Conversational Recommender System for Audio Listening Device Based on Ontology

Audio devices especially headphones, headsets, and earphones, are often used for daily activities. With the rapid development in the market, we can now choose a variety of specifications, shapes, and models that we may not know. To help us choose the right item, we usually use recommender systems such as conversational recommender systems (CRS). However, most CRS systems use the technical specifications of products as a reference to recommend items to users. Users who do not understand technical specifications and do not follow the development of the items find it difficult to choose an item that suits their needs. To overcome this problem is to directly ask about the functionality needs of users so that the recommender system can recommend items appropriately. This research evaluates an ontology-based CRS system as the main structure of the system, then embeds knowledge in it to facilitate conversational interaction with users, provide product recommendations, and explain the results of recommended items in the domain of audio listening devices. System evaluation is based on system performance and user satisfaction, accuracy for system performance gets a value of around 86.57% and gets positive results for user satisfaction. These results show that the system successfully provides recommendations that suit the users.