Recommendations Using Information from Multiple Association Rules: A Probabilistic Approach. Ghoshal, A., Menon, S., & Sarkar, S. Information Systems Research, 26(3):532–551, July, 2015. ![Recommendations Using Information from Multiple Association Rules: A Probabilistic Approach [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex 2 downloads Business analytics has evolved from being a novelty used by a select few to an accepted facet of conducting business. Recommender systems form a critical component of the business analytics toolkit and, by enabling firms to effectively target customers with products and services, are helping alter the e-commerce landscape. A variety of methods exist for providing recommendations, with collaborative filtering, matrix factorization, and association-rule-based methods being the most common. In this paper, we propose a method to improve the quality of recommendations made using association rules. This is accomplished by combining rules when possible and stands apart from existing rule-combination methods in that it is strongly grounded in probability theory. Combining rules requires the identification of the best combination of rules from the many combinations that might exist, and we use a maximum-likelihood framework to compare alternative combinations. Because it is impractical to apply the maximum likelihood framework directly in real time, we show that this problem can equivalently be represented as a set partitioning problem by translating it into an information theoretic context—the best solution corresponds to the set of rules that leads to the highest sum of mutual information associated with the rules. Through a variety of experiments that evaluate the quality of recommendations made using the proposed approach, we show that (i) a greedy heuristic used to solve the maximum likelihood estimation problem is very effective, providing results comparable to those from using the optimal set partitioning solution; (ii) the recommendations made by our approach are more accurate than those made by a variety of state-of-the-art benchmarks, including collaborative filtering and matrix factorization; and (iii) the recommendations can be made in a fraction of a second on a desktop computer, making it practical to use in real-world applications.
@article{ghoshal_recommendations_2015,
title = {Recommendations {Using} {Information} from {Multiple} {Association} {Rules}: {A} {Probabilistic} {Approach}},
volume = {26},
issn = {1047-7047},
url = {http://pubsonline.informs.org/doi/abs/10.1287/isre.2015.0583},
doi = {10.1287/isre.2015.0583},
abstract = {Business analytics has evolved from being a novelty used by a select few
to an accepted facet of conducting business. Recommender systems form a
critical component of the business analytics toolkit and, by enabling
firms to effectively target customers with products and services, are
helping alter the e-commerce landscape. A variety of methods exist for
providing recommendations, with collaborative filtering, matrix
factorization, and association-rule-based methods being the most common.
In this paper, we propose a method to improve the quality of
recommendations made using association rules. This is accomplished by
combining rules when possible and stands apart from existing
rule-combination methods in that it is strongly grounded in probability
theory. Combining rules requires the identification of the best
combination of rules from the many combinations that might exist, and we
use a maximum-likelihood framework to compare alternative combinations.
Because it is impractical to apply the maximum likelihood framework
directly in real time, we show that this problem can equivalently be
represented as a set partitioning problem by translating it into an
information theoretic context—the best solution corresponds to the set of
rules that leads to the highest sum of mutual information associated with
the rules. Through a variety of experiments that evaluate the quality of
recommendations made using the proposed approach, we show that (i) a
greedy heuristic used to solve the maximum likelihood estimation problem
is very effective, providing results comparable to those from using the
optimal set partitioning solution; (ii) the recommendations made by our
approach are more accurate than those made by a variety of
state-of-the-art benchmarks, including collaborative filtering and matrix
factorization; and (iii) the recommendations can be made in a fraction of
a second on a desktop computer, making it practical to use in real-world
applications.},
number = {3},
urldate = {2015-09-19},
journal = {Information Systems Research},
author = {Ghoshal, Abhijeet and Menon, Syam and Sarkar, Sumit},
month = jul,
year = {2015},
pages = {532--551},
}
Downloads: 2
