Research Of Personalized Recommendation On Air Travel Booking Website

In recent years, electronic commerce has become popular with the rapid development of Internet. Air travel booking website is an important category of e-commerce site, and the business is confronted with more and more intense competition. Therefore, more and more merchants are concerning on how to preserve old users and attract new users in order to increase sales of e-commerce effectively.To solve these problems, the key is getting user-interest pattern in order to better understand the customer's visit behavior, and providing more personalized recommendation service for customers. Currently, many sites have achieved a certain degree of personalized recommendation service, but for non-registered users, personalized recommendation is still in the initial stage. Since all users will leave a lot Web server's log which contains user visit information in the process of accessing the site, how to dig out useful user visit pattern from these massive logs(Web log mining), has become a hot and difficult topic of current research.In this paper, the problem is studied from the perspective of Web log mining based on an airline e-ticketing website. First, the paper outlines the web mining. Introduce cluster analysis and sequential pattern mining in the process of web log mining; establish theoretical principle for the following application. Second, detailed discuss data preprocessing process of web log mining, including data cleaning, user identification, session identification and path supplement and so on. To better realize session identification, the paper proposes a new method making use of access time and session reconstruction. In this method, the initial session sets are generated based on the access time. Then, the quality of session sets are optimized using a method of union and rupture. Third, the tradition matrix clustering algorithm is optimized, getting weight matrix cluster algorithm. This algorithm disperses user visit times, to obtain the weighted UrlID-UserID visit matrix. The algorithm not only better reflects the user visit, but also helps simplify data processing. Next, simplify each kind of user's session sequence, and discover each kind of user's maximum forward path database. Use PrefixSpan algorithm to find each kind of user frequent visit path. Finally, apply the session identification, clustering and sequence pattern mining's results to e-commerce recommendation system, and design a personalized recommendation system about air travel booking website.