{"id":{"repo_id":"vt","oai_identifier":"oai:vtechworks.lib.vt.edu:10919/140897"},"canonical_url":"https://search.dev.ndltd.org/etd/vt/oai:vtechworks.lib.vt.edu:10919/140897","repository":{"repo_id":"vt","name":"Virginia Tech","base_url":"https://vtechworks.lib.vt.edu/oai/request"},"display":{"title":"Reinforcement Learning–Based Discrete Prompt Optimization for Neuro-Symbolic Structured Simplification of Complex Game Descriptions with Large Language Models","abstract":"This thesis investigates how large language models can be trained to perform structured simplification of complex, free-form game descriptions for the GameChangineer platform. The work formalizes simplification as a discrete prompt optimization problem and introduces a neuro-symbolic pipeline that maps raw natural language into controlled GameChangineer sentences via scenario normalization, retrieval-augmented code generation, and AST-based FACTS extraction. A reinforcement learning framework based on Proximal Policy Optimization optimizes discrete prompt edits using task-specific rewards that combine grammar compliance, semantic agreement with the FACTS contract, and compiler validity of the resulting games. Experiments on diverse arcade-style game descriptions show that the proposed GC-Repair and sentence correction agents significantly improve grammar-constrained generation, robustness to noisy user input, and end-to-end code correctness compared to direct LLM rewriting baselines.","abstract_html":"This thesis investigates how large language models can be trained to perform structured simplification of complex, free-form game descriptions for the GameChangineer platform. The work formalizes simplification as a discrete prompt optimization problem and introduces a neuro-symbolic pipeline that maps raw natural language into controlled GameChangineer sentences via scenario normalization, retrieval-augmented code generation, and AST-based FACTS extraction. A reinforcement learning framework based on Proximal Policy Optimization optimizes discrete prompt edits using task-specific rewards that combine grammar compliance, semantic agreement with the FACTS contract, and compiler validity of the resulting games. Experiments on diverse arcade-style game descriptions show that the proposed GC-Repair and sentence correction agents significantly improve grammar-constrained generation, robustness to noisy user input, and end-to-end code correctness compared to direct LLM rewriting baselines.","abstract_has_math":false,"creators":["Bhatt, Shubham Satyaprakash"],"institution":"Virginia Tech","degree_name":"Master of Science","degree_level":"masters","degree_discipline":"Computer Engineering","degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":[],"committee_chairs":["Hsiao, Michael S.","Abbott, Amos L."],"committee_members":["Wang, Yue J."],"year":2026,"date_issued":"2026-01-20","date_published":"2026-01-20","updated_at":"2026-07-22T22:18:53Z","subjects":["Text Simplification","Neuro-Symbolic NLP","Retrieval-Augmented Generation","Reinforcement Learning","Controlled Natural Language"],"languages":["en"],"rights":["In Copyright"],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["vt_gsexam:45464"],"render_values":[{"text":"vt_gsexam:45464","href":null,"code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/10919/140897","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.committeechair","label":"Committee Chair","values":["Hsiao, Michael S.","Abbott, Amos L."]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Wang, Yue J."]},{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:creator","label":"Author","values":["Bhatt, Shubham Satyaprakash"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-01-21T09:00:22Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2026-01-21T09:00:22Z"]},{"key":"dc:date.issued","label":"Date","values":["2026-01-20"]},{"key":"dc:publisher","label":"Institution","values":["Virginia Tech"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["masters"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Virginia Polytechnic Institute and State University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Text Simplification","Neuro-Symbolic NLP","Retrieval-Augmented Generation","Reinforcement Learning","Controlled Natural Language"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["In Copyright"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["vt_gsexam:45464"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10919/140897"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This thesis investigates how large language models can be trained to perform structured simplification of complex, free-form game descriptions for the GameChangineer platform. The work formalizes simplification as a discrete prompt optimization problem and introduces a neuro-symbolic pipeline that maps raw natural language into controlled GameChangineer sentences via scenario normalization, retrieval-augmented code generation, and AST-based FACTS extraction. A reinforcement learning framework based on Proximal Policy Optimization optimizes discrete prompt edits using task-specific rewards that combine grammar compliance, semantic agreement with the FACTS contract, and compiler validity of the resulting games. Experiments on diverse arcade-style game descriptions show that the proposed GC-Repair and sentence correction agents significantly improve grammar-constrained generation, robustness to noisy user input, and end-to-end code correctness compared to direct LLM rewriting baselines."]