There are always new challenges in the extraction of object (entity) relations contained in unstructured or semi structured text documents found in the Internet, due to the volume of the documents, the evolution of the text language, and the fast Internet growth. In this paper, the authors present the description and the experimental results of a novel role based approach in mining the entities and its relations. The proposed method defines a new concept of entity relationship which treat entities relation as the relation of the main object and its supporting object. The relation between these objects are extracted through pattern learning process that utilize the Indonesia WordNet as an external knowledge. Based on the performance evaluation of the proposed method, it can be confirmed that it is feasible to apply the method in the area. The feasibility of the method is measured by the accuracy of the extraction process in 10 experiments. The average F-score values for the experiments are 0.895 and 0.795 in main object extraction and supporting object extraction respectively.
|Number of pages||13|
|Journal||Journal of Theoretical and Applied Information Technology|
|Publication status||Published - 2015|
- Object extraction relation
- Object interaction
- Pattern learning
- Tuple scoring