【摘要】 随着信息社会尤其是互联网的发展,人们对文本检索的要求越来越高.作为对传统关键词匹配技术的改进,智能检索研究已经成为热点,并将是支撑下一代互联网的核心技术之一.将语义计算技术应用于文本检索,是智能检索的重要方向.文中在文本检索的两个关键技术(“标引”和“相似度计算”)中引入语义计算技术,用浅层语义来指导检索过程,提高检索准确率.针对“标引”技术,提出了语义树模型;针对“相似度计算”,基于语义张量的概念,结合自然语言处理的一些技术,提出三个可计算的窗口模型来近似语义张量的核心思想.以上工作在一定程度上实现了语义计算的功能.利用TREC数据集进行的评测表明,采用了语义计算技术后,文本检索的准确率可以提高10%左右.更多还原

【Abstract】 With the development of internet, users need a great improvement of the performance of text retrieval. As a kind of improvement of traditional keyword-based matching technologies, intelligent retrieval has become a research highlight, and will become one of the core technologies supporting internet of next generation. Introducing semantic computation into text retrieval is an important direction in the field of intelligent retrieval. This paper introduces semantic computations into two critical technologies in text retrieval(i.e. indexing and similarity computation), using shallow semantics to guide the process of retrieval, so that to improve the precision of text retrieval. For indexing, semantic tree model is put forward. For similarity computation, based on the concept of semantic tensor, using some NLP technologies, three window-based models are proposed to approximate the concept of semantic tensor. Based on the above technologies, semantic computation is realized in some degree in text retrieval. The evaluations in TREC data show that, the precision of text retrieval can be improved around 10% after using the two kinds of semantic computation technologies.更多还原