On the 24th of November MindCraft.ai took part in the Data Science Ukraine Conference with a topic “NER or Recognize and Classify Documents and Named Entities are to Use Word-Based Models with Embeddings” Andy Bosyi, CEO at MindCraft, showed the audience how to use character-based and other models and stack them in order to achieve better results in the no-data area.
Also, he explained in detail and on real examples how a character-based model can over-fit classical NLP libraries like spaCy.
The topic generated many questions from the audience. In general, the conference was very interesting, offering practical cases and experience. At MindCraft we are happy to have a chance to contribute to the development of the Data Science industry in Ukraine.