An Optimized Ranking Strategy for Expert Search on the Web with NLP Techniques

The most important thought of this strategy is to find experts for variety of domains on the web, where large numbers of WebPages and people names are considered. It has mainly two difficult issues: WebPages could be of unreliable quality and full of noises; the expertise evidences scattered in WebPages are usually formless and uncertain. We propose to control the large amount of co-occurrence information to evaluate relevance and reputation of a person name for a query topic. The objective is to design a system providing functionality of the expert search engine. NLP techniques can be applied for the same with name queries. Performance is optimized by effective crawling and deep parsing of web pages in order to adjust the association scores among people names and words. Global occurrences of experts are exercised so as to support the accuracy and relevancy of results. The proposed system also tries to boost performance by user rating based on user feedback. Finally we propose a unified approach for Person name Extraction where crawled data from web is applied to module which uses 3 class models which is used for building code for developing sequence models.