Abstract: To resolve the problem that general information retrieval models need exact match, a new query expansion was provided. With strong memory, learning and association ability, the chaotic neural network was used to learn from users query operations. Then users initial query condition was expanded and reconstructed so that the different retrieval results could be acquired to satisfy the users query motivation. The experimental results prove that the new model has better performance than the traditional neural network information retrieval models.

Key words: Chaotic Neural Network (CNN), Information Retrieval (IR), query expansion

摘要： 针对传统的信息检索模型只能进行精确匹配的问题，提出一种基于混沌神经网络模型的查询扩展方法，利用混沌神经网络较强的记忆性、学习性和联想性，对用户查询行为进行学习，从而对用户的初始查询进行扩展和重构，以得到符合不同用户的检索结果。与传统的神经网络信息检索模型的对比实验表明，新模型具有更高的查全率和查准率。

关键词: 混沌神经网络, 信息检索, 查询扩展