中文分词中间件的设计与实现

hinese word segmentation is one of the fundamental technology in Chinese information processing, and it is frequently used in many fields. Application developers need to compare the efficiency and accuracy of different Chinese word segmentation systems before decide which one to adopt. There are many differences between Chinese word segmentation systems in many respects, such as programming language, algorithm, dictionary, function and interface, because these systems are developed by different project groups without a common standard. These differences lead to the increase of the learning cost and development cost of application developers. In addition, because of coupling, the developers have to reprogram if they want to replace one segementation system with the other segmentation system . In order to cut down the development cost, reduce coupling, this paper designs and implements a middleware for Chinese word segmentation. The middleware supports three Chinese word segmentation systems, including ICTCLAS, Jieba and Pychseg, by default. The middleware has good expansibility to support the other Chinese word segmentation systems. The middle provides uniform, standard, high-level interfaces to the upper layer application developers and integrators. Besides, the middleware also hides the heterogeneity of the underlying various Chinese word segmentation systems. In addition, application developers just need simple configuration to change one Chinese word segmentation system to the others. The middleware can be integrated into all kinds of complex applications. DCS, a search engine for digital resources, was developed to test and verify the performance of the middleware. The results shows that the middleware can be applied to the actual projects.