Modeling item selection and relevance for accurate recommendations

Barbieri, Nicola; Costa, Gianni; Manco, Giuseppe; Ortale, Riccardo

doi:10.1145/2043932.2043941

Cited by 19 publications

(16 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The application of Information Retrieval metrics to recommender systems evaluation has been studied by several authors in the field (Barbieri et al, 2011;Breese et al, 1998;Cremonesi et al, 2010;Herlocker et al, 2004;Shani and Gunawardana, 2011). We elaborate here an experimental design framework that aims to synthesise commonalities and differences between studies, encompassing prior approaches and supporting new variants upon a common methodological grounding.…”

Section: Characterisation Of Design Alternatives In Ir Methodologies mentioning

confidence: 99%

“…For this reason, researchers are turning towards metrics and methodologies from the Information Retrieval (IR) field (Barbieri et al, 2011;Cremonesi et al, 2010;Herlocker et al, 2004), where ranking evaluation has been studied and standardised for decades. Yet, gaps remain between the methodological formalisation of tasks in both fields, which result in divergences in the adoption of IR methodologies, hindering the interpretation and comparability of empirical observations by different authors.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Statistical biases in Information Retrieval metrics for recommender systems

2017

View full text Add to dashboard Cite

There is an increasing consensus in the Recommender Systems community that the dominant error-based evaluation metrics are insufficient, and mostly inadequate, to properly assess the practical effectiveness of recommendations. Seeking to evaluate recommendation rankings—which largely determine the effective accuracy in matching user needs—rather than predicted rating values, Information Retrieval metrics have started to be applied for the evaluation of recommender systems. In this paper we analyse the main issues and potential divergences in the application of Information Retrieval methodologies to recommender system evaluation, and provide a systematic characterisation of experimental design alternatives for this adaptation. We lay out an experimental configuration framework upon which we identify and analyse specific statistical biases arising in the adaptation of Information Retrieval metrics to recommendation tasks, namely sparsity and popularity biases. These biases considerably distort the empirical measurements, hindering the interpretation and comparison of results across experiments. We develop a formal characterisation and analysis of the biases upon which we analyse their causes and main factors, as well as their impact on evaluation metrics under different experimental configurations, illustrating the theoretical findings with empirical evidence. We propose two experimental design approaches that effectively neutralise such biases to a large extent. We report experiments validating our proposed experimental variants, and comparing them to alternative approaches and metrics that have been defined in the literature with similar or related purposes.This work was partially supported by the national Spanish Government (grants nr. TIN2013-47090-C3-2 and TIN2016-80630-P). We wish to express our gratitude to the anonymous reviewers whose insightful and generous feedback guided us in producing an enhanced version of the paper beyond the amendmentof flaws and shortcomings

show abstract

Section: Characterisation Of Design Alternatives In Ir Methodologies mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Statistical biases in Information Retrieval metrics for recommender systems

2017

View full text Add to dashboard Cite

show abstract

“…The case-based component retrieves the past student most similar to the candidate being evaluated. Additionally, a Bayesian probabilistic model for explicit preference data was presented by Barbieri et al (2011). Their model proposes a generative process, which takes into account both item selection and rating emission to bring into communities those users who experience the same items and tend to adopt the same rating pattern.…”

Section: Recent Studies In Recommendation Techniquesmentioning

confidence: 99%

A survey of job recommender systems

Al-Otaibi¹

2012

Int. J. Phys. Sci.

124

View full text Add to dashboard Cite

The Internet-based recruiting platforms become a primary recruitment channel in most companies. While such platforms decrease the recruitment time and advertisement cost, they suffer from an inappropriateness of traditional information retrieval techniques like the Boolean search methods. Consequently, a vast amount of candidates missed the opportunity of recruiting. The recommender system technology aims to help users in finding items that match their personnel interests; it has a successful usage in e-commerce applications to deal with problems related to information overload efficiently. In order to improve the e-recruiting functionality, many recommender system approaches have been proposed. This article will present a survey of e-recruiting process and existing recommendation approaches for building personalized recommender systems for candidates/job matching.

show abstract

“…Things change substantially when considering the precision and recall accuracy metrics. Based on the results in [31,70] (discussed in the previous chapter), we consider here also LDA model, which has been identified as one of the top-performers in terms of recommendation accuracy. Notice that LDA was not included in the analysis of predictive accuracy, as it does not explicitly support a way to compute rating prediction.…”

Section: Discussionmentioning

confidence: 99%

Probabilistic Approaches to Recommendations

Barbieri¹,

Manco²,

Ritacco³

2014

Synthesis Lectures on Data Mining and Knowledge Discovery

Self Cite

View full text Add to dashboard Cite

This means that di↵erent user communities can exhibit di↵erent preference patterns on di↵erent item categories, where the identification of the latter depend on the considered community. This approach better models relationships between users and items and improve existing techniques both in terms of prediction and recommendation accuracy. During the development of this thesis, it has been my privilege to work with brilliant and wonderful people. Among all, my sincere thanks go to Giuseppe Manco, for his support and guidance over the past three years. I am grateful for his collaboration whose benefits go over and above the learning of data mining techniques. His wide perspective, motivation, passion and, above all, his work ethical standards, had a great impact in shaping my enthusiasm towards research.

show abstract

Modeling item selection and relevance for accurate recommendations

Cited by 19 publications

References 10 publications

Statistical biases in Information Retrieval metrics for recommender systems

Statistical biases in Information Retrieval metrics for recommender systems

A survey of job recommender systems

Probabilistic Approaches to Recommendations

Contact Info

Product

Resources

About