Cross-Language Boosting in Pattern-based Semantic Relation Extraction from Text

In this work we propose a novel technique called "Cross-Language Boosting" (C-LB), aimed at increasing the accuracy of pattern-based semantic relation extraction systems: given a pair of terms expressed in a "Target Language" (e.g. in Italian), we can translate the terms in a "Support Language" (e.g. in English) and apply the translated term pair to reliable lexico-syntactic patterns expressed in that language to increase the accuracy of the system. Experiments have been conducted by comparing the results obtained by the SemRelEx system, a hybrid unsupervised system for semantic relation extraction from texts, with and without the support of the C-LB technique, applied to a set of candidate semantically related term pairs automatically extracted from a corpus in the History of Art domain.