Daniel Valcarce

Item-based relevance modelling of recommendations for getting rid of long tail products

The liquidation of long tail items can be assisted by recommender systems.We propose a probabilistic item-based Relevance Model (IRM2).IRM2 outperforms state-of-the-art recommenders for long tail liquidation. Recommender systems are a growing research field due to its immense potential application for helping users to select products and services. Recommenders are useful in a broad range of domains such as films, music, books, restaurants, hotels, social networks, news, etc. Traditionally, recommenders tend to promote certain products or services of a company that are kind of popular among the communities of users. An important research concern is how to formulate recommender systems centred on those items that are not very popular: the long tail products. A special case of those items are the ones that are product of an overstocking by the vendor. Overstock, that is, the excess of inventory, is a source of revenue loss. In this paper, we propose that recommender systems can be used to liquidate long tail products maximising the business profit. First, we propose a formalisation for this task with the corresponding evaluation methodology and datasets. And, then, we design a specially tailored algorithm centred on getting rid of those unpopular products based on item relevance models. Comparison among existing proposals demonstrates that the advocated method is a significantly better algorithm for this task than other state-of-the-art techniques.

A Study of Smoothing Methods for Relevance-Based Language Modelling of Recommender Systems

Language Models have been traditionally used in several fields like speech recognition or document retrieval. It was only recently when their use was extended to collaborative Recommender Systems. In this field, a Language Model is estimated for each user based on the probabilities of the items. A central issue in the estimation of such Language Model is smoothing, i.e., how to adjust the maximum likelihood estimator to compensate for rating sparsity. This work is devoted to explore how the classical smoothing approaches (Absolute Discounting, Jelinek-Mercer and Dirichlet priors) perform in the recommender task. We tested the different methods under the recently presented Relevance-Based Language Models for collaborative filtering, and compared how the smoothing techniques behave in terms of precision and stability. We found that Absolute Discounting is practically insensitive to the parameter value being an almost parameter-free method and, at the same time, its performance is similar to Jelinek-Mercer and Dirichlet priors.

Language Models for Collaborative Filtering Neighbourhoods

Language Models are state-of-the-art methods in Information Retrieval. Their sound statistical foundation and high effectiveness in several retrieval tasks are key to their current success. In this paper, we explore how to apply these models to deal with the task of computing user or item neighbourhoods in a collaborative filtering scenario. Our experiments showed that this approach is superior to other neighbourhood strategies and also very efficient. Our proposal, in conjunction with a simple neighbourhood-based recommender, showed a great performance compared to state-of-the-art methods (NNCosNgbr and PureSVD) while its computational complexity is low.

A Study of Priors for Relevance-Based Language Modelling of Recommender Systems

Probabilistic modelling of recommender systems naturally introduces the concept of prior probability into the recommendation task. Relevance-Based Language Models, a principled probabilistic query expansion technique in Information Retrieval, has been recently adapted to the item recommendation task with success. In this paper, we study the effect of the item and user prior probabilities under that framework. We adapt two priors from the document retrieval field and then we propose other two new probabilistic priors. Evidence gathered from experimentation indicates that a linear prior for the neighbour and a probabilistic prior based on Dirichlet smoothing for the items improve the quality of the item recommendation ranking.

Additive Smoothing for Relevance-Based Language Modelling of Recommender Systems

The use of Relevance-Based Language Models for top-N recommendation has become a promising line of research. Previous works have used collection-based smoothing methods for this task. However, a recent analysis on RM1 (an estimation of Relevance-Based Language Models) in document retrieval showed that this type of smoothing methods demote the IDF effect in pseudo-relevance feedback. In this paper, we claim that the IDF effect from retrieval is closely related to the concept of novelty in recommendation. We perform an axiomatic analysis of the IDF effect on RM2 concluding that this kind of smoothing methods also demotes the IDF effect in recommendation. By axiomatic analysis, we find that a collection-agnostic method, Additive smoothing, does not demote this property. Our experiments confirm that this alternative improves the accuracy, novelty and diversity figures of the recommendations.

