Employing Document Embeddings to Solve the “New Catalog” Problem in User Targeting, and Provide Explanations to the Users
In the current digital era, items that were consumed in a physical form are now available in online platforms that allow users to stream or buy them. However, not all of the items are available in digital form. When the companies that run these platforms acquire the rights to add a new catalog of items, the problem that arises is to identify who, among the customers, should be advertised with this new addition. Indeed, although the items may have existed for a long time, the preferences of the users for these items are not available. In this paper, we propose an approach that selects a set of users to target, to advertise a new catalog. In order to do so, we consider the textual description of these items and employ document embeddings (i.e., vector representations of a document) to model both the new catalog and the users. We also propose an approach to generate an explanation list to a user, represented by the top-n artists she evaluated that are most similar to the one of the new catalog. Experimental results show the effectiveness of both our targeting approach and of the explanation lists.
KeywordsUser targeting Document embeddings Explanation
This work is partially funded by Regione Sardegna under project NOMAD (Next generation Open Mobile Apps Development), through PIA - Pacchetti Integrati di Agevolazione “Industria Artigianato e Servizi” (annualità 2013).
- 4.Christou, D.: Feature extraction using latent dirichlet allocation and neural networks: a case study on movie synopses. CoRR abs/1604.01272 (2016)Google Scholar
- 5.Dhillon, I.S., Sra, S.: Generalized nonnegative matrix approximations with bregman divergences. In: Advances in Neural Information Processing Systems, NIPS 2005, 5–8 December, 2005, Vancouver, British Columbia, Canada, vol. 18, pp. 283–290 (2005)Google Scholar
- 6.Dumais, S.T.: Latent semantic analysis. ARIST 38(1), 188–230 (2004)Google Scholar
- 7.Hoffman, M.D., Blei, D.M., Bach, F.R.: Online learning for latent dirichlet allocation. In: Advances in Neural Information Processing Systems 23: 24th Annual Conference on Neural Information Processing Systems 2010, pp. 856–864. Curran Associates, Inc. (2010)Google Scholar
- 9.Lam, X.N., Vu, T., Le, T.D., Duong, A.D.: Addressing cold-start problem in recommendation systems. In: Proceedings of the 2nd International Conference on Ubiquitous Information Management and Communication, ICUIMC 2008, New York, pp. 208–211. ACM (2008)Google Scholar
- 10.Le, Q.V., Mikolov, T.: Distributed representations of sentences and documents. In: Proceedings of the 31th International Conference on Machine Learning, ICML 2014. JMLR Proceedings, vol. 32, pp. 1188–1196 (2014). JMLR.orgGoogle Scholar
- 12.Mikolov, T., Chen, K., Corrado, G., Dean, J.: Efficient estimation of word representations in vector space. CoRR abs/1301.3781 (2013)Google Scholar
- 13.Mikolov, T., Le, Q.V., Sutskever, I.: Exploiting similarities among languages for machine translation. CoRR abs/1309.4168 (2013)Google Scholar
- 14.Mikolov, T., Sutskever, I., Chen, K., Corrado, G.S., Dean, J.: Distributed representations of words and phrases and their compositionality. In: Proceedings of the 27th Annual Conference on Neural Information Processing Systems 2013, pp. 3111–3119 (2013)Google Scholar
- 15.Mikolov, T., Yih, W., Zweig, G.: Linguistic regularities in continuous space word representations. In: Human Language Technologies: Conference of the North American Chapter of the Association of Computational Linguistics, Proceedings, pp. 746–751. The Association for Computational Linguistics (2013)Google Scholar
- 17.Zhila, A., Yih, W., Meek, C., Zweig, G., Mikolov, T.: Combining heterogeneous models for measuring relational similarity. In: Human Language Technologies: Conference of the North American Chapter of the Association of Computational Linguistics, pp. 1000–1009. The Association for Computational Linguistics (2013)Google Scholar