Conversational Recommender Systems (CRS) convoys a new revolutionary recommendation system and relatively gains more research attention which ensues recommendation as a rich interactive process that enables the users to converse with the recommender via natural language. The most recent approaches in CRS implies on deep learning and Natural Language Processing (NLP). The emerging research scopes of CRS are exploration and exploitation trade-offs in the cold-start recommendation setting, attribute-centric conversational recommendation strategy-focused conversational recommendation, and dialogue understanding and response generation. Applicability of CRS includes Information Retrieval (IR), NLP, and Human-Computer Interaction (HCI).
The challenges in implementing CRS are question-based user preference elicitation, multi-turn conversational recommendation strategies, joint optimization, bias and debiasing methods in CRSs, multi-modal knowledge enrichment, evaluation, and user simulation. Most of the surveys on conversational recommendation systems comprise interaction modalities of CRS, the knowledge and data, the computational tasks, evaluation approaches, and future directions. Thereby, it is requisite to facilitate further enhancements in CRS to establish an applicable, reliable, and predictable conversational recommendation system by concentrating on task formalization, dataset collection, algorithm development, and evaluation.