[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82944-en":3,"doc-seo-82944-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},82944,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Catalyst Papers in Artificial Intelligence Research","Catalyst Papers in Artificial Intelligence Research: A Landscape on ICLR from 2017 to 2025 examines how to identify “catalyst” submissions—papers whose descendants measurably redirect future research. Using 36,113 ICLR papers (2017–2025), the study compares four disruptiveness measures: CD index, node2vec, the direction-aware EDM, and an LLM-based semantic rater. EDM most strongly predicts highly cited papers (AUC 0.83), while peer review scores are essentially orthogonal to future disruptiveness (|ρ|≤0.005).","Catalyst Papers in Artificial Intelligence Research: A Landscape on ICLR from 2017 to 2025  \nFan Huang  \nIndiana University Bloomington [huangfan@acm.org](huangfan@acm.org)  \narXiv :2607 .05401v1 [ cs .DL] 24 May 2026  \nAbstract  \nA small number of methodological contributions, including word2vec, the Transformer, large-scale pre-training, and reinforcement learning from human feedback, have reshaped NLP and AI research over the past decade.  \nOpenReview now makes numeric reviewer scores and accept/reject decisions public for every ICLR submission. Whether such review signals identify trajectory-changing papers at submission time, however, remains untested at corpus scale. We answer this question on  \n36 , 113 papers from ICLR 2017–2025, identifying catalysts: papers whose descendants measurably redirect future research. We compare four disruptiveness measures (the Consolidation/Destabilization (CD) index, node2vec, the direction-aware Embedding Disruptiveness Measure (EDM), and an LLM-based semantic rater) and define a five-type operational catalyst taxonomy (topic initiator, topic bridge, withintopic redirector, simultaneous, and recognitionmisaligned) . EDM leads at identifying highly cited ICLR papers (AUC 0.83 vs. 0.60 for CD, 0.49 for node2vec, and 0 .42 for the LLM rater) . Topic initiators precede a 7.55 × topic-share growth and topic bridges precede an 11.52 × growth in cross-topic citation flow versus yearmatched controls. We found that the peer review scores are essentially orthogonal to future disruptiveness (|ρ|≤0 .005; accepted and rejected papers have indistinguishable mean EDM, p=0 . 11) .  \n1 Introduction  \nA small number of methodological contributions, including word2vec (Mikolov et al., 2013), the Transformer architecture (Vaswani et al., 2017), large-scale pre-training (Radford et al., 2018, 2019), and reinforcement learning from human feedback (Ouyang et al., 2022), have reshaped NLP and AI research over the past decade. OpenReview makes per-paper reviewer scores and accept/reject decisions publicly available for every submission  \nto the International Conference on Learning Representations (ICLR) (González-Márquez and Kobak, 2024) . Identifying which submissions later redirect research trajectories has become a central question in the science of science (Park et al., 2023 ; Wu et al., 2019 ; Fortunato et al., 2018); estimating this potential at submission time, however, remains difficult.  \nThe CD index (Funk and Owen-Smith, 2017) captures only one-hop citation displacement and clusters near zero on sparse networks (Petersen et al., 2024 ; Kim et al., 2026); whether recent embedding- or content-based alternatives (Kim et al., 2026 ; Cohan et al., 2020), combined with peer-review signals, can identify submissions that later reorient a sub-field remains open.  \nPrior work has progressed along two largely separate lines. The bibliometric line introduced the Consolidation/Destabilization (CD) index (Funk and Owen-Smith, 2017), reported a multi-decade decline in disruptiveness (Park et al., 2023), linked team size and atypical combinations to impact (Wu et al., 2019 ; Uzzi et al., 2013), used citationstructure features to anticipate impact (Clausetet al., 2017), and recently proposed the Embedding Disruptiveness Measure (EDM), a direction-aware alternative that is more robust to sparse networks (Kim et al., 2026) . The peer-review line has documented status effects (Merton, 1968 ; Teplitskiyet al., 2022), novelty penalties (Wang et al., 2017), and reviewer inconsistency at ML venues (Cortes and Lawrence, 2021) .  \nThe two lines have rarely been joined: journal corpora (Web of Science, APS) do not include perpaper reviewer scores (Park et al., 2023 ; Kim et al., 2026 ; Clauset et al., 2017), while conference-side work has not connected reviewer signals to longrun trajectory change (Cortes and Lawrence, 2021) . Prior work has examined the relationship between ML-conference review scores and raw citation outcome","cbCaioy04aYSrY0Z","https://ap.wps.com/l/cbCaioy04aYSrY0Z","pdf",2854466,1,23,"English","en",105,"# Introduction\n# Related Work\n## Disruption measures\n# Methods\n## Measurement framework\n# Results\n## Disruptiveness measure comparison\n## Catalyst taxonomy and effects\n# Peer review signal analysis\n# Contributions","[{\"question\":\"What defines a “catalyst” paper in this study?\",\"answer\":\"A catalyst paper is a submission whose descendants measurably redirect later research trajectories at subsequent time points.\"},{\"question\":\"Which disruptiveness measure performs best at identifying highly cited ICLR papers?\",\"answer\":\"The direction-aware Embedding Disruptiveness Measure (EDM) leads, achieving AUC 0.83, outperforming CD (0.60), node2vec (0.49), and the LLM rater (0.42).\"},{\"question\":\"How do peer review signals relate to future disruptiveness?\",\"answer\":\"Peer review scores are essentially orthogonal to future disruptiveness, with near-zero correlation (|ρ|≤0.005), and accepted vs. rejected papers show indistinguishable mean EDM values.\"}]",1784184226,58,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"catalyst-papers-in-artificial-intelligence-research","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/catalyst-papers-in-artificial-intelligence-research/82944/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What defines a “catalyst” paper in this study?","Question",{"text":75,"@type":76},"A catalyst paper is a submission whose descendants measurably redirect later research trajectories at subsequent time points.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which disruptiveness measure performs best at identifying highly cited ICLR papers?",{"text":80,"@type":76},"The direction-aware Embedding Disruptiveness Measure (EDM) leads, achieving AUC 0.83, outperforming CD (0.60), node2vec (0.49), and the LLM rater (0.42).",{"name":82,"@type":73,"acceptedAnswer":83},"How do peer review signals relate to future disruptiveness?",{"text":84,"@type":76},"Peer review scores are essentially orthogonal to future disruptiveness, with near-zero correlation (|ρ|≤0.005), and accepted vs. rejected papers show indistinguishable mean EDM values.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]