Index term weighing computation method based on structural constraint in Chinese information retrieval

Abstract

The invention relates to a feature weighing computation method based on a structural constraint in a Chinese information retrieval. The method comprises the following steps: a. carrying out structuring processing to inquiry and obtaining a structuring inquiry result, wherein the structuring processing comprises one or a plurality of following steps: splitting words, carrying out part-of-speech tagging to splitted works, carrying out shallow parsing to inquiry or carrying out parsing to inquiry; b. determining an index term according to the structuring inquiry result and then determining an inquiry-context property set of the index term according to the structuring inquiry result which is adjacent to the index term and positioned in a word list; c. computing the weighing value of each property in the inquiry-context property set; d. combining the weighing values of all properties into property values of the index terms through a first composite function; and f. combining the property values of the index terms by a second composite function and obtaining the index term weighing. The method can compute the weighing accurately no matter whether the index terms exist in the word list or not. 本发明是有关于一种中文信息检索中基于结构约束的特征权重计算方法，包括以下步骤：a.对查询进行结构化处理，得到结构化查询结果；结构化处理包括：分词、对切分出的词进行词性标注、对查询进行浅层句法分析或对查询进行句法分析中一个或几个；b.根据述结构化查询结果确定索引词，然后根据与所述索引词相邻并位于词列表中的结构化查询的结果，确定所述索引词的查询—上下文属性集；c.计算查询—上下文属性集中每个属性的权重值；d.通过第一组合函数将各个属性的权重值组合成所述索引词的属性值；e.使用第二组合函数对所述索引词的属性值组合，得到所述索引词权重。无论索引词是否在词列表中，本发明的方法都能准确计算出其权重。