Exploring Statistical Language Models for Recommender Systems

Even though there exist multiple approaches to build recommendation algorithms, algebraic techniques based on vector and matrix representations are predominant in the field. Notwithstanding the fact that these algebraic Collaborative Filtering methods have been demonstrated to be very effective in the rating prediction task, they do not generally provide good results in the top-N recommendation task. In this research, we return to the roots of recommender systems and we explore the relationship between Information Filtering and Information Retrieval. We think that probabilistic methods taken from the latter field such as statistical Language Models can be a more effective and formal way for generating personalised ranks of recommendations. We compare our improvements against several algebraic and probabilistic state-of-the-art algorithms and pave the way to future and promising research directions.

Efficient Pseudo-Relevance Feedback Methods for Collaborative Filtering Recommendation

Recently, Relevance-Based Language Models have been demonstrated as an effective Collaborative Filtering approach. Nevertheless, this family of Pseudo-Relevance Feedback techniques is computationally expensive for applying them to web-scale data. Also, they require the use of smoothing methods which need to be tuned. These facts lead us to study other similar techniques with better trade-offs between effectiveness and efficiency. Specifically, in this paper, we analyse the applicability to the recommendation task of four well-known query expansion techniques with multiple probability estimates. Moreover, we analyse the effect of neighbourhood length and devise a new probability estimate that takes into account this property yielding better recommendation rankings. Finally, we find that the proposed algorithms are dramatically faster than those based on Relevance-Based Language Models, they do not have any parameter to tune (apart from the ones of the neighbourhood) and they provide a better trade-off between accuracy and diversity/novelty.

LiMe: linear methods for pseudo-relevance feedback

Retrieval effectiveness has been traditionally pursued by improving the ranking models and by enriching the pieces of evidence about the information need beyond the original query. A successful method for producing improved rankings consists in expanding the original query. Pseudo-relevance feedback (PRF) has proved to be an effective method for this task in the absence of explicit users judgements about the initial ranking. This family of techniques obtains expansion terms using the top retrieved documents yielded by the original query. PRF techniques usually exploit the relationship between terms and documents or terms and queries. In this paper, we explore the use of linear methods for pseudo-relevance feedback. We present a novel formulation of the PRF task as a matrix decomposition problem which we called LiMe. This factorisation involves the computation of an inter-term similarity matrix which is used for expanding the original query. We use linear least squares regression with regularisation to solve the proposed decomposition with non-negativity constraints. We compare LiMe on five datasets against strong state-of-the-art baselines for PRF showing that our novel proposal achieves improvements in terms of MAP, nDCG and robustness index.

Combining Top-N Recommenders with Metasearch Algorithms

Given the diversity of recommendation algorithms, choosing one technique is becoming increasingly difficult. In this paper, we explore methods for combining multiple recommendation approaches. We studied rank aggregation methods that have been proposed for the metasearch task (i.e., fusing the outputs of different search engines) but have never been applied to merge top-N recommender systems. These methods require no training data nor parameter tuning. We analysed two families of methods: voting-based and score-based approaches. These rank aggregation techniques yield significant improvements over state-of-the-art top-N recommenders. In particular, score-based methods yielded good results; however, some voting techniques were also competitive without using score information, which may be unavailable in some recommendation scenarios. The studied methods not only improve the state of the art of recommendation algorithms but they are also simple and efficient.

Axiomatic Analysis of Language Modelling of Recommender Systems

Language Models constitute an effective framework for text retrieval tasks. Recently, it has been extended to various collaborative filtering tasks. In particular, relevance-based language models can be used for generating highly accurate recommendations using a memory-based approach. On the other hand, the query likelihood model has proven to be a successful strategy for neighbourhood computation. Since relevance-based language models rely on user neighbourhoods for producing recommendations, we propose to use the query likelihood model for computing those neighbourhoods instead of cosine similarity. The combination of both techniques results in a formal probabilistic recommender system which has not been used before in collaborative filtering. A thorough evaluation on three datasets shows that the query likelihood model provides better results than cosine similarity. To understand this improvement, we devise two properties that a good neighbourhood algorithm should satisfy. Our axiomatic analysis shows ...