基于形式概念分析的关联数据知识表示与推理研究

关联数据是一种存在广泛、应用价值巨大的数据类型。本项目以关联数据为研究对象，基于“关联”视角，变单一形式背景为关联形式背景，变对象关联为概念关联，采用参数化、多粒度、批增量等策略，以发展属性约简、知识基、蕴含逻辑和推理规则为主要技术手段，重点解决关联数据知识表示的跨论域关联性和灵活性、知识更新的机制，以及知识推理的逻辑基础三个核心科学问题。主要研究内容包括：（1）参数化/多粒度融合形式背景的构建；（2）参数化/多粒度知识表示与推理；（3）知识更新的批增量方法；（4）关联数据的知识表示、获取与推理系统。本项目有望突破现有形式概念分析的研究局限，拓展其数据类型适用性，开创一个面向关联数据挖掘的形式概念分析新分支。研究成果将丰富关联数据知识发现与推理的理论与方法体系，对于扩展电子商务领域的智能推荐技术具有潜在应用价值。

Correlative data is a kind of data type that exists widely and has great practical applications. In this project, we will conduct a thorough and systematic study of correlative data from the “correlative” perspective, i.e., employing correlative formal context instead of single formal context, and emphasizing concept correlation instead of object correlation. To this end, it is necessary to solve the three core scientific problems, namely, cross-domain correlation and flexibility of knowledge representation of correlative data, knowledge updating mechanism, and logical foundation of knowledge reasoning, by adopting many strategies such as parameterization, multi-granularity, and batch incrementation, and by developing many new techniques such as attribute reduction, knowledge basis, implication logic and inference rules. .The main contents of the project include: (1) to construct parameterized and multi-granular fusion contexts from correlative data; (2) to study parameterized and multi-granular knowledge representation and reasoning; (3) to design batch incremental updating methods of parameterized and multi-granular knowledge; (4) to develop a knowledge representation, acquisition and reasoning system for correlative data. .This project is expected to break through the current research limitations of formal concept analysis, expand the applicability of its data types, and open a new branch of formal concept analysis for correlative data mining. The research results will also enrich the theory and method of knowledge discovery and reasoning of correlative data, and have a potential application value for extending intelligent recommendation techniques in the field of e-commerce.