Automatic non-personalized book recommender algorithm for bookstore shelf management

摘要

Generally, the bookstore organizes bookshelf with book's category, writer, recommend, publish date and etc. However, the most important self is the recommend book shelf. The recommend bookshelf often set at the front of the store, because the customers can see the first and obviously. Traditionally the recommend bookshelf was based on the best seller and the newest. The recommend bookshelf was managed by the Non-Personalize recommendation, because the recommendation must affect to the most customers. This paper presented the book recommender algorithm for book recommender shelf. The purchase transaction log and the term in book title was used to predict the next best seller. The ranking of the predicted best seller book was used as the evaluation method, and the result shown that the best seller book was correctly identified up to 43% in the "Cookbooks, Food & Wine" category.