Design And Implementation On The Retrieval System Of CMS

The main objective of content management is a lot of information for the evolution of capital productivity. And the content management system relies on the quality of content-based retrieval implemetion for its high-quality service. The project comes from"Research Theroy and method about dealing with magnanimity infomation"in academy,which the purpose is the design and Implementation of a flexible, which can be expanded content retrieval subsystems, and retrieval framework for integration of different retrieval technology.Meanwhile for types of structured, semi-structured and unstructured text of the three types, implementate a full-text retrieval subsystem.The project select vector space as theory model of the retrieval subsystem, because Lucene is a excellent open source of full-text retrieval of software package, we select Lucene as a implementation base. Vector space syncopate inquire and retrieve into tokens unit-term, what organized and retrieval index.Lucene implementate inverted index frame effectly. Through Lucene's segment, field, document, and term implementating inverted index frame high-effectly.Based on the requirement of magnanimity data manage systerm,we import hash list on the top of Lucene's inverted index of data structure,which design and implementate Barrel_hash data structure based on hash list.And using hash list once locational technology,optimize Lucene find word storage which base on comparative fuction.And using algorithm of weightpower and relevance ranking in vertor space,calculate and rank inquire about result.Through experiment indicate, based hash list of inverted data structure show well on update inverted index and retrieval optimize.