BlueFinder: Recommending Wikipedia Links Using DBpedia Properties - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2013

BlueFinder: Recommending Wikipedia Links Using DBpedia Properties

Diego Torres
  • Function : Author
  • PersonId : 856793
Pascal Molli
Alicia Diaz
  • Function : Author
  • PersonId : 856794

Abstract

DBpedia knowledge base has been built from data extracted from Wikipedia. However, many existing relations among resources in DBpedia are missing links among articles from Wikipedia. In some cases, adding these links into Wikipedia will enrich Wikipedia content and therefore will enable bet- ter navigation. In previous work, we proposed PIA algorithm that predicts the best link to connect two articles in Wikipedia corresponding to those related by a semantic property in DB- pedia and respecting the Wikipedia convention. PIA calcu- lates this link as a path query. After introducing PIA results in Wikipedia, most of them were accepted by the Wikipedia community. However, some were rejected because PIA pre- dicts path queries that are too general. In this paper, we report the BlueFinder collaborative filtering algorithm that fixes PIA miscalculation. It is sensible to the specificity of the resource types. According to the conducted experimentation we found out that BlueFinder is a better solution than PIA because it solves more cases with a better recall.
Fichier principal
Vignette du fichier
authorVersionmain.pdf (586.4 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-00822696 , version 1 (15-05-2013)

Identifiers

  • HAL Id : hal-00822696 , version 1

Cite

Diego Torres, Hala Skaf-Molli, Pascal Molli, Alicia Diaz. BlueFinder: Recommending Wikipedia Links Using DBpedia Properties. Web Science, May 2013, Paris, France. ⟨hal-00822696⟩
202 View
332 Download

Share

Gmail Facebook X LinkedIn More