Cross-Domain Tourist Service Recommendation Through Combinations of Explicit and Latent Features
| Autor: | Mingliang Qi, Yudong Tan, Jian Cao |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Service (business)
Information retrieval Computer science Bayesian probability 02 engineering and technology Business process reengineering Domain (software engineering) Matrix decomposition Product (business) Ranking 020204 information systems 0202 electrical engineering electronic engineering information engineering Selection (linguistics) 020201 artificial intelligence & image processing |
| Zdroj: | Lecture Notes in Computer Science ISBN: 9783319491776 APSCC |
| DOI: | 10.1007/978-3-319-49178-3_7 |
| Popis: | Nowadays, Online Travel Agents (OTA) can provide massive amount of travel services (such as flights, hotels), which also bring selection dilemma to users. Thus, it is critical to apply recommendation technology to help users. However, tourist service recommendation such as hotel recommendation is challenging because of the data sparsity problem. Moreover, generally, only implicit feedbacks (e.g. booking records) are available. In this paper, we propose to combine latent factors and explicit features across multiple domains for tourist service recommendation. Specifically, we extend Heterogeneous Matrix Factorization (HeteroMF) with explicit features, e.g. price of the tourist product. We also learn users’ preferences in different service domains with a transfer matrix to convert users’ preferences from one service domain to another. Furthermore, we train our model with respect to Bayesian Personalized Ranking (BPR) optimization criterion. Experiments on a real-world dataset show that our proposed model significantly outperforms HeteroMF and other baseline methods. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|