Search Engine Ranking Algorithm Based On User Feedback

In the past few years, with the rapid development of the Internet, the role of search engine becomes more and more important, many internet users begin to choose the search engine as their first choice for access to network resources. However, the search engine's performance is not always satisfactory. When the user wants to find some information, search engines will return thousands of search results, in which only a small or no page can satisfy user's demand. How to understand the intent of the user's search, and how to find pages to meet user's requirement and then put the most relevant search result pages in the forefront, that has became a very important subject in the research of search engine.In this paper, we used user behavior in Web 2.0 as a research object, explored ways of user feedback, and proposed the concept of user feedback score. After studied the specific methods and corresponding realization for how user feedback to impact the final ranking of search results, a new sorting algorithm for search results based on neural network was presented. The algorithm used the BP neural network, and the neural network was trained by samples which were selected according to user feedback score. Traditional search results will be put into the trained neural network to compute, and a new ranking will be made according to relevance of the web page which indicated by the calculated results. This algorithm used the neural network's pattern recognition capabilities, combined user feedback and search engine effectively, making search results more in line with the user's search request.A new search engine website is designed and developed, which is based on user-defined relationship between the search keyword and the link address. In order to combine and utilize the user feedback, a secondary sort was implemented to improve the first search result which based on lucene search architecture. Web link with user feedback can be adjusted by resetting boost factor, and the rank of web link without user feedback can be recalculated by using neural network. The search results were properly improved comparing to the original results which can give reference for future research on the design of such algorithms.