Ontological Matchmaking in Recommender Systems

TL;DR

This paper presents an ontology-based matchmaking system for recommender systems in electronic marketplaces, addressing challenges like result presentation, ranking, and multi-provider querying, supported by experimental evaluation.

Contribution

The paper introduces a novel ontology-driven matchmaking system that handles complex real-life scenarios, including result grouping, asynchronous presentation, and multi-provider integration.

Findings

System effectively ranks and groups results based on user criteria

Experimental results show good performance and usability

Matchmaking strategies improve recommendation relevance

Abstract

The electronic marketplace offers great potential for the recommendation of supplies. In the so called recommender systems, it is crucial to apply matchmaking strategies that faithfully satisfy the predicates specified in the demand, and take into account as much as possible the user preferences. We focus on real-life ontology-driven matchmaking scenarios and identify a number of challenges, being inspired by such scenarios. A key challenge is that of presenting the results to the users in an understandable and clear-cut fashion in order to facilitate the analysis of the results. Indeed, such scenarios evoke the opportunity to rank and group the results according to specific criteria. A further challenge consists of presenting the results to the user in an asynchronous fashion, i.e. the 'push' mode, along with the 'pull' mode, in which the user explicitly issues a query, and displays the results. Moreover, an important issue to consider in real-life cases is the possibility of submitting a query to multiple providers, and collecting the various results. We have designed and implemented an ontology-based matchmaking system that suitably addresses the above challenges. We have conducted a comprehensive experimental study, in order to investigate the usability of the system, the performance and the effectiveness of the matchmaking strategies with real ontological datasets.