基于本体的认知推理模型及其应用研究

Simulation of human cognition process is cross-frontier of computer science and cognitive science, and ACT-R cognitive architecture is internationally recognized as an ideal model to simulate the process. Based on the long-term study of ontology and automated reasoning, this project will propose a cognitive reasoning model based on cognitive model, process ontology and rules for problem solving in domains which is absent in cognition and semantics. Firstly, construct a process ontology in some domain by extracting abstraction model and exemplar model from human problem solving processes, combining with ACT-R; Secondly, propose a cognitive reasoning model on process ontology to simulate human cognitive reasoning, in details, this model will be built on natural language processing and automated problem solving, in addition, it will describe declarative knowledge with ontology, and describe procedural knowledge with process ontology and Prolog rules; Finally validate this model by mathematical problem solving. By combining cognitive theory, process model with Prolog rules, this project can further improve the theory of cognitive reasoning, and provide a new method for combination reasoning. Consequently, it will promote knowledge sharing and reusing, realize high performance knowledge reasoning and provide new ideas for problem solving in domains.

模拟人的认知过程是计算机科学与认知科学的交叉前沿，其中ACT-R认知模型是目前国际公认的模拟该过程的理想模型。本项目在长期从事知识本体和自动推理研究工作的基础上，拟针对领域问题求解过程中缺乏认知性和语义的问题，研究一种集认知模型、过程本体和规则于一体的认知推理模型。首先，基于ACT-R认知模型，研究从人类解题过程中提取抽象模型和实例模型来构建领域过程本体；其次，研究基于过程本体的认知推理模型的构建，以此来模拟人的认知推理方式，此模型建立在自然语言处理和问题求解模型自动构建的基础上，拟用领域本体来描述陈述性的知识，用过程本体与Prolog规则结合来描述程序性的知识；最后，通过将其应用于数学问题的求解来验证此模型。本项目将认知理论、过程模型与规则推理结合起来，可进一步完善现有推理模型的认知理论，为认知推理建模提供新的方法，促进知识的共享与重用，实现高性能的知识推理，为领域问题求解提供新思路。

本项目基于ACT-R认知模型，研究从人类解题过程中提取抽象模型和实例模型来构建领域过程本体；研究基于过程本体的认知推理模型的构建，以此来模拟人的认知推理方式，用领域本体来描述陈述性的知识，用过程本体与Prolog规则结合来描述程序性的知识；提出一种集认知模型、过程本体和规则于一体的认知推理模型，并基于此认知推理模型，设计和实现了一个初中数学问题求解原型系统。一方面为认知推理建模提供新的方法，另一方面为领域问题求解提供新思路。