基于遗传算法的词语语义相似度计算研究

语义相似度计算就是把词语间语言学上的信息映射为0到1之间的数值。基于知识本体的语义相似度计算方法,利用知识本体提供的信息,建立词语关系和语义相似度之间的函数关系,该方法可解释性强、使用简单,成为语义相似度计算的一类重要方法。提出了一种基于《同义词词林》的语义相似度计算模型,该模型运用遗传算法探索了《同义词词林》语义编码与语义相似度之间的内在联系,建立了更符合《同义词词林》中所蕴含的语义相似信息的函数关系式。该方法使用遗传算法搜索知识与语义相似度的函数表达式,克服了先验模型中函数形式及调节参数的局限性,所得计算结果与人工判定结果的皮尔逊相关系数为0.8645,为使用人工智能方法挖掘自然语言处理中的规律提供了一种新的思路和方法。

Research on Word Semantic Similarity Calculation Based on Genetic Algorithm

Semantic similarity calculation is to map linguistic information between words to values between 0 and 1.The semantic similarity calculation method based on the ontology uses the information provided by the ontology to establish the functional relationship between the word relationship and the semantic similarity.This method has strong interpretability and simple use,which becomes an important method of semantic similarity calculation.We propose a semantic similarity calculation model based on CiLin.The model uses genetic algorithm to explore the deep relationship between semantic coding and semantic similarity of CiLin,then establishes a function more consistent with the semantic similarity information contained in CiLin.This method uses genetic algorithm to search the function expression of knowledge and semantic similarity,and overcomes the limitation of function form and adjusting parameter in prior model.The Pearson correlation coefficient between the calculated result and the artificial judgment result is 0.8645.It provides a new way of thinking and method for explore the deep law in natural language processing by using artificial intelligence method.