Design And Implementation Of Image-text Recommendation System Based On Search Logs

With the widespread use of the Internet, great changes have taken place when we get access to the rich information. The Internet has greatly decreased the cost of sharing information, people can easily get what we want from the network. But with the explosive increase of data, picking up effective information from the Internet gets challenging. Under the circumstances, search engine draws more and more attention, displaying information from massive data.The search engine recommends related items with the searching query. This recommendation helps to optimize the queries and user interest analysis, so as to offer the personalized recommendation. This query recommendation service is provided a query recommendation system. Compared with other recommendation systems, query recommendation system aims to recommend interesting queries or a more clear expression of the query.This paper designs and implements a query recommendation system based on search logs. Different from the traditional query recommendation system, the results of this system is a combination of images and text, instead of simple text information. Thus, the colorful results of image-text recommendation system are more attractive, and lead to better user influence.This paper designs an image-text recommendation system with a variety of search log mining algorithms. Our method takes different information into account, including URL co-occurrence of user click logs, query vocabulary similarity, and session logs. This optimizes our recommendation system with interesting image-text entries.Experiments shows that our image-text recommendation system yields fantastic results. The user click logs proves that the system meets the actual demand of user interests and information display. From the theory and practice, our image-text recommendation system gets outstanding performance.