Research On Website Product Category Based On Mental Model By Data Mining

"User-centric" is a prerequisite for the success of e-commerce sites. Currently,80%of e-commerce sites’users prefer using the catalog, and the catalog has been an important way for users to access information. Also, taking users’personalized needs into account, most e-commerce sites have introduced the "personalized recommendations" function, the essence of which is to recommend the most satisfied catalogs for users. Overall, the catalog has been the key part of the e-commerce sites, therefore both of the catalog designing and personalized recommending based on catalog should be "user-centric", should meet users’mental models. So, we need to solve the three following questions:How to gain users’mental models of websites? How to design websites catalogs based on users’mental models? How to personalized recommend based on catalog and users’mental models?To address these three key issues, this paper explores a new user research method. We gain users’mental models of websites catalog by data mining; because the log can reflect users’mental models directly. There are two key properties of mental model, which are conceptual similarity and spatiality. We analyze the relationship between users’expected catalog and the catalog of the websites from the conceptual similarity aspect by clustering and pathfinder networks in order to optimize the catalog so that users can be more satisfied with. And we also analyze the website from the spatiality aspect by clustering and multidimensional scaling. Learning from the collaborative filtering recommendation algorithm based on ideological content, this paper classifies users based on logs and regards the directory as the theme, and recommends the most relevant third level catalog in order to improve users’satisfaction, so that the websites can retain more users.