arXiv - CS - Logic in Computer Science最新文献_第3页

Hypergraph rewriting and Causal structure of $λ-$calculus 超图重写与$λ-$微积分的因果结构

arXiv - CS - Logic in Computer Science Pub Date : 2024-09-02 DOI: arxiv-2409.01006

Utkarsh Bajaj

引用次数: 0

The Adjacent Fragment and Quine's Limits of Decision 邻接片段与奎因的决定限度

arXiv - CS - Logic in Computer Science Pub Date : 2024-09-02 DOI: arxiv-2409.01231

Bartosz Bednarczyk, Daumantas Kojelis, Ian Pratt-Hartmann

{"title":"The Adjacent Fragment and Quine's Limits of Decision","authors":"Bartosz Bednarczyk, Daumantas Kojelis, Ian Pratt-Hartmann","doi":"arxiv-2409.01231","DOIUrl":"https://doi.org/arxiv-2409.01231","url":null,"abstract":"We introduce the adjacent fragment AF of first-order logic, obtained by\u0000restricting the sequences of variables occurring as arguments in atomic\u0000formulas. The adjacent fragment generalizes (after a routine renaming) the\u0000two-variable fragment of first-order logic as well as the so-called fluted\u0000fragment. We show that the adjacent fragment has the finite model property, and\u0000that the satisfiability problem for its k-variable sub-fragment is in\u0000(k-1)-NExpTime. Using known results on the fluted fragment, it follows that the\u0000satisfiability problem for the whole adjacent fragment is Tower-complete. We\u0000additionally consider the effect of the adjacency requirement on the well-known\u0000guarded fragment of first-order logic, whose satisfiability problem is\u0000TwoExpTime-complete. We show that the satisfiability problem for the\u0000intersection of the adjacent and guarded adjacent fragments remains\u0000TwoExpTime-hard. Finally, we show that any relaxation of the adjacency\u0000condition on the allowed order of variables in argument sequences yields a\u0000logic whose satisfiability and finite satisfiability problems are undecidable.","PeriodicalId":501208,"journal":{"name":"arXiv - CS - Logic in Computer Science","volume":"49 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142192508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Support + Belief = Decision Trust 支持+信念=决策信任

arXiv - CS - Logic in Computer Science Pub Date : 2024-09-01 DOI: arxiv-2409.00666

Alessandro Aldini, Agata Ciabattoni, Dominik Pichler, Mirko Tagliaferri

引用次数: 0

Harnessing the Power of Semi-Structured Knowledge and LLMs with Triplet-Based Prefiltering for Question Answering 利用基于三重预过滤的半结构化知识和 LLMs 的力量进行问题解答

arXiv - CS - Logic in Computer Science Pub Date : 2024-09-01 DOI: arxiv-2409.00861

Derian Boer, Fabian Koch, Stefan Kramer

{"title":"Harnessing the Power of Semi-Structured Knowledge and LLMs with Triplet-Based Prefiltering for Question Answering","authors":"Derian Boer, Fabian Koch, Stefan Kramer","doi":"arxiv-2409.00861","DOIUrl":"https://doi.org/arxiv-2409.00861","url":null,"abstract":"Large Language Models (LLMs) frequently lack domain-specific knowledge and\u0000even fine-tuned models tend to hallucinate. Hence, more reliable models that\u0000can include external knowledge are needed. We present a pipeline, 4StepFocus,\u0000and specifically a preprocessing step, that can substantially improve the\u0000answers of LLMs. This is achieved by providing guided access to external\u0000knowledge making use of the model's ability to capture relational context and\u0000conduct rudimentary reasoning by themselves. The method narrows down\u0000potentially correct answers by triplets-based searches in a semi-structured\u0000knowledge base in a direct, traceable fashion, before switching to latent\u0000representations for ranking those candidates based on unstructured data. This\u0000distinguishes it from related methods that are purely based on latent\u0000representations. 4StepFocus consists of the steps: 1) Triplet generation for\u0000extraction of relational data by an LLM, 2) substitution of variables in those\u0000triplets to narrow down answer candidates employing a knowledge graph, 3)\u0000sorting remaining candidates with a vector similarity search involving\u0000associated non-structured data, 4) reranking the best candidates by the LLM\u0000with background data provided. Experiments on a medical, a product\u0000recommendation, and an academic paper search test set demonstrate that this\u0000approach is indeed a powerful augmentation. It not only adds relevant traceable\u0000background information from information retrieval, but also improves\u0000performance considerably in comparison to state-of-the-art methods. This paper\u0000presents a novel, largely unexplored direction and therefore provides a wide\u0000range of future work opportunities. Used source code is available at\u0000https://github.com/kramerlab/4StepFocus.","PeriodicalId":501208,"journal":{"name":"arXiv - CS - Logic in Computer Science","volume":"75 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142192511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

