Abstract
This paper presents a statistical type inference algorithm for ontology alignment, which assigns DBpedia entities with a new type (class). To infer types for a specific entity, the algorithm first identifies types that co-occur with the type the entity already has, and subsequently prunes the set of candidates for the most confident one. The algorithm has one parameter for balancing specificity/reliability of the resulting type selection. The proposed algorithm is used to complement the types in the LHD dataset, which is RDF knowledge base populated by identifying hypernyms from the free text of Wikipedia articles. The majority of types assigned to entities in LHD 1.0 are DBpedia resources. Through the statistical type inference, the number of entities with a type from DBpedia Ontology is increased significantly: by 750 thousand entities for the English dataset, 200.000 for Dutch and 440.000 for German. The accuracy of the inferred types is at 0.65 for English (as compared to 0.86 for LHD 1.0 types). A byproduct of the mapping process is a set of 11.000 mappings from DBpedia resources to DBpedia Ontology classes with associated confidence values. The number of the resulting mappings is an order of magnitude larger than what can be achieved with standard ontology alignment algorithms (Falcon, LogMapLt and YAM++), which do not utilize the type co-occurrence information. The presented algorithm is not restricted to the LHD dataset, it can be used to address generic type inference problems in presence of class membership information for a large number of instances.- Anthology ID:
- L14-1552
- Volume:
- Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)
- Month:
- May
- Year:
- 2014
- Address:
- Reykjavik, Iceland
- Editors:
- Nicoletta Calzolari, Khalid Choukri, Thierry Declerck, Hrafn Loftsson, Bente Maegaard, Joseph Mariani, Asuncion Moreno, Jan Odijk, Stelios Piperidis
- Venue:
- LREC
- SIG:
- Publisher:
- European Language Resources Association (ELRA)
- Note:
- Pages:
- 3517–3523
- Language:
- URL:
- http://www.lrec-conf.org/proceedings/lrec2014/pdf/703_Paper.pdf
- DOI:
- Cite (ACL):
- Tomáš Kliegr and Ondřej Zamazal. 2014. Towards Linked Hypernyms Dataset 2.0: complementing DBpedia with hypernym discovery. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14), pages 3517–3523, Reykjavik, Iceland. European Language Resources Association (ELRA).
- Cite (Informal):
- Towards Linked Hypernyms Dataset 2.0: complementing DBpedia with hypernym discovery (Kliegr & Zamazal, LREC 2014)
- PDF:
- http://www.lrec-conf.org/proceedings/lrec2014/pdf/703_Paper.pdf