Learning of Chained Rules for Construction of Conceptual Graphs

Authors: Sonia Ordoņez Salinas

 Polibits, 41, pp. 39-45, 2010.

Abstract:  The paper presents a method for learning of the chained rules for generation of the new rules. Application of these rules allows for construction of conceptual graphs. The proposal is based on the inclusion of the chained rules and on the supervised method.  The rules are defined on the basis of the three elements: a. syntactic role of a word in the phrase b. Conceptual graphs standards, c. definition of the object that works as a black box for conceptual graphs. The experiments were conducted for titles and comments that form part of the collection of medical images ImageClefmed 2008. For marking the relations we used the metathesaurus UMLS and the tool MMTx, while for classification we used Weka. We evaluate the obtained rules.

Keywords:  Learning of chained rules; UMLS; conceptual graphs; medical image annotation

PDF:  Learning of Chained Rules for Construction of Conceptual Graphs, Alternative link