},{"key":"dc:description.abstractgeneral","label":"General Abstract","values":["This thesis explores how to help people describe video games in everyday English while still producing reliable computer code behind the scenes. Many students and beginners write game ideas as long, tangled sentences that are hard for both humans and software tools to turn into clear rules. I design a system that first asks a large language model (an AI system trained to work with text) to break these complex descriptions into smaller, simpler sentences and then convert them into executable game code. The system checks and repairs its own code, extracts the key facts about the game, and finally rewrites those facts back into clean, structured English that a teaching platform called GameChangineer can understand. This approach aims to make it easier for learners to practice computational thinking, receive precise feedback on their ideas, and quickly turn rough game concepts into working educational games."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Science"]},{"key":"dc:format.medium","label":"Dc Format Medium","values":["ETD"]},{"key":"dc:title","label":"Title","values":["Reinforcement Learning–Based Discrete Prompt Optimization for Neuro-Symbolic Structured Simplification of Complex Game Descriptions with Large Language Models"]}]}],"canonical_facts":{"dc:contributor.committeechair":["Hsiao, Michael S.","Abbott, Amos L."],"dc:contributor.committeemember":["Wang, Yue J."],"dc:contributor.department":["Electrical and Computer Engineering"],"dc:creator":["Bhatt, Shubham Satyaprakash"],"dc:date.accessioned":["2026-01-21T09:00:22Z"],"dc:date.available":["2026-01-21T09:00:22Z"],"dc:date.issued":["2026-01-20"],"dc:description.abstract":["This thesis investigates how large language models can be trained to perform structured simplification of complex, free-form game descriptions for the GameChangineer platform. The work formalizes simplification as a discrete prompt optimization problem and introduces a neuro-symbolic pipeline that maps raw natural language into controlled GameChangineer sentences via scenario normalization, retrieval-augmented code generation, and AST-based FACTS extraction. A reinforcement learning framework based on Proximal Policy Optimization optimizes discrete prompt edits using task-specific rewards that combine grammar compliance, semantic agreement with the FACTS contract, and compiler validity of the resulting games. Experiments on diverse arcade-style game descriptions show that the proposed GC-Repair and sentence correction agents significantly improve grammar-constrained generation, robustness to noisy user input, and end-to-end code correctness compared to direct LLM rewriting baselines."],"dc:description.abstractgeneral":["This thesis explores how to help people describe video games in everyday English while still producing reliable computer code behind the scenes. Many students and beginners write game ideas as long, tangled sentences that are hard for both humans and software tools to turn into clear rules. I design a system that first asks a large language model (an AI system trained to work with text) to break these complex descriptions into smaller, simpler sentences and then convert them into executable game code. The system checks and repairs its own code, extracts the key facts about the game, and finally rewrites those facts back into clean, structured English that a teaching platform called GameChangineer can understand. This approach aims to make it easier for learners to practice computational thinking, receive precise feedback on their ideas, and quickly turn rough game concepts into working educational games."],"dc:description.degree":["Master of Science"],"dc:format.medium":["ETD"],"dc:identifier.other":["vt_gsexam:45464"],"dc:identifier.uri":["https://hdl.handle.net/10919/140897"],"dc:language.iso":["en"],"dc:publisher":["Virginia Tech"],"dc:rights":["In Copyright"],"dc:rights.uri":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:subject":["Text Simplification","Neuro-Symbolic NLP","Retrieval-Augmented Generation","Reinforcement Learning","Controlled Natural Language"],"dc:title":["Reinforcement Learning–Based Discrete Prompt Optimization for Neuro-Symbolic Structured Simplification of Complex Game Descriptions with Large Language Models"],"dc:type":["Thesis"],"thesis:degree_discipline":["Computer Engineering"],"thesis:degree_level":["masters"],"thesis:degree_name":["Master of Science"],"thesis:institution_name":["Virginia Polytechnic Institute and State University"]},"updated_at":"2026-07-22T22:18:53Z"}