From neighbors to global neighbors in collaborative filtering

Boumaza, Amine; Brun, Armelle

doi:10.1145/2330163.2330214

Cited by 8 publications

(9 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The system aims to effectively adapt and respond to immediate changes in users preference. Boumaza and Brun [4] present experiments and results on a standard benchmark data-set from the recommender system community that support the choice of the evolutionary approach and show that it leads to a high accuracy of recommendations and a high coverage, while dramatically reducing the size of the model (by 84%), also show that the evolutionary approach produces results able to generate accurate recommendations to unseen users, while easily allowing the insertion of new users in the system with little overhead.…”

Section: Related Workmentioning

confidence: 85%

Learning User’s Characteristics in Collaborative Filtering through Genetic Algorithms: Some New Results

Velez-Langs

Jiménez

2014

Advance Trends in Soft Computing

View full text Add to dashboard Cite

Abstract. This work presents an alternative approach (Genetic Algorithms approach) to traditional treatment of Recommender Systems (RSs). The work examines genetic algorithms possibilities to offer adaptive characteristics to these systems trough learning. The main goal, in addition to give a general view about RSs capabilities and possibilities, it is to provide a new example mechanism for to extend RSs learning capabilities (from user's personal characteristics), with the purpose of improve the effectiveness at time of to find recommendations and appropriate suggestions for particular individuals.Keywords: Recommender Systems, Genetic Algorithms, User-Adapted Interaction. IntroductionUsually in Internet, that is the main source of information for people that search answers and solutions for many situations, we use search engines typing key words, being too much the employed time for this task without finding the expected results. One alternative is to use Recommender Systems (RSs) [19,26,3] that offer to users an approach with their preferences, which are capable of suggesting the acquisition of any product. These systems filter the information, being classified in two main categories, depending on the information that use to suggest items. Those that only use information on the items and information with respect to objective user are call "Content Based" where the types, needs and inclinations of the users are determined in design time [9]. Alternative there are systems that do not use information on the items, but they do the suggestion using the known preferences of a group of users to predict the strangers preferences of a new user, the recommendations for this new user are based on these predictions [24]. This category is named "Collaborative Filtering" (CF).

show abstract

Section: Related Workmentioning

confidence: 85%

Learning User’s Characteristics in Collaborative Filtering through Genetic Algorithms: Some New Results

Velez-Langs

Jiménez

2014

Advance Trends in Soft Computing

View full text Add to dashboard Cite

show abstract

“…Some researches adjust Pearson coefficient, Cosine coefficient with information of time, items or users' interest in user-based CF algorithms and achieved good results [3][4].The other researches use the social network information to improve the calculation of similarity between users in the CF algorithms,such as information on friendship, tag data of social network and trustworthiness between users [5]. Accroding to the user-item diagram and user behavior data, Konstas I (2009) accessed to the friendship between users to improve the similarity, which have achieved good results and provided a new ides for the development of collaborative recommendation technology based on social networks [6].R. Zheng et.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Inspired by the idea of the global nearest neighbor in literature [10], based on the analysis of the different part user in long-tail distribution, this study attempts to construct a global core subset of users, containing only 30-40% users, but unchanging the performance of the collaborative recommendation algorithm largely. It provides a new method to solve the real-time and cold start problem in online recommendation system.…”

Section: Literature Reviewmentioning

confidence: 99%

Constructing the Core User Set for Collaborative Recommendation Based on Samples Selection Idea

Li¹,

Yu²,

Shuyan³

2016

IJUNESST

View full text Add to dashboard Cite

As an effective recommendation technology to solve "information overload" problem, collaborative filtering has widly attracted attention of scholars from various fields. First, this paper proposes a method for measuring the recommending ability of user based on popularity and long-tailed distribution. Then, a global core user set is constructed for recommadition based on the recommending ability and samples selection idea in data mining, aimming to take advantage of users in the different part of the long tail distribution and reduce computing complexity of the algorithm without lowing the recommended performance. Experimental results show that the algorithm is effective and can be used to solve the real-time problem and cold start recommendation.

show abstract

“…Approaches like (Boumaza and Brun, 2012), use a global set of neighbors, to predict the ratings for all the users. Therefore the space occupied by the model is reduced.…”

Section: Neighbors Selection In Collaborative Filteringmentioning

confidence: 99%

Using Collaborative Filtering to Overcome the Curse of Dimensionality when Clustering Users in a Group Recommender System

Boratto

Carta

2014

Proceedings of the 16th International Conference on Enterprise Information Systems

View full text Add to dashboard Cite

A characteristic of most datasets is that the number of data points is much lower than the number of dimensions (e.g., the number of movies rated by a user is much lower than the number of movies in a dataset). Dealing with high-dimensional and sparse data leads to problems in the classification process, known as curse of dimensionality. Previous researches presented approaches that produce group recommendations by clustering users in contexts where groups are not available. In the literature it is widely-known that clustering is one of the classification forms affected by the curse of dimensionality. In this paper we propose an approach to remove sparsity from a dataset before clustering users in group recommendation. This is done by using a Collaborative Filtering approach that predicts the missing data points. In such a way, it is possible to overcome the curse of dimensionality and produce better clusterings. Experimental results show that, by removing sparsity, the accuracy of the group recommendations strongly increases with respect to a system that works on sparse data.

show abstract

From neighbors to global neighbors in collaborative filtering

Cited by 8 publications

References 19 publications

Learning User’s Characteristics in Collaborative Filtering through Genetic Algorithms: Some New Results

Learning User’s Characteristics in Collaborative Filtering through Genetic Algorithms: Some New Results

Constructing the Core User Set for Collaborative Recommendation Based on Samples Selection Idea

Using Collaborative Filtering to Overcome the Curse of Dimensionality when Clustering Users in a Group Recommender System

Contact Info

Product

Resources

About