The paper “BPEmb: Tokenization-free Pre-trained Subword Embeddings in 275 Languages” by Benjamin Heinzerling and Michael Strube has been accepted for LREC 2018, which will take place in Miyazaki, Japan, from May 7th to 12th, 2018. The paper comes with an innovative natural language resource, which can be downloaded from this GitHub page. BTW: The European…Details
Natural Language Processing (NLP)
The Natural Language Processing (NLP) group develops methods, algorithms, and tools for the automatic analysis of natural language. The group focuses on questions related to processing, understanding and generating discourse. It works on coreference resolution with a particular focus on appropriate representations for the task. It develops models of local coherence, most recently an unsupervised, graph-based version of the entity grid with applications in text generation, assessing readability and automatic summarization. The group is also interested in word sense disambiguation and its relations to discourse processing.