[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81794-en":3,"doc-seo-81794-105":29,"detail-sidebar-cat-0-en-105":90},{"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":4,"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},81794,137441390410,"Hazel","https://ap-avatar.wpscdn.com/avatar/2000252f4ab5702993?_k=1776741390130283984",8,"Research & Report","Agentic generation of verifiable rules for deterministic, self-expanding reaction classification","Computer-assisted synthesis planning decomposes target molecules into precursors via large libraries of reaction rules, each with a deterministic, interpretable label. Chemistry’s long-tailed distribution makes manual rule encoding infeasible and forces existing tools to depend on fixed, non-adaptive rulesets. A fully automated multi-agent LLM pipeline classifies reactions and generates new rules from 665,901 US patent reactions under a corpus verification loop, expanding taxonomy from 68 to 14,073 classes without human curation.","arXiv :2607 .0 106 1v2 [ cs .AI ] 5 Jul 2026  \nAgentic generation of verifiable rules for deterministic, self-expanding reaction classification  \nDaniel Armstrong1 , Maarten Dobbelaere1,2 , Valentas Olikauskas1,3 , Helena Avila1,3 , Octavian Susanu1 , Jérôme Waser1,3 , Philippe Schwaller1,3  \n1École Polytechnique Fédérale de Lausanne (EPFL), Switzerland  \n2 Ghent University, Belgium  \n3National Centre of Competence in Research (NCCR) Catalysis, Switzerland  \n{daniel.armstrong,philippe.schwaller}@epfl.ch  \nAbstract  \nComputer-assisted synthesis planning breaks target molecules into accessible precursors using large libraries of reaction rules that assign each transformation a deterministic, interpretable label. But chemistry is long-tailed, making manual encoding intractable, and existing tools rely on fixed rulesets that cannot adapt to new chemistries. Here we present a fully automated pipeline in which a multi-agent framework of large language models (LLMs) classifies reactions and writes the rules themselves across 665,901 US patent reactions, generating each rule under a verification loop that tests it against the corpus. It expands a standard taxonomy from 68 to 14,073 classes without human curation. With a lightweight fingerprint classifier, it classifies 97.7% of unseen reactions, matching a leading proprietary classifier while resolving chemistry more finely and extending on demand to chemistry outside its training distribution. The result is a living reactivity database and a general route to turning generative models into reliable, self-expanding symbolic systems.  \n1 Introduction  \nChemical synthesis planning, in which a complex target molecule is decomposed recursively into simpler building blocks, remains a core challenge in drug discovery and materials design. In their seminal 1969 work, Corey and Wipke proposed computer-assisted synthesis planning (CASP), applying encoded mechanistic rules to suggest synthetic pathways 1–4. To date, this rule-based approach remains common; the most comprehensive implementations contain tens of thousands of manually designed reaction rules with hardcoded protection and incompatibility logic 5. Yet the distribution of chemical reactions follows a power law 6,7 , and the long tail of rare transformations, each demanding the same careful encoding as a commonplace amide coupling, makes exhaustive manual coverage practically intractable. Automating the extraction and generalisation of reaction rules is particularly timely, as the utilisation of Large Language Models (LLMs) in chemistry matures 8,9 . Such models may offer a new way to encode the synthetic toolbox, replacing human logic with the automatic inference of symbolic reaction transforms from chemical data.  \nThe automation of reaction rule generation requires solving two distinct problems. First, a reaction must be assigned to a named class within a structured taxonomy. Carey et al. 10 introduced a semantic hierarchy of ten superclasses to analyse common industrial pharmaceutical transformations, and together with the subsequent medicinal chemistry analysis by Roughley and Jordan 11 , this work informed the Royal Society of Chemistry’s RXNO ontology of named reactions 12. Second, the generalised transformation within each class must be encoded as a computable reaction rule. Such  \nPreprint.  \nrules are typically expressed as SMIRKS, a text based format which encodes a graph transformation around the reaction centre and its local atomic environment 13. However, constructing these patterns depends heavily on chemical intuition; even a single transformation requires careful specification of atom mappings, stereochemical constraints, and functional group compatibility, and the effort scales poorly to the thousands of reaction types observed in practice.  \nData-driven approaches to reaction classification have emerged over the past decade, enabled in large part by the publicly available USPTO reaction dataset introduced by 14,15 . ","cbCaidEmWKEeBYGp","https://ap.wps.com/l/cbCaidEmWKEeBYGp","pdf",1399338,1,40,"English","en",105,"# Abstract\n# Introduction","[{\"question\":\"What problem does the paper address in reaction rule generation?\",\"answer\":\"Manual encoding of reaction rules is infeasible for the long-tailed space of chemical transformations, and existing tools depend on fixed rulesets that cannot adapt to new chemistries.\"},{\"question\":\"How does the proposed pipeline generate reaction rules?\",\"answer\":\"A multi-agent framework of LLMs classifies reactions and writes rules themselves, using a verification loop that tests each generated rule against the patent reaction corpus.\"},{\"question\":\"What performance and taxonomy expansion results are reported?\",\"answer\":\"The approach expands a taxonomy from 68 to 14,073 reaction classes without human curation and uses a lightweight fingerprint classifier to classify 97.7% of unseen reactions, matching a leading proprietary classifier while refining chemistry granularity and extending to out-of-distribution chemistry on demand.\"}]",1784176191,101,{"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":85,"head_meta":87,"extra_data":89,"updated_unix":27},"agentic-generation-of-verifiable-rules-for-deterministic-self-expanding-reaction-classification","",{"@graph":35,"@context":84},[36,53,67],{"@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/agentic-generation-of-verifiable-rules-for-deterministic-self-expanding-reaction-classification/81794/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"What problem does the paper address in reaction rule generation?","Question",{"text":74,"@type":75},"Manual encoding of reaction rules is infeasible for the long-tailed space of chemical transformations, and existing tools depend on fixed rulesets that cannot adapt to new chemistries.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"How does the proposed pipeline generate reaction rules?",{"text":79,"@type":75},"A multi-agent framework of LLMs classifies reactions and writes rules themselves, using a verification loop that tests each generated rule against the patent reaction corpus.",{"name":81,"@type":72,"acceptedAnswer":82},"What performance and taxonomy expansion results are reported?",{"text":83,"@type":75},"The approach expands a taxonomy from 68 to 14,073 reaction classes without human curation and uses a lightweight fingerprint classifier to classify 97.7% of unseen reactions, matching a leading proprietary classifier while refining chemistry granularity and extending to out-of-distribution chemistry on demand.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,118,121,126,129,133],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":21,"slug":117},7,"Healthcare","healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":119,"slug":120},30,"research-report",{"id":122,"doc_module":4,"doc_module_name":45,"category_name":123,"show_sort_weight":124,"slug":125},9,"Religion & Spirituality",20,"religion-spirituality",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":127,"show_sort_weight":124,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":45,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":45,"category_name":135,"show_sort_weight":105,"slug":136},19,"General","general"]