Identification of category associations using a multilabel classifier - Publikacja - MOST Wiedzy

Wyszukiwarka

Identification of category associations using a multilabel classifier

Abstrakt

Description of the data using categories allows one to describe it on a higher abstraction level. In this way, we can operate on aggregated groups of the information, allowing one to see relationships that do not appear explicit when we analyze the individual objects separately. In this paper we present automatic identification of the associations between categories used for organization of the textual data. As experimental data we used a network of English Wikipedia articles and their associated categories, that have been preprocessed by a dedicated filtering method for noise reduction. The main contribution of the paper is the introduction of the method based on supervised machine learning for mining relations between these categories. We describe existing in the literature category proximity metrics as well as introduce three new ones, based on observing the properties of a multilabel Support Vector Machine classifier. The first metric uses classifier predictions, the second uses its errors, and the third is based on its model. Comparison to the existing state-of-the-art methods, and to manual assessments, confirm that the proposed methods are useful and are more flexible than typical approaches. We show how different metrics allow us to introduce new significant relations between categories. Aggregated results of mining categories’ associations have been used to build a semantic network that shows a practical application of the research. The proposed method for finding associations can be extended with using other approaches than SVM classification, and can find (other than presented in the paper) applications for mining categories in text repositories. Eg.: it can be used for extending the prediction of the rating in recommender systems or as a method of missing data imputation.

Cytowania

  • 4

    CrossRef

  • 5

    Web of Science

  • 4

    Scopus

Pełna treść

pełna treść publikacji nie jest dostępna w portalu

Informacje szczegółowe

Kategoria:
Publikacja w czasopiśmie
Typ:
artykuł w czasopiśmie wyróżnionym w JCR
Opublikowano w:
EXPERT SYSTEMS WITH APPLICATIONS nr 61, strony 327 - 342,
ISSN: 0957-4174
Język:
angielski
Rok wydania:
2016
Opis bibliograficzny:
Szymański J., Rzeniewicz J.: Identification of category associations using a multilabel classifier// EXPERT SYSTEMS WITH APPLICATIONS. -Vol. 61, (2016), s.327-342
DOI:
Cyfrowy identyfikator dokumentu elektronicznego (otwiera się w nowej karcie) 10.1016/j.eswa.2016.05.039
Weryfikacja:
Politechnika Gdańska

wyświetlono 33 razy

Publikacje, które mogą cię zainteresować

Meta Tagi