Research And Implementation Of Rulebase And Its Maintance In Relation Words Automatic Identification System

The research of Chinese information processing includes three important parts, which to the concrete is word processing, words processing and sentence processing. As an important unit of the Chinese sentence, processing of Compound sentence is an important component for Chinese information. Present research of which mainly to the extract of relation words and the study of this paper focus the research of rules, which is the fundamental part of automatically identifies.At present the research about compound sentense mainly includes two aspects:the identification of sub-clause and compound sentence. This paper studies how to establish the rulebase on the basis of the theory of the rule-based expert system, and research how to maintenance it, which is the theoretical basis to realize automatic identification of relational word. The innovation of this paper lies in the extraction process of the rules and the storage technology in which use different tables for different constraints, and proposed the maintenance method according the actual situation.According to the complexity of the Chinese langager, this paper first analyzes Chinese corpus of compound sentence, from which to explore the expression method of rules, which is to examine how the language phenomenon summarized to the rules described in text form.Discussion the character representation of the rules in describing which, including the data structure and classification according to the chareateristics.Due to the text rules could not identified by machine, this paper drafted unified formal standards to form the text rules into logical rules which can be identified by machine.Secondly, establish the rulebase according to the features of the rules.Mainly discusses the design of the physical structure and the physical storage of the rule base, by which makes the rules engine can call rules in the rule base to achieve the purpose of automatic identification. Last, during the use of the rule base, because of the operation to it, as to modify, delete and updated and so on, the consistency of rule base is difficult to be guaranteed, and with the increase of the rules and the content of it is more and more complexity, there are many disadvantages in artificial maintenance. Therefore we establish the maintenance system to detect the consistency and integrity of the rules, in order to ensure the accuracy and usefulness of the rule base. Finally the experiment tests the maintenance tool for the rulebase and ensures it by mannual check. The result shows that the detection method based on abstract inference tree model is reliable and feasiable.