{"id":{"repo_id":"trento","oai_identifier":"oai:iris.unitn.it:11572/457410"},"canonical_url":"https://search.dev.ndltd.org/etd/trento/oai:iris.unitn.it:11572/457410","repository":{"repo_id":"trento","name":"Università degli Studi di Trento","base_url":"https://iris.unitn.it/oai/request"},"display":{"title":"Reflexive Composition: Bidirectional Enhancement of Language Models and Knowledge Graphs","abstract":"Large Language Models (LLMs) have significantly advanced natural language processing, yet they con- tinue to face limitations such as hallucinations, factual inconsistencies, and restricted domain-specific knowledge. Knowledge Graphs (KGs), by contrast, provide structured and verifiable information but are expensive to build and maintain manually. This thesis introduces Reflexive Composition, a bidirectional integration framework in which LLMs and KGs iteratively refine each other’s outputs. The framework consists of three interconnected components: (1) LLM2KG, where LLMs assist in the construction and updating of domain-specific knowledge graphs; (2) Human-in-the-Loop (HITL) validation, which supports structured expert review; and (3) KG2LLM, which conditions LLM outputs on verified knowledge to reduce hallucinations and improve consistency. The methodology is evaluated across three case studies: temporal knowledge management, privacy- preserving data integration, and historical bias mitigation. Results include a 23% increase in knowledge extraction accuracy (F1 score from 0.65 to 0.80), a 28.7% reduction in LLM hallucination rates, and measurable improvements in validation efficiency through structured workflows. Reflexive Composition offers a reproducible approach for improving the reliability, scalability, and transparency of AI systems in dynamic or high-risk domains.","abstract_html":"Large Language Models (LLMs) have significantly advanced natural language processing, yet they con- tinue to face limitations such as hallucinations, factual inconsistencies, and restricted domain-specific knowledge. Knowledge Graphs (KGs), by contrast, provide structured and verifiable information but are expensive to build and maintain manually. This thesis introduces Reflexive Composition, a bidirectional integration framework in which LLMs and KGs iteratively refine each other’s outputs. The framework consists of three interconnected components: (1) LLM2KG, where LLMs assist in the construction and updating of domain-specific knowledge graphs; (2) Human-in-the-Loop (HITL) validation, which supports structured expert review; and (3) KG2LLM, which conditions LLM outputs on verified knowledge to reduce hallucinations and improve consistency. The methodology is evaluated across three case studies: temporal knowledge management, privacy- preserving data integration, and historical bias mitigation. Results include a 23% increase in knowledge extraction accuracy (F1 score from 0.65 to 0.80), a 28.7% reduction in LLM hallucination rates, and measurable improvements in validation efficiency through structured workflows. Reflexive Composition offers a reproducible approach for improving the reliability, scalability, and transparency of AI systems in dynamic or high-risk domains.","abstract_has_math":false,"creators":["Mehta, Virendra Kumar"],"institution":"Università degli studi di Trento","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Giunchiglia, Fausto","Casati, Fabio"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-06-20","date_published":"2025-06-20","updated_at":"2026-07-24T05:04:22Z","subjects":["Large Language Models, Knowledge Graphs, LLM Hallucination Reduction, Knowledge Graph Evolu-tion, Bias Mitigation"],"languages":["eng"],"rights":["info:eu-repo/semantics/openAccess","license:Tutti i diritti riservati (All rights reserved)","license uri:iris.PRI01"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["http://dx.doi.org/10.15168/11572_457410","10.15168/11572_457410"],"render_values":[{"text":"http://dx.doi.org/10.15168/11572_457410","href":"http://dx.doi.org/10.15168/11572_457410","code":true},{"text":"10.15168/11572_457410","href":"https://doi.org/10.15168/11572_457410","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/11572/457410","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Mehta, Virendra Kumar","Giunchiglia, Fausto","Casati, Fabio"]},{"key":"dc:creator","label":"Author","values":["Mehta, Virendra Kumar"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025-06-20"]},{"key":"dc:publisher","label":"Institution","values":["Università degli studi di