Distributed Information Retrieval Research And Application

In the modern society, more and more data give us a challenge to solve how to improve the retrieval efficiency when we search, browse, evaluate and process these data in a large mobile world. In this dissertation, we focus on the following issues on text information retrieval, such as Text Concept Barycenter Model based on attribute theory, semantic judgement based on barycenter model, clustering algorithm to browse the database, Web information mining and distributed information retrieval based on mobile agent.In more detail, the contributions of the dissertation are as follows:(1)Construct a Text Concept Barycenter Model(TCBM). By analyzing the basic attributes of the documents and the relationship of these attributes, a conceptual uni-semigroup can describe the relation of concept sets based on the conjunctive operation. And it has been proved that TCBM is the isomorphic model as the conceptual uni-semigroup.(2)Based on this model, a text is made up of several concept vectors and represented as a barycenter. At the same time, a match criterion used to compute the similarity of texts and query is also described. By using this match criterion, we can also identify the semantic similarity of concepts.(3)Based on the high autonomy and the mobility of mobile agent, a mobile multiagent system prototype is proposed to solve the problem of browsing and organizing the distributed information in the distributed environment. Within this prototype, we layout a task agent, build information retrieval agent and create task object. And complete seveval retrieval functions.(4)According to the distributed concepts, we put forward an equal-value vector that can describe the condition of concepts in the whole text set. Then use an equal-value layer cluster algorithm to classify the set.(5)In order to discover the information of distributed environment, we analyze the features of Web pages and formally describe the configuration of the pages. By using the anchor tags in the HTML files, a common pattern that can help users create a new, perfect Web pages is extracted.(6)Based on the above research, we develop a system that is a subsystem of Content-based Multimedia Information Retrieval System.There are still many open aspects of information retrieval need to be discussed. As an attempt, we have to do further work.