FRAMEWORK FOR THE SEMI-AUTOMATIC CONSTRUCTION OF A TEST COLLECTION USED IN EXTRACTING RELATIONSHIPS BETWEEN TECHNICAL TERMS

摘要

The present invention relates to a framework for the semi-automatic construction of test collections used in extracting relationships between technical terms, and provides a framework for the semi-automatic construction of test collections used in extracting relationships between technical terms, wherein a practical test collection is constructed by systematically processing technical terms present in texts and associative relationships between the same on the basis of a language resource, and by going beyond the existing approach for constructing small-scale test collections based on limited texts and objects and instead making effective use of a large-scale academic database, and of specialist dictionaries in various fields and up-to-date machine-learning algorithms. The present invention provides the advantages that efforts on the part of constructors can be minimised and variations in results which occur due to the dispositions of constructors can be reduced by automating tasks which are standardised and time-consuming.