Mining Relation Extraction Based on Pattern Learning Approach

Mujiono Sadikin


Semantically, objects in unstructured document are related each other to perform a certain entity relation. This certain entity relation such: drug-drug interaction through their compounds, buyer-seller relationship through the goods or services, etc. Motivated by those kind of interaction, this study proposes a method to extract those objects and their interactions. It is presented a general framework of object-interaction mining of large corpora. The framework is started with the initial step in extracting a single object in the unstructured document. In this study, the initial step is a pattern learning method that is applied to drug-label documents to extract drug-names. We utilize an existing external knowledge to identify a certain regular expressions surrounding the targeted object and the probabilities of those regular expression, to perform the pattern learning process. The performance of this pattern learning approach is promising to apply in this relation extraction area. As presented in the results of this study, the best f-score performance of this method is 0.78 f-score. With adjusting of some parameters and or improving the method, the performance can be potentially improved.


Unstructured document object; Interaction object; Relation information extraction; Pattern learning; Drug-name;

Full Text:




  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

shopify stats IJEECS visitor statistics