Comparison of Recommender System Algorithms Focusing on the New-item and User-bias Problem

Hauger, Stefan; Tso, Karen H. L.; Schmidt-Thieme, Lars

doi:10.1007/978-3-540-78246-9_62

Cited by 13 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…genre, actors) this information can be used for recommendation. Especially when little rating information on a specific user or item is present like in the new-user and new-item problem (aka 'cold start problem'), attribute information can improve the prediction quality [12,5]. In this paper we deal with problems without any attribute information.…”

Section: Related Workmentioning

confidence: 99%

Online-updating regularized kernel matrix factorization models for large-scale recommender systems

Rendle

Schmidt-Thieme

2008

Proceedings of the 2008 ACM Conference on Recommender Systems

Self Cite

186

View full text Add to dashboard Cite

Regularized matrix factorization models are known to generate high quality rating predictions for recommender systems. One of the major drawbacks of matrix factorization is that once computed, the model is static. For real-world applications dynamic updating a model is one of the most important tasks. Especially when ratings on new users or new items come in, updating the feature matrices is crucial.In this paper, we generalize regularized matrix factorization (RMF) to regularized kernel matrix factorization (RKMF). Kernels provide a flexible method for deriving new matrix factorization methods. Furthermore with kernels nonlinear interactions between feature vectors are possible. We propose a generic method for learning RKMF models. From this method we derive an online-update algorithm for RKMF models that allows to solve the new-user/ new-item problem. Our evaluation indicates that our proposed online-update methods are accurate in approximating a full retrain of a RKMF model while the runtime of online-updating is in the range of milliseconds even for huge datasets like Netflix.

show abstract

Section: Related Workmentioning

confidence: 99%

Online-updating regularized kernel matrix factorization models for large-scale recommender systems

Rendle

Schmidt-Thieme

2008

Proceedings of the 2008 ACM Conference on Recommender Systems

Self Cite

186

View full text Add to dashboard Cite

show abstract

“…Content information, such as item attributes, were exploited to address such issues in previous methods [1], [2], [3]. However, items with similar attributes may be of different interest for the same user.…”

Section: Introductionmentioning

confidence: 99%

Addressing the Item Cold-Start Problem by Attribute-Driven Active Learning

Zhu

Lin

et al. 2020

IEEE Trans. Knowl. Data Eng.

111

View full text Add to dashboard Cite

In recommender systems, cold-start issues are situations where no previous events, e.g. ratings, are known for certain users or items. In this paper, we focus on the item cold-start problem. Both content information (e.g. item attributes) and initial user ratings are valuable for seizing users' preferences on a new item. However, previous methods for the item cold-start problem either 1) incorporate content information into collaborative filtering to perform hybrid recommendation, or 2) actively select users to rate the new item without considering content information and then do collaborative filtering. In this paper, we propose a novel recommendation scheme for the item cold-start problem by leverage both active learning and items' attribute information. Specifically, we design useful user selection criteria based on items' attributes and users' rating history, and combine the criteria in an optimization framework for selecting users. By exploiting the feedback ratings, users' previous ratings and items' attributes, we then generate accurate rating predictions for the other unselected users. Experimental results on two real-world datasets show the superiority of our proposed method over traditional methods.

show abstract

“…Plusieurs autres travaux se sont orientés vers la construction de modèles unificateurs qui prennent en compte les caractéristiques des utilisateurs et des documents ; (Basu et al, 1998) ont proposé l'application d'un classifieur d' attributs sur les utilisateurs couplé avec des informations sur le contenu des documents (ici le genre, les acteurs, le directeur… d'un film). Ainsi, il est possible de recommander de nouveaux documents sur la base des préférences des utilisateurs vis-à-vis de ces attributs sans disposer d'aucune évaluation au départ (Hauger et al, 2007). (Shein et al, 2002) proposent plusieurs modèles de probabilités où l'idée de base est l'analyse par la sémantique latente pour identifier de possibles liens sémantiques (une affinité particulière par exemple) cachés liant un document et un utilisateur.…”

Section: Les Systèmes De Filtrage Hybrideunclassified

Le filtrage collaboratif et le web 2.0. Etat de l'art

Oufaida¹,

Nouali²

2008

Document numérique

View full text Add to dashboard Cite

Le présent article fait le point sur l'état de l'art des systèmes de filtrage d'information. Il présente les différentes techniques de filtrage proposées dans la littérature dont le filtrage à base de contenu, le filtrage collaboratif et les modèles de filtrage hybrides. Il présente également les différentes limitations dont souffre toujours ce type de systèmes notamment la rareté des votes et le problème du démarrage à froid. Enfin, il discute les évolutions majeures prévisibles avec l'introduction des aspects du web 2.0 tels que les ontologies, les réseaux sociaux et les annotations. ABSTRACT. This paper overviews the recommender systems' field. It first presents different recommendation methods and techniques proposed in the literature from the content based recommender systems to the collaborative and hybrid recommendation models. It also describes the main limitations of the existing techniques such as the sparsity and the cold start problems. Finally, the new generation of semantic and social recommender systems are discussed which include several web 2.0 based extensions to overcome such weaknesses.

show abstract

Comparison of Recommender System Algorithms Focusing on the New-item and User-bias Problem

Cited by 13 publications

References 7 publications

Online-updating regularized kernel matrix factorization models for large-scale recommender systems

Online-updating regularized kernel matrix factorization models for large-scale recommender systems

Addressing the Item Cold-Start Problem by Attribute-Driven Active Learning

Le filtrage collaboratif et le web 2.0. Etat de l'art

Contact Info

Product

Resources

About