Skip to main content Skip to main navigation


Proceedings of the 5th Workshop on Semantic Deep Learning

Dagmar Gromann; Luis Espinosa-Anke; Thierry Declerck; Jose Camacho-Collados; Mohammad Taher Pilehvar (Hrsg.)
Workshop on Semantic Deep Learning (SemDeep-2019), located at 28th International Joint Conference on Artificial Intelligence, August 10-16, Macao, China, ISBN 978-1-950737-19-2, Association for Computational Linguistics (ACL), Stroudsburg, PA 18360, USA, 8/2019.


Welcome to the 5th Workshop on Semantic Deep Learning (SemDeep-5), held in conjunction with IJ-CAI 2019 (Macau, China). As a series of workshops and a special issue, SemDeep has been aiming tobring together Semantic Web and Deep Learning research as well as industrial communities. It seeks tooffer a platform for joining computational linguistics and formal approaches to represent information andknowledge, and thereby opens the discussion for new and innovative application scenarios for neural andsymbolic approaches to NLP, such as neural-symbolic reasoning.SemDeep-5 features a shared task on evaluating meaning representations, the Word in Context (WiC)challenge. It represents a joint task of semantic structure in the organization of senses and their represen-tation. In addition to providing a reliable benchmark for studying an important linguistic phenomenon,WiC is directly related to applications such as word sense disambiguation, entity linking, and semanticsearch. In brief, the task consists in determining whether a given word is used in the same or differentsenses given two different contexts. For the WiC challenge there were seven participant systems and threepapers could be accepted. Ansell et al. present an ELMo-inspired approach to tackle this challenge thatintroduces a new similarity measure for an adapted version of contextualized representations. Loureiroand Jorge combine word sense disambiguation with contextual embeddings and sense embeddings. Finally, Soler et al. utilize word and sentence embeddings paired with in-context substitute annotations.


Weitere Links