Trento","place:TRENTO"]},{"key":"dc:relation","label":"Dc Relation","values":["firstpage:1","lastpage:178","numberofpages:178"]},{"key":"dc:type","label":"Dc Type","values":["info:eu-repo/semantics/doctoralThesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Large Language Models, Knowledge Graphs, LLM Hallucination Reduction, Knowledge Graph Evolu-tion, Bias Mitigation"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess","license:Tutti i diritti riservati (All rights reserved)","license uri:iris.PRI01"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://hdl.handle.net/11572/457410","http://dx.doi.org/10.15168/11572_457410","10.15168/11572_457410"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Large Language Models (LLMs) have significantly advanced natural language processing, yet they con- tinue to face limitations such as hallucinations, factual inconsistencies, and restricted domain-specific knowledge. Knowledge Graphs (KGs), by contrast, provide structured and verifiable information but are expensive to build and maintain manually. This thesis introduces Reflexive Composition, a bidirectional integration framework in which LLMs and KGs iteratively refine each other’s outputs. The framework consists of three interconnected components: (1) LLM2KG, where LLMs assist in the construction and updating of domain-specific knowledge graphs; (2) Human-in-the-Loop (HITL) validation, which supports structured expert review; and (3) KG2LLM, which conditions LLM outputs on verified knowledge to reduce hallucinations and improve consistency. The methodology is evaluated across three case studies: temporal knowledge management, privacy- preserving data integration, and historical bias mitigation. Results include a 23% increase in knowledge extraction accuracy (F1 score from 0.65 to 0.80), a 28.7% reduction in LLM hallucination rates, and measurable improvements in validation efficiency through structured workflows. Reflexive Composition offers a reproducible approach for improving the reliability, scalability, and transparency of AI systems in dynamic or high-risk domains."]},{"key":"dc:title","label":"Title","values":["Reflexive Composition: Bidirectional Enhancement of Language Models and Knowledge Graphs"]}]}],"canonical_facts":{"dc:contributor":["Mehta, Virendra Kumar","Giunchiglia, Fausto","Casati, Fabio"],"dc:creator":["Mehta, Virendra Kumar"],"dc:date":["2025-06-20"],"dc:description":["Large Language Models (LLMs) have significantly advanced natural language processing, yet they con- tinue to face limitations such as hallucinations, factual inconsistencies, and restricted domain-specific knowledge. Knowledge Graphs (KGs), by contrast, provide structured and verifiable information but are expensive to build and maintain manually. This thesis introduces Reflexive Composition, a bidirectional integration framework in which LLMs and KGs iteratively refine each other’s outputs. The framework consists of three interconnected components: (1) LLM2KG, where LLMs assist in the construction and updating of domain-specific knowledge graphs; (2) Human-in-the-Loop (HITL) validation, which supports structured expert review; and (3) KG2LLM, which conditions LLM outputs on verified knowledge to reduce hallucinations and improve consistency. The methodology is evaluated across three case studies: temporal knowledge management, privacy- preserving data integration, and historical bias mitigation. Results include a 23% increase in knowledge extraction accuracy (F1 score from 0.65 to 0.80), a 28.7% reduction in LLM hallucination rates, and measurable improvements in validation efficiency through structured workflows. Reflexive Composition offers a reproducible approach for improving the reliability, scalability, and transparency of AI systems in dynamic or high-risk domains."],"dc:identifier":["https://hdl.handle.net/11572/457410","http://dx.doi.org/10.15168/11572_457410","10.15168/11572_457410"],"dc:language":["eng"],"dc:publisher":["Università degli studi di Trento","place:TRENTO"],"dc:relation":["firstpage:1","lastpage:178","numberofpages:178"],"dc:rights":["info:eu-repo/semantics/openAccess","license:Tutti i diritti riservati (All rights reserved)","license uri:iris.PRI01"],"dc:subject":["Large Language Models, Knowledge Graphs, LLM Hallucination Reduction, Knowledge Graph Evolu-tion, Bias Mitigation"],"dc:title":["Reflexive Composition: Bidirectional Enhancement of Language Models and Knowledge Graphs"],"dc:type":["info:eu-repo/semantics/doctoralThesis"]},"updated_at":"2026-07-24T05:04:22Z"}