Review-based recommendation of attractive sentences in a novel for effective browsing

Abstract

Recently, the electronic book (e-book) market is growing rapidly and people have been able to choose e-books that they would like to read from a large amount of e-books. Therefore, techniques for finding efficiently one or more sufficient books that have something worth reading from vast numbers of e-books are demanded. In order to support a user to select books, many techniques for searching and recommending books have been proposed. However, the user would have to decide whether each book in candidates is worth reading. In this paper, we introduce a method for supporting a user to decide whether he/she read an e-book novel or not effectively. Our method recommends a user sentences that would attract and/or interest the user in an e-book novel. In our method, firstly, the attractiveness of each term in a novel is calculated based on reviews about the novel on the web. Then, the attractiveness of each sentence in the novel is calculated based on the attractiveness of the terms. This paper shows the experimental results of our method and discusses its effectiveness.