Search engine; Semantic web; YahooBoss API; Machine learning; Information retrieval; 2. Rocchio-based Re-ranking Method;
Using YahooBoss API and Rocchio-based method.
Methodology: We provide a web interface that can “remember” the links clicked by user. The search engine will extract information from snippets associated with clicked links to know what user is interested in. These user’s interests are then used in re-ranking Rocchio-based methods to order the result for the next queries of user. The raw results can be retrieved using YahooBOSS API. By this way, our search engine could avoid the ambiguity.
For example, supposed that a user quite often click on links that related to music. The user now makes a query “Yamaha”, which could be music instruments, motorbike, or company’s name. Our search engine will definitely return results related to music with higher priority than those related to car or company’s name. The search engine does it by firstly getting the snippet results from yahoo, and then re-ranks the result.
System specification:
The database system: MySQL
Web server: Apache
Web client: Firefox, IE, Chrome.
System architecture:
The main components of our peronalized search engine are depicted as in the attachment figure.
