Key Techniques Of Search Engine Based On Ontology Services

With the rapid development of the service economy, modern services areincreasingly different from the traditional services. More and more information ofservices and service providers are advertised through the Internet. The service providerspublish their service information to the Internet and manage the information bythemselves or even complete the service discovery and trading on the network. Sodiscovery and extraction services and service providers from the Internet have became ademand. However, due to the service belong to different areas. Service providerspublish and manage their own service information, which is scattered in various cornersof the Internet and the type and quantity is enormous. This phenomenon gave atremendous challenge to how to quickly and accurately find the service and serviceproviders. This thesis puts forward an effective method to solve this problem.In this thesis, ontology-based service search engine (OBSSE) avoids search basedon keywords matching of traditional search engine. Firstly OBSSE create an domainontology includes service class, service provider class, domain theme class and so on.Secondly, after using the crawler downloads pages, identifying the web page’s theme inuse of the method that link text matches with theme library and theme characteristiclibrary. Then use the web content and HTML structure to recognize the theme of webpage. Through filtering the web page twice, it will improve the theme identificationaccuracy. Thirdly, parsing web page is mainly to extract service providers’ basicproperties and providers’ relationship from pages. Using the technology of namerecognition to identify the personal service providers. Using intelligent method, such asspeech tagging, to identify providers’ basic properties. And using the link relationshipand abstract services’ relationship to identify the relationship of providers. Finally, putthe service providers’ properties and relationship into database. This thesis’s mainlyresearch contents are service-oriented semantic crawler and ontology-based informationextraction. Crawler can identify service-related pages from the Internet and downloadthem to the local. Information extraction can extract service providers, their propertiesand their relationships, and save them into structured database. These information isdata resource of retrieval service.This thesis has done a lot of experiments whose resources are all web sites. These experiments mainly test the accuracy and recall rate of this search engine throughextracting information of service providers’ basic properties. The experimental datashow that, OBSSE’s identification rate about information of service providers canbasically reach more than90%, even up to98%.