Human Language Technology Days was an international conference held on September 27-28 in Warsaw. It was organized by the Institute of Computer Science of the Polish Academy of Sciences and the University of Łódź as a part of CESAR project (Central and South-East European Resources) – one of several EU financed projects that aim at developing common platform of open tools and resources for European languages.
Presentations on the first day of the conference proved that although there are still some domains of language technology that need extensive research, there are some areas that are already quite mature and have been successfully introduced in well recognized commercial products such as iPhone Siri, Google Translate or IBM Watson. The last one is the supercomputer which won Jeopardy! quiz in 2011, overcoming two human masters of this game by giving accurate answers to the most questions. The advanced language processing and knowledge representation in this system was explained by Włodek Zadrożny – researcher in DeepQA project under which the Watson was developed. Now the system is being adjusted to be used as diagnosis assistance in medicine. Another interesting talk was given by Enrique Alfonseca, a researcher in Google. He presented a recently started service Knowledge Graph (which based on semantic database Freebase) and some new ideas that are currently under developement, such as automatic text summarization, which is useful in Google News to make a short description for many similar documents on the same topic.
The second day of the conference focused on the current language technology research in Poland. Wide range of natural language processing areas was introduced in 13 presentations, such as finite state automata construction and representation (used e.g. for dictionary compression), various services for linguisticians or speech recognition and question answering systems.
All presentations can be watched at http://www.hltdays.pl/video.