| With the advance of information technology, search engine has been applied in various areas, and it becomes one of the best methods that people get information. The popular search engines, like Google and Baidu, etc., are general purpose search engines for public needs. The information that they retrieve is based on unstructable information including web pages, blogs, and document files etc. But to enterprices, there lack suitable search engine for them to use, espiecally for structable information retrieval.Face to such a big market, by combining the feature of search engine and data warehouse, we design and develop the search engine system based on data warehouse. The system uses that standard data model of data warehouse and search engine model, and allows user to use natural language as query input. The enterprices can offer a new information retrieval form to query and display information, which reduces threshold of user’s IT skill requirements, and improves information sharing and deeply mining.This thesis’s main working and achievement are as follows.1. For structured data storage, a method is proposed for semantics level abstraction encapsulation, which is based on data warehouse mul-dimensional model. The method normalizes the storage structure of data.2. We optimize the search engine index structure to fit structure data query, and realize reversed word based on regular expression to increase parsing accuracy.3. The query parser is implemented by combing Chinese parsing algorithm and IKAnalyzer open source model, which makes end users to use nature language as query input and directly input business phrases to query. It realizes NLP expression parsing with complier technique to query structure data and automaticly create SQL statement.4. We optimize search engine, and obtain a design patten with more effective search interface.The Data Warehouse Search Engine system has been deployed in data center of Shanghai Tobacoo Corp. and achieved expected effect. |
PDF Full Text Request