The Music Retrieval System Based on the Frequently-Used Rules of Chinese Text

Abstract

The conventional music retrieval system is mostly based on accurate matching of keywords, and it is difficult to satisfy the demand of intelligent retrieval. This paper presents a system of music retrieval based on the frequently-used rules of Chinese text. The system not only retains the conventional retrieval, but also is added the frequently-used rules of Chinese text that is “the attribute of the entity” (such as “the singer of the song”) into it. Firstly, the system segments the words of user’s query requests and checks that whether the user’s inputs comply with the rule of “the attribute of the entity”, and then it decides to adopt what kind of retrieval that is retrieval of text rules or conventional retrieval. The experiment proves that the system is better to satisfy the demand of user’s query requests than conventional music retrieval system while coping with the text rule of “the attribute of the entity”.