{"_id":"CxWoeywFKKQEpnnFL","bibbaseid":"ghoshal-menon-sarkar-recommendationsusinginformationfrommultipleassociationrulesaprobabilisticapproach-2015","authorIDs":[],"author_short":["Ghoshal, A.","Menon, S.","Sarkar, S."],"bibdata":{"bibtype":"article","type":"article","title":"Recommendations Using Information from Multiple Association Rules: A Probabilistic Approach","volume":"26","issn":"1047-7047","url":"http://pubsonline.informs.org/doi/abs/10.1287/isre.2015.0583","doi":"10.1287/isre.2015.0583","abstract":"Business analytics has evolved from being a novelty used by a select few to an accepted facet of conducting business. Recommender systems form a critical component of the business analytics toolkit and, by enabling firms to effectively target customers with products and services, are helping alter the e-commerce landscape. A variety of methods exist for providing recommendations, with collaborative filtering, matrix factorization, and association-rule-based methods being the most common. In this paper, we propose a method to improve the quality of recommendations made using association rules. This is accomplished by combining rules when possible and stands apart from existing rule-combination methods in that it is strongly grounded in probability theory. Combining rules requires the identification of the best combination of rules from the many combinations that might exist, and we use a maximum-likelihood framework to compare alternative combinations. Because it is impractical to apply the maximum likelihood framework directly in real time, we show that this problem can equivalently be represented as a set partitioning problem by translating it into an information theoretic context—the best solution corresponds to the set of rules that leads to the highest sum of mutual information associated with the rules. Through a variety of experiments that evaluate the quality of recommendations made using the proposed approach, we show that (i) a greedy heuristic used to solve the maximum likelihood estimation problem is very effective, providing results comparable to those from using the optimal set partitioning solution; (ii) the recommendations made by our approach are more accurate than those made by a variety of state-of-the-art benchmarks, including collaborative filtering and matrix factorization; and (iii) the recommendations can be made in a fraction of a second on a desktop computer, making it practical to use in real-world applications.","number":"3","urldate":"2015-09-19","journal":"Information Systems Research","author":[{"propositions":[],"lastnames":["Ghoshal"],"firstnames":["Abhijeet"],"suffixes":[]},{"propositions":[],"lastnames":["Menon"],"firstnames":["Syam"],"suffixes":[]},{"propositions":[],"lastnames":["Sarkar"],"firstnames":["Sumit"],"suffixes":[]}],"month":"July","year":"2015","pages":"532–551","bibtex":"@article{ghoshal_recommendations_2015,\n\ttitle = {Recommendations {Using} {Information} from {Multiple} {Association} {Rules}: {A} {Probabilistic} {Approach}},\n\tvolume = {26},\n\tissn = {1047-7047},\n\turl = {http://pubsonline.informs.org/doi/abs/10.1287/isre.2015.0583},\n\tdoi = {10.1287/isre.2015.0583},\n\tabstract = {Business analytics has evolved from being a novelty used by a select few\nto an accepted facet of conducting business. Recommender systems form a\ncritical component of the business analytics toolkit and, by enabling\nfirms to effectively target customers with products and services, are\nhelping alter the e-commerce landscape. A variety of methods exist for\nproviding recommendations, with collaborative filtering, matrix\nfactorization, and association-rule-based methods being the most common.\nIn this paper, we propose a method to improve the quality of\nrecommendations made using association rules. This is accomplished by\ncombining rules when possible and stands apart from existing\nrule-combination methods in that it is strongly grounded in probability\ntheory. Combining rules requires the identification of the best\ncombination of rules from the many combinations that might exist, and we\nuse a maximum-likelihood framework to compare alternative combinations.\nBecause it is impractical to apply the maximum likelihood framework\ndirectly in real time, we show that this problem can equivalently be\nrepresented as a set partitioning problem by translating it into an\ninformation theoretic context—the best solution corresponds to the set of\nrules that leads to the highest sum of mutual information associated with\nthe rules. Through a variety of experiments that evaluate the quality of\nrecommendations made using the proposed approach, we show that (i) a\ngreedy heuristic used to solve the maximum likelihood estimation problem\nis very effective, providing results comparable to those from using the\noptimal set partitioning solution; (ii) the recommendations made by our\napproach are more accurate than those made by a variety of\nstate-of-the-art benchmarks, including collaborative filtering and matrix\nfactorization; and (iii) the recommendations can be made in a fraction of\na second on a desktop computer, making it practical to use in real-world\napplications.},\n\tnumber = {3},\n\turldate = {2015-09-19},\n\tjournal = {Information Systems Research},\n\tauthor = {Ghoshal, Abhijeet and Menon, Syam and Sarkar, Sumit},\n\tmonth = jul,\n\tyear = {2015},\n\tpages = {532--551},\n}\n\n","author_short":["Ghoshal, A.","Menon, S.","Sarkar, S."],"key":"ghoshal_recommendations_2015","id":"ghoshal_recommendations_2015","bibbaseid":"ghoshal-menon-sarkar-recommendationsusinginformationfrommultipleassociationrulesaprobabilisticapproach-2015","role":"author","urls":{"Paper":"http://pubsonline.informs.org/doi/abs/10.1287/isre.2015.0583"},"metadata":{"authorlinks":{}},"downloads":2},"bibtype":"article","biburl":"https://api.zotero.org/users/6655/collections/TJPPJ92X/items?key=VFvZhZXIoHNBbzoLZ1IM2zgf&format=bibtex&limit=100","creationDate":"2020-03-27T02:34:35.362Z","downloads":2,"keywords":[],"search_terms":["recommendations","using","information","multiple","association","rules","probabilistic","approach","ghoshal","menon","sarkar"],"title":"Recommendations Using Information from Multiple Association Rules: A Probabilistic Approach","year":2015,"dataSources":["5Dp4QphkvpvNA33zi","jfoasiDDpStqkkoZB","BiuuFc45aHCgJqDLY"]}