Research Of Rule Parser In Relation Words Of Compound Sentences Automatic Identification System

Chinese information processing, as a branch of computational linguistics, it plays an increasingly important role in Chinese information construction. Generally, people divide the process of Chinese information processing into three stages:word processing stage, words processing stage, sentence and text processing stage. Now, our country has made breakthrough progress in the word and words processing, and is efforts to achieving the sentence and text processing stage.Compound sentences as the entity unit of Chinese grammar, include rich and complex information, thus becomes the research emphasis of Chinese information processing. The relation word plays an important role in the field of modern Chinese compound sentences, what is an important topic of Chinese syntax and semantics researching. People tend to "seize the marks" to understand the compound sentences, computer also do by "seize the marks" to understand the compound sentences.In order to recognize the relation words of the compound sentences correctly, the rules for relation words are built in the relation words automatic identification system, which were used to decide whether the quasi-relationship word is relation word or not. In this paper, based on the high readability and flexibility of rules, rule parser is made, which is an important functional module of the relation words automatic identification system. Rule parser mainly contains two interfaces:1、The interface between rule base and rule engine. The interface by a new algorithm of character string matching—contains matching algorithm, what realize the matching between quasi-relationship words sequence identified by rule engine and strings of the rule base, and get all of the matching substrings to ensure the integrity of the matching.2、The interface between rule base and feature analyzer. The interface realize the parsing the rule constraints by calling the feature analyzer, and it can obtain the results of best matching rule by analytical results and recognize the quasi-relationship words according to the results.The features of rule parser are:extracting all the rules in the rule base automatically, completing the parsing task automatically, high accuracy, fast execution. It also simplify the work of rule engine that rule engine only need to evaluate the conclusions and recognize the quasi-relationship words according to the results of rule parser.