Learning complex users’ preferences for recommender systems
Recommender systems (RSs) have emerged as very useful tools to help customers with their decision-making process, find items of their interest, and alleviate the information overload problem. There are two different lines of approaches in RSs: (1) general recommenders with the main goal of discovering long-term users’ preferences, and (2) sequential recommenders with the main focus of capturing short-term users’ preferences in a session of user-item interaction (here, a session refers to a record of purchasing multiple items in one shopping event). While considering short-term users’ preferences may satisfy their current needs and interests, long-term users’ preferences provide users with the items that they may interact with, eventually. In this thesis, we first focus on improving the performance of general RSs. Most of the existing general RSs tend to exploit the users’ rating patterns on common items to detect similar users. The data sparsity problem (i.e. the lack of available information) is one of the major challenges for the current general RSs, and they may fail to have any recommendations when there are no common items of interest among users. We call this problem ‘data sparsity with no feedback on common items’ (DSW-n-FCI). To overcome this problem, we propose a personality-based RS in which similar users are identified based on the similarity of their personality traits. Next, we focus on one of the major difficulties that sequential recommenders are confronted with, which is how to model a noisy session. Current studies may assume that all the adjacent items in a session are highly dependent, which may not be practical in real-world scenarios because of the uncertainty of the customers’ shopping behaviour. A user-item interaction session may contain some irrelevant items which in turn may lead to false dependencies. Furthermore, long-term users’ preferences may be ignored by most of the existing sequential recommenders. To address this issue, we propose an attention-based framework to discriminately learn the dependencies among items in both long-term and short-term users’ preferences. Finally, sequential recommenders assume that each user-item interaction in a session is independent, which may be a very simplistic assumption. In real-world cases, people may have a particular purpose for buying successive items in a session. Unfortunately, a user’s behaviour pattern is not completely exploited by most of the current sequential recommenders, and they neglect the distinction between users’ purposes and their preferences in both users’ long-term and short-term item sets. We propose an approach using a purpose-specific attention unit (PSAU) to attend to important items differently depending on the user purposes and preferences for the next-item recommendation task. All of the proposed models in this thesis were tested on real-world datasets, and the experimental results demonstrated the effectiveness of these approaches as compared to the state-of-the-art RSs.