{"id":{"repo_id":"sdstate","oai_identifier":"oai:openprairie.sdstate.edu:etd2-3151"},"canonical_url":"https://search.ndltd.org/etd/sdstate/oai:openprairie.sdstate.edu:etd2-3151","repository":{"repo_id":"sdstate","name":"South Dakota State University","base_url":"https://openprairie.sdstate.edu/do/oai/"},"display":{"title":"A Framework for Reliable Domain AI: Knowledge-Grounded Data and Logic-Graph Reasoning","abstract":"Large language models (LLMs) perform well on general tasks, yet their deployment in specialized, high-stakes domains such as public health remains unreliable. Models hallucinate facts, combine multi-step evidence incorrectly, and retrieve context that corrupts downstream reasoning. Three root causes underlie these failures: training data that underrepresents rare but critical domain themes, reasoning that is opaque and unverifiable, and evidence use that cannot be re-checked when the evidence changes. This dissertation addresses all three. First, we introduce HDBSCAN_SSA, a coarse-to-fine topic-modeling pipeline that couples density-based clustering with a novel Semantic Seed Aggregation algorithm. Using samples drawn from a larger corpus of approximately 469,000 COVID-19 tweets, it discovers more coherent topics than LDA, k-means, and BERTopic, and topic-guided sampling under fixed labeling budgets of 10%, 20%, and 40% consistently improves downstream sentiment-classification accuracy and macro-F1 over random and BERTopic-based sampling. Second, we propose a two-stage logic-graph reasoning interface in which an LLM extracts a structured logic graph from a claim and its evidence table, and a graph attention network classifies the graph. On PubHealthTab, a public-health table-based fact-checking dataset of 1,942 claim-table pairs, the method reaches 88.1% accuracy and 0.853 macro-F1, outperforming chain-of-thought, zero-shot chain-of-thought, and self-ask baselines, with explanation quality (ROUGE-1 0.342, ROUGE-L 0.235) exceeding direct LLM outputs. Third, we propose a grounded verification framework that makes structured reasoning re-checkable against current evidence. A multi-agent pipeline binds extracted operations to exact table cells, and a symbolic layer checks whether the bound values still support the stated reasoning result, overriding the verdict when they do not. The framework achieves 95.6% clean accuracy, 9.8 percentage points above the strongest prompting baseline. Under evidence-side attacks it corrects 95.2% of 209 cell-swap and 91.1% of 538 numerical perturbations, and it resolves 74.9% of 538 claim-side negations through re-extraction. Together, these contributions form a data-to-reasoning-to-verification pipeline for domain AI whose outputs are grounded, inspectable, and robust.","abstract_html":"Large language models (LLMs) perform well on general tasks, yet their deployment in specialized, high-stakes domains such as public health remains unreliable. Models hallucinate facts, combine multi-step evidence incorrectly, and retrieve context that corrupts downstream reasoning. Three root causes underlie these failures: training data that underrepresents rare but critical domain themes, reasoning that is opaque and unverifiable, and evidence use that cannot be re-checked when the evidence changes. This dissertation addresses all three. First, we introduce HDBSCAN_SSA, a coarse-to-fine topic-modeling pipeline that couples density-based clustering with a novel Semantic Seed Aggregation algorithm. Using samples drawn from a larger corpus of approximately 469,000 COVID-19 tweets, it discovers more coherent topics than LDA, k-means, and BERTopic, and topic-guided sampling under fixed labeling budgets of 10%, 20%, and 40% consistently improves downstream sentiment-classification accuracy and macro-F1 over random and BERTopic-based sampling. Second, we propose a two-stage logic-graph reasoning interface in which an LLM extracts a structured logic graph from a claim and its evidence table, and a graph attention network classifies the graph. On PubHealthTab, a public-health table-based fact-checking dataset of 1,942 claim-table pairs, the method reaches 88.1% accuracy and 0.853 macro-F1, outperforming chain-of-thought, zero-shot chain-of-thought, and self-ask baselines, with explanation quality (ROUGE-1 0.342, ROUGE-L 0.235) exceeding direct LLM outputs. Third, we propose a grounded verification framework that makes structured reasoning re-checkable against current evidence. A multi-agent pipeline binds extracted operations to exact table cells, and a symbolic layer checks whether the bound values still support the stated reasoning result, overriding the verdict when they do not. The framework achieves 95.6% clean accuracy, 9.8 percentage points above the strongest prompting baseline. Under evidence-side attacks it corrects 95.2% of 209 cell-swap and 91.1% of 538 numerical perturbations, and it resolves 74.9% of 538 claim-side negations through re-extraction. Together, these contributions form a data-to-reasoning-to-verification pipeline for domain AI whose outputs are grounded, inspectable, and robust.","abstract_has_math":false,"creators":["Bai, Yangxiao"],"institution":null,"degree_name":"Doctor of Philosophy (PhD)","degree_level":"Dissertation - Open Access","degree_discipline":"Electrical Engineering and Computer Science","degree_department":null,"school":null,"contributors":["Sungyong Jung"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":null,"date_issued":"","date_published":null,"updated_at":"2026-11-19T09:06:13Z","subjects":[],"languages":[],"rights":[],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://openprairie.sdstate.edu/etd2/2148","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://openprairie.sdstate.edu/etd2/2148","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://openprairie.sdstate.edu/do/oai/?