You-Only-Randomize-Once: Shaping Statistical Properties in Constraint-based PCG 只随机一次：在基于约束的 PCG 中塑造统计特性

arXiv - CS - Logic in Computer Science Pub Date : 2024-09-01 DOI: arxiv-2409.00837

Jediah Katz, Bahar Bateni, Adam M. Smith

{"title":"You-Only-Randomize-Once: Shaping Statistical Properties in Constraint-based PCG","authors":"Jediah Katz, Bahar Bateni, Adam M. Smith","doi":"arxiv-2409.00837","DOIUrl":"https://doi.org/arxiv-2409.00837","url":null,"abstract":"In procedural content generation, modeling the generation task as a\u0000constraint satisfaction problem lets us define local and global constraints on\u0000the generated output. However, a generator's perceived quality often involves\u0000statistics rather than just hard constraints. For example, we may desire that\u0000generated outputs use design elements with a similar distribution to that of\u0000reference designs. However, such statistical properties cannot be expressed\u0000directly as a hard constraint on the generation of any one output. In contrast,\u0000methods which do not use a general-purpose constraint solver, such as Gumin's\u0000implementation of the WaveFunctionCollapse (WFC) algorithm, can control output\u0000statistics but have limited constraint propagation ability and cannot express\u0000non-local constraints. In this paper, we introduce You-Only-Randomize-Once\u0000(YORO) pre-rolling, a method for crafting a decision variable ordering for a\u0000constraint solver that encodes desired statistics in a constraint-based\u0000generator. Using a solver-based WFC as an example, we show that this technique\u0000effectively controls the statistics of tile-grid outputs generated by several\u0000off-the-shelf SAT solvers, while still enforcing global constraints on the\u0000outputs.1 Our approach is immediately applicable to WFC-like generation\u0000problems and it offers a conceptual starting point for controlling the design\u0000element statistics in other constraint-based generators.","PeriodicalId":501208,"journal":{"name":"arXiv - CS - Logic in Computer Science","volume":"75 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142192512","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Unifying Model Execution and Deductive Verification with Interaction Trees in Isabelle/HOL 用 Isabelle/HOL 中的交互树统一模型执行和演绎验证

arXiv - CS - Logic in Computer Science Pub Date : 2024-08-28 DOI: arxiv-2408.15817

