A sequence-based filtering method for exhibition booth visit recommendations

As exhibitions are known to play important roles in marketing and sales promotion, the exhibition industry has grown significantly not only in the exhibition event size and frequency but also in the number of participating firms and visitors. While the challenge in assessing economic returns from exhibitions is being studied, it is agreed that the eventual success of an exhibition resides largely in its ability to meet the visitors' needs. Visitors use an exhibition as a source of information when searching for products or services. Though an exhibition provides an information-rich environment, however, visitors often get lost in the abundance of information. A specialized recommender system can be a good solution to information overload as it can guide visitors to right exhibition booths and help them collect necessary information. Traditional collaborative-filtering recommender systems, however, use only customers' rating or purchase records so that they do not capture exhibition visitors' temporal visit sequences and dynamic preferences. Moreover, due to the computation overhead, they cannot generate real-time recommendation in ubiquitous environments for exhibitions. In order to overcome these drawbacks, this study proposes a booth recommendation procedure that takes into consideration not only booth visit records but also visit sequences. Experiment results show that the proposed procedure achieves higher recommendation accuracy, faster computation, and more diversity than a typical collaborative-filtering recommender system. From the results, we conclude that the proposed booth recommendation procedure is suitable for real-time recommendation in ubiquitous exhibition environments.