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Aopenprairie.sdstate.edu%3Aetd2-3151","prefix":"oai_etdms"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Sungyong Jung"]},{"key":"dc:creator","label":"Author","values":["Bai, Yangxiao"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2026-08-05T07:00:00Z"]},{"key":"dc:date.created","label":"Dc Date Created","values":["2026-01-01T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical Engineering and Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation - Open Access"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy (PhD)"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://openprairie.sdstate.edu/etd2/2148"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Large language models (LLMs) perform well on general tasks, yet their deployment in specialized, high-stakes domains such as public health remains unreliable. Models hallucinate facts, combine multi-step evidence incorrectly, and retrieve context that corrupts downstream reasoning. Three root causes underlie these failures: training data that underrepresents rare but critical domain themes, reasoning that is opaque and unverifiable, and evidence use that cannot be re-checked when the evidence changes. This dissertation addresses all three. First, we introduce HDBSCAN_SSA, a coarse-to-fine topic-modeling pipeline that couples density-based clustering with a novel Semantic Seed Aggregation algorithm. Using samples drawn from a larger corpus of approximately 469,000 COVID-19 tweets, it discovers more coherent topics than LDA, k-means, and BERTopic, and topic-guided sampling under fixed labeling budgets of 10%, 20%, and 40% consistently improves downstream sentiment-classification accuracy and macro-F1 over random and BERTopic-based sampling. Second, we propose a two-stage logic-graph reasoning interface in which an LLM extracts a structured logic graph from a claim and its evidence table, and a graph attention network classifies the graph. On PubHealthTab, a public-health table-based fact-checking dataset of 1,942 claim-table pairs, the method reaches 88.1% accuracy and 0.853 macro-F1, outperforming chain-of-thought, zero-shot chain-of-thought, and self-ask baselines, with explanation quality (ROUGE-1 0.342, ROUGE-L 0.235) exceeding direct LLM outputs. Third, we propose a grounded verification framework that makes structured reasoning re-checkable against current evidence. A multi-agent pipeline binds extracted operations to exact table cells, and a symbolic layer checks whether the bound values still support the stated reasoning result, overriding the verdict when they do not. The framework achieves 95.6% clean accuracy, 9.8 percentage points above the strongest prompting baseline. Under evidence-side attacks it corrects 95.2% of 209 cell-swap and 91.1% of 538 numerical perturbations, and it resolves 74.9% of 538 claim-side negations through re-extraction. Together, these contributions form a data-to-reasoning-to-verification pipeline for domain AI whose outputs are grounded, inspectable, and robust."]},{"key":"dc:title","label":"Title","values":["A Framework for Reliable Domain AI: Knowledge-Grounded Data and Logic-Graph Reasoning"]}]}],"canonical_facts":{"dc:contributor":["Sungyong Jung"],"dc:creator":["Bai, Yangxiao"],"dc:date.available":["2026-08-05T07:00:00Z"],"dc:date.created":["2026-01-01T08:00:00Z"],"dc:description.abstract":["Large language models (LLMs) perform well on general tasks, yet their deployment in specialized, high-stakes domains such as public health remains unreliable. Models hallucinate facts, combine multi-step evidence incorrectly, and retrieve context that corrupts downstream reasoning. Three root causes underlie these failures: training data that underrepresents rare but critical domain themes, reasoning that is opaque and unverifiable, and evidence use that cannot be re-checked when the evidence changes. This dissertation addresses all three. First, we introduce HDBSCAN_SSA, a coarse-to-fine topic-modeling pipeline that couples density-based clustering with a novel Semantic Seed Aggregation algorithm. Using samples drawn from a larger corpus of approximately 469,000 COVID-19 tweets, it discovers more coherent topics than LDA, k-means, and BERTopic, and topic-guided sampling under fixed labeling budgets of 10%, 20%, and 40% consistently improves downstream sentiment-classification accuracy and macro-F1 over random and BERTopic-based sampling. Second, we propose a two-stage logic-graph reasoning interface in which an LLM extracts a structured logic graph from a claim and its evidence table, and a graph attention network classifies the graph. On PubHealthTab, a public-health table-based fact-checking dataset of 1,942 claim-table pairs, the method reaches 88.1% accuracy and 0.853 macro-F1, outperforming chain-of-thought, zero-shot chain-of-thought, and self-ask baselines, with explanation quality (ROUGE-1 0.342, ROUGE-L 0.235) exceeding direct LLM outputs. Third, we propose a grounded verification framework that makes structured reasoning re-checkable against current evidence. A multi-agent pipeline binds extracted operations to exact table cells, and a symbolic layer checks whether the bound values still support the stated reasoning result, overriding the verdict when they do not. The framework achieves 95.6% clean accuracy, 9.8 percentage points above the strongest prompting baseline. Under evidence-side attacks it corrects 95.2% of 209 cell-swap and 91.1% of 538 numerical perturbations, and it resolves 74.9% of 538 claim-side negations through re-extraction. Together, these contributions form a data-to-reasoning-to-verification pipeline for domain AI whose outputs are grounded, inspectable, and robust."],"dc:identifier":["https://openprairie.sdstate.edu/etd2/2148"],"dc:rights":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:title":["A Framework for Reliable Domain AI: Knowledge-Grounded Data and Logic-Graph Reasoning"],"thesis:degree_discipline":["Electrical Engineering and Computer Science"],"thesis:degree_level":["Dissertation - Open Access"],"thesis:degree_name":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-11-19T09:06:13Z"}