Cross-Language Learning for Arabic Relation Extraction

Relation Extraction from Arabic text is a difficult and challenging task. Pattern-based methods often employ precise and accurate linguistics rules; however, they need huge amount of manual works to annotate corpora with desired tags. On the other hand, supervised methods need large corpus with semantic tags; which in turn imposes extra load for preparing annotated data. In this paper, a cross-language method for relation extraction is proposed, which uses the training data of other languages and trains a model for relation extraction from Arabic text. The task is supervised learning, in which several lexical and syntactic features are considered. The proposed method mainly relies on the Universal Dependency (UD) parsing and the similarity of UD trees in different languages. Regarding UD parse trees, all the features for training classifiers are extracted and represented in a universal space. To incorporate different features in training the classifier, a combination of kernel functions is proposed. Result of experiments on ACE-2004 data set reveals 63.5% F1 for Arabic test data.