C4.5: programs for machine learning
C4.5: programs for machine learning
| Hi-index | 0.00 |
Terminology Acquisition (TA) methods are viable solutions for the knowledge bottleneck problem that confines knowledge-intensive information access systems (such as Information Extraction systems) to restricted application scenarios. TA can be seen as a way to inspect large text collections for extracting concise domain knowledge. In this paper we argue that major insights over the notion of term can be obtained by investigating a more domain-based term definition. We propose a decision tree learning approach as an interesting model of the human TA activity. An incremental model is proposed to study the evolution of the term definition during the TA process over a particular implicit domain model. The experimental apparatus is based on robust text processing tools that support a large scale investigation. The good results suggest that the proposed automatic TA model can support the development of conceptual domain dictionaries as required by knowledge-based information systems.