Research On Query Expansion Technology Of Content-based Flash Retrieval

The rapid development of information technology has changed our lives dramatically, andprovides a great convenience for our life and learning. Entering information-based society paidmore attention to the achievement of information education, and multimedia has been widelyapplied to various fields of education, which greatly improved the quality of education andeducational efficiency. Application of multimedia teaching depends on rich teaching resources.As a teaching resource, Flash, with small storage capacity, easy processing, convenient networktransmission, is more widely applied to teaching. In order to facilitate the search of teachingresources for masses of users, multimedia search engines have come into being. But it is rare touse search engine that specific to Flash animations, and most of search engines are based onkeyword. So the retrieval results, not conducive to users find the resources they want quickly,are quite broad. On this account, a more accurate retrieval Flash search engine is developed.Content-based Flash search engine is developed based on the above purpose. On the basisof reach about the Flash-contained structure, content properties were downloaded and stored inthe database. Coded and optimized in the database, the search interface was divided into twoparts: basic and advanced search. According to the user’s retrieval requests, correspondingrecords were selected in the database and displayed in the results.On the basis of deep understanding of retrieval system, the research progress of the wholeproject was combed and a website was set up conveniently for people who were interested inthe progress of the project. Established search engine has achieved its function basically, andobtained corresponding results according to the user’s retrieval condition, but the retrievalperformance remained to be further improved. Flash retrieval system was further optimized, andquery expansion technology was used to improve the retrieval performance of the system.Query expansion technology is to use a certain strategy to enter the query words and extend toform a new before the use of retrieval system. When the new query word is searched, morerelevant records could be obtained. Combining retrieval system for query expansion is mainlycontains two parts: keywords of the basic retrieval and primary colors of advanced retrieval. The two parts are extended by synonym dictionaries. The difference is that the former transfersdirectly synonym dictionary to the program to realize the extension of words. And the latterutilizes color dictionary built by itself. The dictionary is established from five models, such asRGB, HSV, HSI, Lab, and Luv, and formed by the degree of similarities between each model.Then the best color dictionary is selected from the five models and used as the last colordictionary to retrieve.In addition,the performance of retrieval system is improved by consummating database, forexample, the use of inverted file and program optimization.Preliminary experimental resultsshow that query expansion has improved the performance of retrieval system as a whole.