Simon Foster, Chung-Kil Hur, Jim Woodcock

{"title":"Unifying Model Execution and Deductive Verification with Interaction Trees in Isabelle/HOL","authors":"Simon Foster, Chung-Kil Hur, Jim Woodcock","doi":"arxiv-2408.15817","DOIUrl":"https://doi.org/arxiv-2408.15817","url":null,"abstract":"Model execution allows us to prototype and analyse software engineering\u0000models by stepping through their possible behaviours, using techniques like\u0000animation and simulation. On the other hand, deductive verification allows us\u0000to construct formal proofs demonstrating satisfaction of certain critical\u0000properties in support of high-assurance software engineering. To ensure\u0000coherent results between execution and proof, we need unifying semantics and\u0000automation. In this paper, we mechanise Interaction Trees (ITrees) in\u0000Isabelle/HOL to produce an execution and verification framework. ITrees are\u0000coinductive structures that allow us to encode infinite labelled transition\u0000systems, yet they are inherently executable. We use ITrees to create\u0000verification tools for stateful imperative programs, concurrent programs with\u0000message passing in the form of the CSP and Circus languages, and abstract\u0000system models in the style of the Z and B methods. We demonstrate how ITrees\u0000can account for diverse semantic presentations, such as structural operational\u0000semantics, a relational program model, and CSP's failures-divergences trace\u0000model. Finally, we demonstrate how ITrees can be executed using the Isabelle\u0000code generator to support the animation of models.","PeriodicalId":501208,"journal":{"name":"arXiv - CS - Logic in Computer Science","volume":"131 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142192514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Logic-Enhanced Language Model Agents for Trustworthy Social Simulations 用于可信社会模拟的逻辑增强语言模型代理

arXiv - CS - Logic in Computer Science Pub Date : 2024-08-28 DOI: arxiv-2408.16081

Agnieszka Mensfelt, Kostas Stathis, Vince Trencsenyi

{"title":"Logic-Enhanced Language Model Agents for Trustworthy Social Simulations","authors":"Agnieszka Mensfelt, Kostas Stathis, Vince Trencsenyi","doi":"arxiv-2408.16081","DOIUrl":"https://doi.org/arxiv-2408.16081","url":null,"abstract":"We introduce the Logic-Enhanced Language Model Agents (LELMA) framework, a\u0000novel approach to enhance the trustworthiness of social simulations that\u0000utilize large language models (LLMs). While LLMs have gained attention as\u0000agents for simulating human behaviour, their applicability in this role is\u0000limited by issues such as inherent hallucinations and logical inconsistencies.\u0000LELMA addresses these challenges by integrating LLMs with symbolic AI, enabling\u0000logical verification of the reasoning generated by LLMs. This verification\u0000process provides corrective feedback, refining the reasoning output. The\u0000framework consists of three main components: an LLM-Reasoner for producing\u0000strategic reasoning, an LLM-Translator for mapping natural language reasoning\u0000to logic queries, and a Solver for evaluating these queries. This study focuses\u0000on decision-making in game-theoretic scenarios as a model of human interaction.\u0000Experiments involving the Hawk-Dove game, Prisoner's Dilemma, and Stag Hunt\u0000highlight the limitations of state-of-the-art LLMs, GPT-4 Omni and Gemini 1.0\u0000Pro, in producing correct reasoning in these contexts. LELMA demonstrates high\u0000accuracy in error detection and improves the reasoning correctness of LLMs via\u0000self-refinement, particularly in GPT-4 Omni.","PeriodicalId":501208,"journal":{"name":"arXiv - CS - Logic in Computer Science","volume":"24 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142192513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Complete Axiomatisation of Equivalence for Discrete Probabilistic Programming 离散概率编程等价性的完整公理化

arXiv - CS - Logic in Computer Science Pub Date : 2024-08-27 DOI: arxiv-2408.14701

Robin Piedeleu, Mateo Torres-Ruiz, Alexandra Silva, Fabio Zanasi

引用次数: 0

Enhancing Analogical Reasoning in the Abstraction and Reasoning Corpus via Model-Based RL 通过基于模型的 RL 增强抽象与推理语料库中的类比推理能力

arXiv - CS - Logic in Computer Science Pub Date : 2024-08-27 DOI: arxiv-2408.14855

Jihwan Lee, Woochang Sim, Sejin Kim, Sundong Kim

引用次数: 0

arXiv - CS - Logic in Computer Science Pub Date : 2024-08-27 DOI: arxiv-2408.15087

Lorenzo Ceragioli, Fabio Gadducci, Giuseppe Lomurno, Gabriele Tedeschi

引用次数: 0