Authorship Identification for Literary Book Recommendations

Proceedings of the 27th International Conference on Computational Linguistics, pages 390–400 Santa Fe, New Mexico, USA, August 20-26, 2018. 390 Authorship Identification for Literary Book Recommendations Haifa Alharthi, Diana Inkpen, Stan Szpakowicz University of Ottawa, Ottawa, Ontario, Canada {halha060,Diana.Inkpen,szpak@eecs}.uottawa.ca Abstract Book recommender systems can help promote the practice of reading for pleasure, which has been declining in recent years. One factor that influences reading preferences is writing style. We propose a system that recommends books after learning their authors’ style. To our knowledge, this is the first work that applies the information learned by an author-identification model to book recommendations. We evaluated the system according to a top-k recommendation scenario. Our system gives better accuracy when compared with many state-of-the-art methods. We also conducted a qualitative analysis by checking if similar books/authors were annotated similarly by experts. 1 Introduction Recommender systems (RSs) are useful for internet users who may find it hard to choose from the multitude of available products and services. RSs predict how likely the target user is to be interested in an item which might have been unknown to her. In this work, we consider book recommender systems, which could be useful in libraries, schools, and on e-learning portals. Nowadays, with the introduction of e-books, readers can access inexpensive resources with little effort. It was expected that the act of reading for pleasure would become widespread, but statistics demonstrate the opposite; it is declining, particularly among young people.1 A number of studies have shown the benefits of reading. In a comparison between the well-being of 7,500 Canadian adult readers and non-readers, the former have been found statistically significantly more likely to report better health or mental health, to volunteer, and to feel strongly satisfied with life (Hill, 2013). Fiction has been shown to stimulate profound social communication (Mar and Oatley, 2008). Exposure to fiction correlates with a greater ability for empathy and social support (Mar et al., 2009). It also has a lingering biological effect, notably in the connectivity of the brain (Berns et al., 2013). It is therefore a worthwhile endeavor to deploy artificial intelligence techniques to spark people’s interest in books by recommending the right type of books. There are two main types of traditional recommender systems: collaborative filtering (CF) and content- based recommendation (CB). When a CF system predicts if a user likes a book, it relies on the ratings of all active users in the system. CF depends on a rating matrix which could be large and sparse, and that might lead to inferior recommendations. Sparsity occurs when items/users do not have enough ratings. It can be noticed in library records that many books are never checked out (as is, e.g., the case of 75% of the books in the library of Changsha University of Science and Technology (Yang et al., 2009)). A few other problems face CF systems: new items which are difficult to recommend until enough ratings have been received, the long tail when most items are overlooked because of few …

Content truncated to 500 words