Improving cold-start recommendations using item-based stereotypes

Abstract

Recommender systems (RSs) have become key components driving the success of e-commerce and other platforms where revenue and customer satisfaction is dependent on the user’s ability to discover desirable items in large catalogues. As the number of users and items on a platform grows, the computational complexity and the sparsity problem constitute important challenges for any recommendation algorithm. In addition, the most widely studied filtering-based RSs, while effective in providing suggestions for established users and items, are known for their poor performance for the new user and new item (cold-start) problems. Stereotypical modelling of users and items is a promising approach to solving these problems. A stereotype represents an aggregation of the characteristics of the items or users which can be used to create general user or item classes. We propose a set of methodologies for the automatic generation of stereotypes to address the cold-start problem. The novelty of the proposed approach rests on the findings that stereotypes built independently of the user-to-item ratings improve both recommendation metrics and computational performance during cold-start phases. The resulting RS can be used with any machine learning algorithm as a solver, and the improved performance gains due to rate-agnostic stereotypes are orthogonal to the gains obtained using more sophisticated solvers. The paper describes how such item-based stereotypes can be evaluated via a series of statistical tests prior to being used for recommendation. The proposed approach improves recommendation quality under a variety of metrics and significantly reduces the dimension of the recommendation model.

Improving Cold Start Stereotype-Based Recommendation Using Deep Learning

“…The research gap that this paper wishes to fill is founded upon two distinct pillars. The first involves extending the methodology of stereotypes in pure cold start scenarios, which were independently developed in recent works [1], [11], and integrating them into deep learning frameworks. In the conclusions and future work of [1], the author anticipated testing the cold start stereotype approach with more sophisticated algorithms.…”

“…We also present metrics that go beyond the simpler accuracy, and intend to demonstrate the effect on accuracy, serendipity and fairness of the methodology researched. In summary, the primary contributions of this paper are as follows: bridging the identified research gap as outlined in the future work of [1]; offering additional evidence regarding the advantages and limitations of employing stereotypes; demonstrating the practical benefits and risks associated with implementing the stereotype framework within a deep learning solver; and emphasizing the evidence, benefits, and potential risks linked with utilizing a deep learning framework in the context of cold start scenarios. These techniques are readily applicable in various real-world applications of recommendation engines that encounter cold start issues when user and item features are accessible.…”

“…Cold start is one of the most challenging areas of operation of a RS, and several studies are dedicated to addressing it. Reference [1] reviews the new user and new item problems, and the existing literature, and introduces a novel way to address and improve cold start recommendations via stereotypes. The study suggests that the creation of general representations of users and items, known as stereotypes, from the available information, without relying on individual preferences or ratings, facilitates the development of a coordinate system that enhances two crucial aspects.…”

“…One of the most challenging tasks that a RS needs to perform is to provide recommendations to new users joining the platform (users that have not expressed any preferences), or recommend new items injected in the platform (items that have not been rated by any user). These two problems known in the literature as cold starts have been discussed in detail by the author in [1] and [2] and references within.…”

“…The rate of advancement of the application of deep learning to cold starts has been less pronounced and cold start remains an area where ad hoc approaches, often datasetdependent approaches, are developed. In this article we wish to bridge such gap by combining the stereotype based approach introduced in [1] with a deep learning framework. The stereotype based approach can be viewed as a ''model free'' manner to address cold-start recommendations, and in this context also as a way to reduce the dimensionality and computational cost of a deep neural network.…”