A new similarity measure for collaborative filtering based recommender systems

A new similarity measure for collaborative filtering based recommender systems. Gazdar, A. & Hidri, L. Knowledge-Based Systems, 188:105058, January, 2020.

Paper doi abstract bibtex 4 downloads

The objective of a recommender system is to provide customers with personalized recommendations while selecting an item among a set of products (movies, books, etc.). The collaborative filtering is the most used technique for recommender systems. One of the main components of a recommender system based on the collaborative filtering technique, is the similarity measure used to determine the set of users having the same behavior with regard to the selected items. Several similarity functions have been proposed, with different performances in terms of accuracy and quality of recommendations. In this paper, we propose a new simple and efficient similarity measure. Its mathematical expression is determined through the following paper contributions: 1) transforming some intuitive and qualitative conditions, that should be satisfied by the similarity measure, into relevant mathematical equations namely: the integral equation, the linear system of differential equations and a non-linear system and 2) resolving the equations to achieve the kernel function of the similarity measure. The extensive experimental study driven on a benchmark datasets shows that the proposed similarity measure is very competitive, especially in terms of accuracy, with regards to some representative similarity measures of the literature.

@article{gazdar_new_2020,
	title = {A new similarity measure for collaborative filtering based recommender systems},
	volume = {188},
	issn = {0950-7051},
	url = {http://www.sciencedirect.com/science/article/pii/S0950705119304484},
	doi = {10.1016/j.knosys.2019.105058},
	abstract = {The objective of a recommender system is to provide customers with
personalized recommendations while selecting an item among a set of
products (movies, books, etc.). The collaborative filtering is the most
used technique for recommender systems. One of the main components of a
recommender system based on the collaborative filtering technique, is the
similarity measure used to determine the set of users having the same
behavior with regard to the selected items. Several similarity functions
have been proposed, with different performances in terms of accuracy and
quality of recommendations. In this paper, we propose a new simple and
efficient similarity measure. Its mathematical expression is determined
through the following paper contributions: 1) transforming some intuitive
and qualitative conditions, that should be satisfied by the similarity
measure, into relevant mathematical equations namely: the integral
equation, the linear system of differential equations and a non-linear
system and 2) resolving the equations to achieve the kernel function of
the similarity measure. The extensive experimental study driven on a
benchmark datasets shows that the proposed similarity measure is very
competitive, especially in terms of accuracy, with regards to some
representative similarity measures of the literature.},
	journal = {Knowledge-Based Systems},
	author = {Gazdar, Achraf and Hidri, Lotfi},
	month = jan,
	year = {2020},
	keywords = {Collaborative filtering, Neighborhood based CF, Recommendation systems, Similarity measure},
	pages = {105058},
}

Downloads: 4

{"_id":"WmgG5drXtqHjGkjm8","bibbaseid":"gazdar-hidri-anewsimilaritymeasureforcollaborativefilteringbasedrecommendersystems-2020","authorIDs":[],"author_short":["Gazdar, A.","Hidri, L."],"bibdata":{"bibtype":"article","type":"article","title":"A new similarity measure for collaborative filtering based recommender systems","volume":"188","issn":"0950-7051","url":"http://www.sciencedirect.com/science/article/pii/S0950705119304484","doi":"10.1016/j.knosys.2019.105058","abstract":"The objective of a recommender system is to provide customers with personalized recommendations while selecting an item among a set of products (movies, books, etc.). The collaborative filtering is the most used technique for recommender systems. One of the main components of a recommender system based on the collaborative filtering technique, is the similarity measure used to determine the set of users having the same behavior with regard to the selected items. Several similarity functions have been proposed, with different performances in terms of accuracy and quality of recommendations. In this paper, we propose a new simple and efficient similarity measure. Its mathematical expression is determined through the following paper contributions: 1) transforming some intuitive and qualitative conditions, that should be satisfied by the similarity measure, into relevant mathematical equations namely: the integral equation, the linear system of differential equations and a non-linear system and 2) resolving the equations to achieve the kernel function of the similarity measure. The extensive experimental study driven on a benchmark datasets shows that the proposed similarity measure is very competitive, especially in terms of accuracy, with regards to some representative similarity measures of the literature.","journal":"Knowledge-Based Systems","author":[{"propositions":[],"lastnames":["Gazdar"],"firstnames":["Achraf"],"suffixes":[]},{"propositions":[],"lastnames":["Hidri"],"firstnames":["Lotfi"],"suffixes":[]}],"month":"January","year":"2020","keywords":"Collaborative filtering, Neighborhood based CF, Recommendation systems, Similarity measure","pages":"105058","bibtex":"@article{gazdar_new_2020,\n\ttitle = {A new similarity measure for collaborative filtering based recommender systems},\n\tvolume = {188},\n\tissn = {0950-7051},\n\turl = {http://www.sciencedirect.com/science/article/pii/S0950705119304484},\n\tdoi = {10.1016/j.knosys.2019.105058},\n\tabstract = {The objective of a recommender system is to provide customers with\npersonalized recommendations while selecting an item among a set of\nproducts (movies, books, etc.). The collaborative filtering is the most\nused technique for recommender systems. One of the main components of a\nrecommender system based on the collaborative filtering technique, is the\nsimilarity measure used to determine the set of users having the same\nbehavior with regard to the selected items. Several similarity functions\nhave been proposed, with different performances in terms of accuracy and\nquality of recommendations. In this paper, we propose a new simple and\nefficient similarity measure. Its mathematical expression is determined\nthrough the following paper contributions: 1) transforming some intuitive\nand qualitative conditions, that should be satisfied by the similarity\nmeasure, into relevant mathematical equations namely: the integral\nequation, the linear system of differential equations and a non-linear\nsystem and 2) resolving the equations to achieve the kernel function of\nthe similarity measure. The extensive experimental study driven on a\nbenchmark datasets shows that the proposed similarity measure is very\ncompetitive, especially in terms of accuracy, with regards to some\nrepresentative similarity measures of the literature.},\n\tjournal = {Knowledge-Based Systems},\n\tauthor = {Gazdar, Achraf and Hidri, Lotfi},\n\tmonth = jan,\n\tyear = {2020},\n\tkeywords = {Collaborative filtering, Neighborhood based CF, Recommendation systems, Similarity measure},\n\tpages = {105058},\n}\n\n","author_short":["Gazdar, A.","Hidri, L."],"key":"gazdar_new_2020","id":"gazdar_new_2020","bibbaseid":"gazdar-hidri-anewsimilaritymeasureforcollaborativefilteringbasedrecommendersystems-2020","role":"author","urls":{"Paper":"http://www.sciencedirect.com/science/article/pii/S0950705119304484"},"keyword":["Collaborative filtering","Neighborhood based CF","Recommendation systems","Similarity measure"],"metadata":{"authorlinks":{}},"downloads":4},"bibtype":"article","biburl":"https://api.zotero.org/users/6655/collections/TJPPJ92X/items?key=VFvZhZXIoHNBbzoLZ1IM2zgf&format=bibtex&limit=100","creationDate":"2020-10-23T07:37:04.720Z","downloads":4,"keywords":["collaborative filtering","neighborhood based cf","recommendation systems","similarity measure"],"search_terms":["new","similarity","measure","collaborative","filtering","based","recommender","systems","gazdar","hidri"],"title":"A new similarity measure for collaborative filtering based recommender systems","year":2020,"dataSources":["5Dp4QphkvpvNA33zi","jfoasiDDpStqkkoZB","BiuuFc45aHCgJqDLY"]}