Relevant Feedback Based Accurate and Intelligent Retrieval on Capturing User Intention for Personalized Websites

A Web Search Personalization Based on Probability of Semantic Similarity between User Log and Query with Web Page

“…Moreover, ranking enhancements have not been addressed effectively in the Personalization Search, which has happened to a research route for many scientific researchers. Even user "behavior-based techniques" have improved ranking performance [1], [3]. For example, "click models" are precisely considered for personalized searches, where "clicks" of a realistic time period for a particular document suppose the user interest in such results [27], [28], while, it may not be applicable to the former users as such.…”

“…Also, if many users use the search engine, many "personalized PageRank" cannot be calculated and saved offline. But the experiment concludes that web search personalization using existing works [1], [7], [18], [19] results using "PageRank scores" is able to enhance the web search, however, the number of personalized results is limited because of computational requirements. In fact, instead of "personalizing" rankings to a particular person, this query changes the rankings depended on the topic of the query and query perspective.…”

“…However, because there is so much information to search, traditional search engine technology is becoming less and less useful. Many studies show that the majority of search terms for search engines are not short and clear, and users can have completely different intentions for the same query [1], [2], [3]. This can be described by the search term "office" or "Jaguar".…”

“…However, personalization can be improved through regular web searches rather than explicitly relying on specific user interests. Even the "Google" and other web search engines are currently attempting a personalized search [1], [5], [6].…”

“…In this situation, personalization based on the user's long-period interests might not offer satisfactory performance, since similar results may be returned repeatedly. Several works [1], [3], [13], [23] have considered using a user's active perspective to indicate short-period information requirements. The search perspective is either incorporated into the user profile or is made up of a different short-period user model or profile and is utilized to estimates the user's information needs.…”