Construction and Implementation of the Quantifier-Noun Collocation Knowledge Base for Proofreading Based on Multiple Knowledge Sources

摘要

Due to specific semantic constraints between quantifiers and nouns and the frequent phenomena of quantifier-noun collocation error in real text, this paper proposes a new model of extracting quantifier-noun collocation and a new representation method of knowledge base. In this paper, by rules and statistical methods and the use of "The Grammatical Knowledge-base of Contemporary Chinese" and the "People's Daily" corpus resources, word-level quantifier-noun collocation knowledge base was established. On this basis, by analyzing the characteristics of the nouns and combining semantic class attributes in the "The Semantic Knowledge-base of Contemporary Chinese", semantic-level quantifier-noun collocation knowledge base was established. The experiment results show that the accuracy of checking quantifier-noun collocation errors reaches 77.38% by using quantifier-noun collocation knowledge base built by this method, and the scale of this knowledge base is expanded effectively.