Coming from the Laboratory of Applied and Computational Linguistics Kyiv National Linguistic University, 5 years immersed in R&D in the field of AI and NLP. Today he represents department of Data Science at YouScan — systems of social media monitoring and research.
Spheres of interests are automation of linguistic analysis, text summary, classification and clustering of short texts, and also methods of vector representation of words and meanings, ML methods in NLP.
Topic: Embracing multiple senses in word space models
Short Description: Problems of polysemy and homonymy are rarely considered in the context of words vector representations constructing.
Most language models do not take into account the possibility of coincidence in the writing of words with different meanings. Researches demonstrate that separation by parts of speech gives an increase in the basic quality of NLP systems. We’ll try to discuss other ways to solve this problem, using live data from social networking services.