A study on the role of uninterested items in group recommendations

With the increase in the availability of network connectivity, e-commerce has grown rapidly. User products and reviews have grown multi-folds for these online markets, leading to the web’s data explosion. A recommender system overcomes this problem and provides personalised content to the user by mining a plethora of choices available on the web. In the group recommendations scenario, group members will have different tastes and achieving consensus is a nontrivial task. The common set of items taken from users’ preferences is vital to provide recommendations. In practice, recommendations are generated by considering only common interest items, with the remaining items not playing a role in the recommendation. The items that are not of interest to a user based on her implicit or explicit feedback are called uninterested items. We hypothesise that these uninterested items may influence the overall recommendations, which can help us better interpret users’ preferences. In this work, we also include the importance of uninterested items of the group members while generating the recommendation list. We have performed experiments over automatically identified groups and random groups to check the efficacy of the proposed models. We conducted experiments on publicly available real-world datasets. We found that incorporating uninterested item in group recommendations play a positive role in group recommendations while considering order and flexibility in user preferences. We observed that the overall group satisfaction scores are better in an automatically identified group.