="item-title">About Us</div>
680      <div class="item-content">
681        <div class="content-item">
682          <div v-for="(item,index) in aboutList" class="item">
683            <a :href="item.href" target="_blank">{{item.title}}</a>
684          </div>
685          <div class="item">
686            <a href="https://www.tup.com.cn/en/index" target="_blank">Tsinghua University Press</a>
687          </div>
688        </div>
689      </div>
690    </div>
691    <div class="panel-item item-3" v-if="publishList.length">
692      <div class="item-title">Publish with Us</div>
693      <div class="item-content">
694        <div class="content-item">
695          <div class="item" v-for="(item,index) in publishList">
696            <a :href="item.href" target="_blank">{{item.title}}</a>
697          </div>
698        </div>
699      </div>
700    </div>
701    <div class="panel-item item-4" v-if="supportList.length || contactUs">
702      <div class="item-title">Contact Us</div>
703      <div class="item-content">
704        <div class="content-item">
705          <div class="contact-us" v-html="contactUs"></div>
706          <div v-for="(item,index) in supportList" class="item">
707            <a :href="item.href" target="_blank">{{item.title}}</a>
708          </div>
709        </div>
710      </div>
711    </div>
712  </div>
713  <a id="cd_top" href="javascript:;" class="page-feed-top cd-top"></a>
714  <div class="v4-copyright">
715  <p class="Copyright-des"></p>
716  <p class="mobile-display-none">
717    <a href="https://beian.miit.gov.cn/" target="_blank" style="margin: 0 8px">京ICP备 10035462号-42</a>
718    <a href="https://beian.mps.gov.cn/#/query/webSearch?code=11010802044758" style="margin: 0 8px" rel="noreferrer" target="_blank"
719      ><img src="/assets/img/jing.jpg" style="width: 14px; vertical-align: text-bottom; margin-right: 5px" />京公网安备11010802044758号</a
720    >
721  </p>
722  
722<script>
723    $('.Copyright-des').html('Copyright  © ' + new Date().getFullYear() + '  Tsinghua University Press Ltd.');
724  </script>
724
725</div>
726  
726<script src="/assets/js/v4-footer.js"></script>
726
727</div>
728      </div>
729    </div>
730  </body>
731  
731<script src="/assets/bootstrap/js/bootstrap.min.js"></script>
731
732  
733  
733<script src="/assets/js/codehim.dropdown-81676e5e87203e8d06abec0ee348f306.js"></script>
733
734  
734<script src="/assets/js/menu-8b5b9e3d73fc1a44426d00d5bd542ec7.js"></script>
734
735  
735<script src="/assets/js/footer-92affce77153f0b1aed686e8e0b226af.js"></script>
735
736  
736<script src="/assets/js/yxpage.min-90717f053efe3704e8c38c5c2655dc6b.js"></script>
736
737  
737<script>
738    var contextPath = '';
739    var pageNo = '1';
740    var pageSize = '10';
741    var total = '188';
742    var type = '1';
743
744    $('#yx-page').pagination({
745      total: total,
746      pageSize: pageSize,
747      current: pageNo,
748      jump: true,
749      coping: true,
750      showPage: false,
751      keepShowPN: true,
752      jumpBtn: 'Go',
753      callback: function (api) {
754        pageNoChange(api.getCurrent(), api.getPageSize());
755      },
756    });
757
758    // 切换分页
759    function pageNoChange(pageNo, pageSize) {
760      location.href = contextPath + '/home/news_article_page?pageNo=' + pageNo + '&pageSize=' + pageSize + '&type=' + type;
761    }
762  </script>
762
763</html>

Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.