Litcius/Paper detail

Rating Prediction With Review Network Feedback: A New Direction in Recommendation

Supriyo Mandal, Abyayananda Maiti

2021IEEE Transactions on Computational Social Systems17 citationsDOI

Abstract

Recommendation systems usually make a personalized recommendation with explicit feedback (i.e., ratings, reviews, and description on products) or implicit feedback (i.e., searching activity, clicking products, and viewing products). Implicit feedback indicates a customer’s preferences, and explicit feedback indicates the satisfaction level from the purchased products. All these feedbacks are direct and generated by individual customers. Interpersonal relations or interactions among customers can be regarded as indirect feedback that can influence customers greatly. We introduce a new class of feedback named review network feedback to bridge this gap. This feedback is based on the concept of a review network where customers are nodes and their interaction in terms of reviewing the same product creates the edges. Review network feedback embodied the reliability and positional influence of customers in their review network. Extensive experiments on Amazon.com online review datasets establish the superiority of our model over popular baselines when we consider review network feedback.

Topics & Concepts

Computer scienceReliability (semiconductor)Bridge (graph theory)Interpersonal communicationClass (philosophy)Recommender systemProduct (mathematics)Human–computer interactionData miningArtificial intelligenceInformation retrievalPsychologyPhysicsPower (physics)Internal medicineMathematicsMedicineQuantum mechanicsGeometrySocial psychologyRecommender Systems and TechniquesComplex Network Analysis TechniquesAdvanced Graph Neural Networks
Rating Prediction With Review Network Feedback: A New Direction in Recommendation | Litcius