Mohamed Amir Yosef

AIDA-Social: Entity Linking on the Social Stream

Named Entity Linking (NEL) in microblogs is a challenging task due to the use of cryptic abbreviations, insufficient contextual information, and the time-varying importance of entities. We propose three techniques to target these challenges: Mention Normalization, Contextual Enrichment, and Temporal Entity Importance. By combining these novel techniques, we achieve 13% improvement in precision over a state-of-the-art NEL tool.

Named Entity Disambiguation for Resource-Poor Languages

Named entity disambiguation (NED) is the task of linking ambiguous names in natural language text to canonical entities like people, organizations or places, registered in a knowledge base. The problem is well-studied for English text, but few systems have considered resource-poor languages that lack comprehensive name-entity dictionaries, entity descriptions, and large annotated training corpora. In this paper we address the NED problem for languages with limited amount of annotated corpora as well as structured resource such as Arabic. We present a method that leverages structured English resources to enrich the components of a language-agnostic NED system and enable effective NED for other languages. We achieve this by fusing data from several multilingual resources and the output of automatic translation/transliteration systems. We show the viability and quality of our approach by synthesizing NED systems for Arabic, Spanish and Italian.

AIDArabic A Named-Entity Disambiguation Framework for Arabic Text

There has been recently a great progress in the field of automatically generated knowledge bases and corresponding disambiguation systems that are capable of mapping text mentions onto canonical entities. Efforts like the before mentioned have enabled researchers and analysts from various disciplines to semantically “understand” contents. However, most of the approaches have been specifically designed for the English language and - in particular - support for Arabic is still in its infancy. Since the amount of Arabic Web contents (e.g. in social media) has been increasing dramatically over the last years, we see a great potential for endeavors that support an entity-level analytics of these data. To this end, we have developed a framework called AIDArabic that extends the existing AIDA system by additional components that allow the disambiguation of Arabic texts based on an automatically generated knowledge base distilled from Wikipedia. Even further, we overcome the still existing sparsity of the Arabic Wikipedia by exploiting the interwiki links between Arabic and English contents in Wikipedia, thus, enriching the entity catalog as well as disambiguation context.

EDRAK: Entity-Centric Data Resource for Arabic Knowledge

Online Arabic content is growing very rapidly, with unmatched growth in Arabic structured resources. Systems that perform standard Natural Language Processing (NLP) tasks such as Named Entity Disambiguation (NED) struggle to deliver decent quality due to the lack of rich Arabic entity repositories. In this paper, we introduce EDRAK, an automatically generated comprehensive Arabic entity-centric resource. EDRAK contains more than two million entities together with their Arabic names and contextual keyphrases. Manual evaluation confirmed the quality of the generated data. We are making EDRAK publicly available as a valuable resource to help advance research in Arabic NLP and IR tasks such as dictionary-based NamedEntity Recognition, entity classification, and entity summarization.