Hierarchical document categorization using associative networks

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    1 Citation (Scopus)
    8 Downloads (Pure)

    Abstract

    Associative networks are a connectionist language model with the ability to handle dynamic data. We used two associative networks to categorize random sets of related Wikipedia articles with only their raw text as input. We then compared the resulting categorization to a gold standard: the manual categorization by Wikipedia authors and used a neural network as a baseline. We also determined a human rating by having a group of judges rank the four categorization methods by correctness and by usefulness with regards to finding information. Based on these tests, we determined that associative networks produce results that are clearly better than the neural network baseline, coming close to the gold standard in terms of usefulness and correctness. Furthermore, automated testing suggests these results continue to hold for large datasets.
    Original languageEnglish
    Title of host publicationProceedings of the IASTED Multiconferences
    Subtitle of host publicationArtificial Intelligence and Applications (AIA 2013)
    EditorsE.P. Klement, W. Borutzky, T. Fahringer, M.H. Hamza, V. Uskov
    Place of PublicationAnaheim
    PublisherACTA Press
    Pages028
    Number of pages8
    ISBN (Print)978-0-88986-955-4
    DOIs
    Publication statusPublished - Feb 2013
    Event12th IASTED International Conference on Artificial Intelligence and Applications, AIA 2013 - Innsbruck, Austria
    Duration: 11 Feb 201313 Feb 2013
    Conference number: 12

    Publication series

    Name
    PublisherACTA Press

    Conference

    Conference12th IASTED International Conference on Artificial Intelligence and Applications, AIA 2013
    Abbreviated titleAIA
    CountryAustria
    CityInnsbruck
    Period11/02/1313/02/13

    Fingerprint

    Neural networks
    Testing

    Keywords

    • EWI-23553
    • Document Clustering
    • Hierarchical Categorization
    • METIS-297757
    • Connectionist Language Model
    • Associative Networks
    • Automatic Categorization
    • IR-87091

    Cite this

    Bloom, N., Theune, M., & de Jong, F. M. G. (2013). Hierarchical document categorization using associative networks. In E. P. Klement, W. Borutzky, T. Fahringer, M. H. Hamza, & V. Uskov (Eds.), Proceedings of the IASTED Multiconferences: Artificial Intelligence and Applications (AIA 2013) (pp. 028). Anaheim: ACTA Press. https://doi.org/10.2316/P.2013.793-028
    Bloom, Niels ; Theune, Mariet ; de Jong, Franciska M.G. / Hierarchical document categorization using associative networks. Proceedings of the IASTED Multiconferences: Artificial Intelligence and Applications (AIA 2013). editor / E.P. Klement ; W. Borutzky ; T. Fahringer ; M.H. Hamza ; V. Uskov. Anaheim : ACTA Press, 2013. pp. 028
    @inproceedings{5658a9fcd62a4092ae54b9475f5caf4c,
    title = "Hierarchical document categorization using associative networks",
    abstract = "Associative networks are a connectionist language model with the ability to handle dynamic data. We used two associative networks to categorize random sets of related Wikipedia articles with only their raw text as input. We then compared the resulting categorization to a gold standard: the manual categorization by Wikipedia authors and used a neural network as a baseline. We also determined a human rating by having a group of judges rank the four categorization methods by correctness and by usefulness with regards to finding information. Based on these tests, we determined that associative networks produce results that are clearly better than the neural network baseline, coming close to the gold standard in terms of usefulness and correctness. Furthermore, automated testing suggests these results continue to hold for large datasets.",
    keywords = "EWI-23553, Document Clustering, Hierarchical Categorization, METIS-297757, Connectionist Language Model, Associative Networks, Automatic Categorization, IR-87091",
    author = "Niels Bloom and Mariet Theune and {de Jong}, {Franciska M.G.}",
    year = "2013",
    month = "2",
    doi = "10.2316/P.2013.793-028",
    language = "English",
    isbn = "978-0-88986-955-4",
    publisher = "ACTA Press",
    pages = "028",
    editor = "E.P. Klement and W. Borutzky and T. Fahringer and M.H. Hamza and V. Uskov",
    booktitle = "Proceedings of the IASTED Multiconferences",
    address = "Canada",

    }

    Bloom, N, Theune, M & de Jong, FMG 2013, Hierarchical document categorization using associative networks. in EP Klement, W Borutzky, T Fahringer, MH Hamza & V Uskov (eds), Proceedings of the IASTED Multiconferences: Artificial Intelligence and Applications (AIA 2013). ACTA Press, Anaheim, pp. 028, 12th IASTED International Conference on Artificial Intelligence and Applications, AIA 2013, Innsbruck, Austria, 11/02/13. https://doi.org/10.2316/P.2013.793-028

    Hierarchical document categorization using associative networks. / Bloom, Niels; Theune, Mariet; de Jong, Franciska M.G.

    Proceedings of the IASTED Multiconferences: Artificial Intelligence and Applications (AIA 2013). ed. / E.P. Klement; W. Borutzky; T. Fahringer; M.H. Hamza; V. Uskov. Anaheim : ACTA Press, 2013. p. 028.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    TY - GEN

    T1 - Hierarchical document categorization using associative networks

    AU - Bloom, Niels

    AU - Theune, Mariet

    AU - de Jong, Franciska M.G.

    PY - 2013/2

    Y1 - 2013/2

    N2 - Associative networks are a connectionist language model with the ability to handle dynamic data. We used two associative networks to categorize random sets of related Wikipedia articles with only their raw text as input. We then compared the resulting categorization to a gold standard: the manual categorization by Wikipedia authors and used a neural network as a baseline. We also determined a human rating by having a group of judges rank the four categorization methods by correctness and by usefulness with regards to finding information. Based on these tests, we determined that associative networks produce results that are clearly better than the neural network baseline, coming close to the gold standard in terms of usefulness and correctness. Furthermore, automated testing suggests these results continue to hold for large datasets.

    AB - Associative networks are a connectionist language model with the ability to handle dynamic data. We used two associative networks to categorize random sets of related Wikipedia articles with only their raw text as input. We then compared the resulting categorization to a gold standard: the manual categorization by Wikipedia authors and used a neural network as a baseline. We also determined a human rating by having a group of judges rank the four categorization methods by correctness and by usefulness with regards to finding information. Based on these tests, we determined that associative networks produce results that are clearly better than the neural network baseline, coming close to the gold standard in terms of usefulness and correctness. Furthermore, automated testing suggests these results continue to hold for large datasets.

    KW - EWI-23553

    KW - Document Clustering

    KW - Hierarchical Categorization

    KW - METIS-297757

    KW - Connectionist Language Model

    KW - Associative Networks

    KW - Automatic Categorization

    KW - IR-87091

    U2 - 10.2316/P.2013.793-028

    DO - 10.2316/P.2013.793-028

    M3 - Conference contribution

    SN - 978-0-88986-955-4

    SP - 028

    BT - Proceedings of the IASTED Multiconferences

    A2 - Klement, E.P.

    A2 - Borutzky, W.

    A2 - Fahringer, T.

    A2 - Hamza, M.H.

    A2 - Uskov, V.

    PB - ACTA Press

    CY - Anaheim

    ER -

    Bloom N, Theune M, de Jong FMG. Hierarchical document categorization using associative networks. In Klement EP, Borutzky W, Fahringer T, Hamza MH, Uskov V, editors, Proceedings of the IASTED Multiconferences: Artificial Intelligence and Applications (AIA 2013). Anaheim: ACTA Press. 2013. p. 028 https://doi.org/10.2316/P.2013.793-028