A maximum entropy approach for collaborative filtering

John Browning, David Jonathan Miller

    Research output: Contribution to journalArticle

    8 Citations (Scopus)

    Abstract

    Collaborative filtering (CF) involves predicting the preferences of a user for a set of items given partial knowledge of the user's preferences for other items, while leveraging a database of profiles for other users. CF has applications e.g. in predicting Web sites a person will visit and in recommending products. Fundamentally, CF is a pattern recognition task, but a formidable one, often involving a huge feature space, a large data set, and many missing features. Even more daunting is the fact that a CF inference engine must be capable of predicting any (user-selected) items, given any available set of partial knowledge on the user's other preferences. In other words, the model must be designed to solve any of a huge (combinatoric) set of possible inference tasks. CF techniques include memory-based, classification-based, and statistical modelling approaches. Among these, modelling approaches scale best with large data sets and are the most adept at handling missing features. The disadvantage of these methods lies in the statistical assumptions (e.g. feature independence), which may be unjustified. To address this shortcoming we propose a new model-based CF method, based on the maximum entropy principle. For the MS Web application, the new method is demonstrated to outperform a number of CF approaches, including naive Bayes and latent variable (cluster) models, support vector machines (SVMs), and the (Pearson) correlation method.

    Original languageEnglish (US)
    Pages (from-to)199-209
    Number of pages11
    JournalJournal of VLSI Signal Processing Systems for Signal, Image, and Video Technology
    Volume37
    Issue number2-3
    DOIs
    StatePublished - Jun 1 2004

    Fingerprint

    Collaborative filtering
    Collaborative Filtering
    Maximum Entropy
    Entropy
    Large Data Sets
    Partial
    Maximum Entropy Principle
    Inference engines
    Pearson Correlation
    Inference Engine
    Naive Bayes
    Correlation methods
    Statistical Modeling
    User Preferences
    Latent Variables
    Web Application
    Feature Space
    Combinatorics
    Pattern Recognition
    Pattern recognition

    All Science Journal Classification (ASJC) codes

    • Signal Processing
    • Information Systems
    • Electrical and Electronic Engineering

    Cite this

    @article{2199a81a7d15474c9b8f149b0d000f82,
    title = "A maximum entropy approach for collaborative filtering",
    abstract = "Collaborative filtering (CF) involves predicting the preferences of a user for a set of items given partial knowledge of the user's preferences for other items, while leveraging a database of profiles for other users. CF has applications e.g. in predicting Web sites a person will visit and in recommending products. Fundamentally, CF is a pattern recognition task, but a formidable one, often involving a huge feature space, a large data set, and many missing features. Even more daunting is the fact that a CF inference engine must be capable of predicting any (user-selected) items, given any available set of partial knowledge on the user's other preferences. In other words, the model must be designed to solve any of a huge (combinatoric) set of possible inference tasks. CF techniques include memory-based, classification-based, and statistical modelling approaches. Among these, modelling approaches scale best with large data sets and are the most adept at handling missing features. The disadvantage of these methods lies in the statistical assumptions (e.g. feature independence), which may be unjustified. To address this shortcoming we propose a new model-based CF method, based on the maximum entropy principle. For the MS Web application, the new method is demonstrated to outperform a number of CF approaches, including naive Bayes and latent variable (cluster) models, support vector machines (SVMs), and the (Pearson) correlation method.",
    author = "John Browning and Miller, {David Jonathan}",
    year = "2004",
    month = "6",
    day = "1",
    doi = "10.1023/B:VLSI.0000027485.11890.15",
    language = "English (US)",
    volume = "37",
    pages = "199--209",
    journal = "Journal of Signal Processing Systems",
    issn = "1939-8018",
    publisher = "Springer New York",
    number = "2-3",

    }

    A maximum entropy approach for collaborative filtering. / Browning, John; Miller, David Jonathan.

    In: Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology, Vol. 37, No. 2-3, 01.06.2004, p. 199-209.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - A maximum entropy approach for collaborative filtering

    AU - Browning, John

    AU - Miller, David Jonathan

    PY - 2004/6/1

    Y1 - 2004/6/1

    N2 - Collaborative filtering (CF) involves predicting the preferences of a user for a set of items given partial knowledge of the user's preferences for other items, while leveraging a database of profiles for other users. CF has applications e.g. in predicting Web sites a person will visit and in recommending products. Fundamentally, CF is a pattern recognition task, but a formidable one, often involving a huge feature space, a large data set, and many missing features. Even more daunting is the fact that a CF inference engine must be capable of predicting any (user-selected) items, given any available set of partial knowledge on the user's other preferences. In other words, the model must be designed to solve any of a huge (combinatoric) set of possible inference tasks. CF techniques include memory-based, classification-based, and statistical modelling approaches. Among these, modelling approaches scale best with large data sets and are the most adept at handling missing features. The disadvantage of these methods lies in the statistical assumptions (e.g. feature independence), which may be unjustified. To address this shortcoming we propose a new model-based CF method, based on the maximum entropy principle. For the MS Web application, the new method is demonstrated to outperform a number of CF approaches, including naive Bayes and latent variable (cluster) models, support vector machines (SVMs), and the (Pearson) correlation method.

    AB - Collaborative filtering (CF) involves predicting the preferences of a user for a set of items given partial knowledge of the user's preferences for other items, while leveraging a database of profiles for other users. CF has applications e.g. in predicting Web sites a person will visit and in recommending products. Fundamentally, CF is a pattern recognition task, but a formidable one, often involving a huge feature space, a large data set, and many missing features. Even more daunting is the fact that a CF inference engine must be capable of predicting any (user-selected) items, given any available set of partial knowledge on the user's other preferences. In other words, the model must be designed to solve any of a huge (combinatoric) set of possible inference tasks. CF techniques include memory-based, classification-based, and statistical modelling approaches. Among these, modelling approaches scale best with large data sets and are the most adept at handling missing features. The disadvantage of these methods lies in the statistical assumptions (e.g. feature independence), which may be unjustified. To address this shortcoming we propose a new model-based CF method, based on the maximum entropy principle. For the MS Web application, the new method is demonstrated to outperform a number of CF approaches, including naive Bayes and latent variable (cluster) models, support vector machines (SVMs), and the (Pearson) correlation method.

    UR - http://www.scopus.com/inward/record.url?scp=3543101946&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=3543101946&partnerID=8YFLogxK

    U2 - 10.1023/B:VLSI.0000027485.11890.15

    DO - 10.1023/B:VLSI.0000027485.11890.15

    M3 - Article

    AN - SCOPUS:3543101946

    VL - 37

    SP - 199

    EP - 209

    JO - Journal of Signal Processing Systems

    JF - Journal of Signal Processing Systems

    SN - 1939-8018

    IS - 2-3

    ER -