Research On Smart Search And Abstract Extraction Technologies Based On Large Database

The introduction of full-text retrieval technology has greatly improved the mass-data retrieval efficiency of a database. However, in many mandarin application systems, especially some commanding information systems, the full-text retrieval technology is not widely used due to special reasons which prevents them from connecting to the Internet. The commanding information systems cannot execute a combined searching among global data in multiple database tables and from more than one field. They are also unable to flexibly display the contents that conform to the specific operational commands of commander’s interest. Therefore, there are urgent needs to devise a general searching engine that perfectly support a mandarin application system and intelligently apply full-text retrieval technology and abstract extraction. Mandarin intelligent searching and abstract extraction features should also be integrated into the system in order to facilitate the users with instantaneous searching, which improves their ability of information processing, quick reacting and decision making.The research focuses on the flexibility of the Chinese data searching in current application systems. Such systems is typified by XX commanding system, which cannot effectively deal with data in Chinese language. With the large database retrieval technologies as the core, the thesis has built a multi-table and multi-field searching engine designed for a global database. In this way, the realization and application of massive database searching are innovated and improved. Users can then retrieve useful information from massive data in a quicker way.Firstly, the large database retrieval mechanism was studied, the global intelligent search technology is proposed and a searching engine workflow is designed to improve the result display of most existing database facing mass-data retrieval. Secondly, the document de-formatting and the XML-based text link technologies are studied to achieve combined data-searching with multi-table and multi-field queries. Thirdly, the abstract extraction method based on regular expressions and the improved scheduling algorithm based on Oracle Text are proposed, which can effectively improve the quality of full-text search results. Fourthly, the query results display techniques are studied. The intelligent searching engine can then adjust the query results to meet specific requirements of users and display them in the most proper way.Finally, intelligent searching and abstract extraction tools are designed, the full-text index is established and a better man-machine interface is developed in the XX commanding system. With these improvements, users or developers of large database applications do not have to know the location of achieve data before searching for them. A general multi-table and multi-field searching within a global database is thus achieved. Automatic sorting, key-word highlighting, abstract displaying and other functions are also realized. According to their types and characteristics, query results can be presented in the most direct and appropriate way to users, so that the database search function can meet the demand of operational commanding, drilling, maneuvering and other office working needs of the XX commanding system.