As no of search engines and websites increase, need of the hour is to have a efficient and user friendly search methods. Now a days lot websites owners indulge in Black Hat SEO techniques in order to reach of the search results. Even though search engines like Google, keeps on updating the search engine algorithm like Panda and Penguin updates, mechanisms to assisting web search and retrieval process are limited, due to lack of access to semantics of a documents and also due to the problems that exists in providing suitable search patterns. These limitations can be overcomed up to an extent by latest advancement in intelligent search by providing search engines with more intelligence and with user’s underlying knowledge. This projects aims to improve the search intelligence by using the natural language tool kit. By using natural language processing the user gets more accurate and correct results. The search engine is implemented using the languages python and PHP.
To improve the relevance shown for search queries, an additional factors like user rating and webpage usability is considered in the ranking of Web pages. Motivation behind this move is due to the shortcomings in existing search engines. They are
- Semantic gap between the user inputs and search engines keyword based results
- Proprietary Ranking algorithms used by search engines like Google, which can't be personalized for a user
- Lack of feedback mechanism regarding the relevancy of the returned pages
This project tries to addresses the issues specified above by allowing users to pass queries as taxonomies to a search engine which supplement the search term with antonyms and synonyms. It then transforms the search query into formats expected by search engines. Once search results are retrieved, application do post processing of the results along with other relevant components. Users can also specify the search engine, ranking of relevant pages which will be used in future decision making for the similar search query. This project implements a relevancy evaluation scheme based on user feedback, which boost above approaches. Application allows user's to create a taxonomy based on his search preferences. Context for the Web search is provided by this taxonomy. Results from different search engines based on the search query is populated into taxonomy. User rating of a web page is the primary component that effect the decision making. Evaluation of Web pages and the associated ranking is done by decision analytic techniques on 6 user based evaluation components which represent multiple evaluation criteria.
System architecture is given below.
Various stages of this project are
- Identifying the problem
- Additional information needed for Assessment
- Formulating Search queries
- Passing queries to search engine and retrieving the results
- Rating of results
- Decision Making on search results