The research on Human Language Processing aims to automatise speech and text-based interactions. In the MILAB, we have defined two focus areas for Human Language Processing research. We are continuing the research efforts of the last two decades, and develop AI solutions for processing the Hungarian language. Hungarian is a stereotype of morphologically rich and free word order languages, hence our research findings can also be utilised at the developments on other morphologically rich languages - and most of the languages in Europe are morphologically rich opposed to the most researched language English -.
Another objective of MILAB is to bridge the gap between research and real-world applications . We are working on AI problems which will enable the effective industrial application of language technology research solutions. Such an important research problem is cheap fine-tuning of applications. The language of every domain (like medical, financial, legal etc.) has its own characteristics and dictionary, moreover each and every application has its own requirements, own definition of task. Today, AI-based human language processing solutions work fine if a big enough labelled domain- and task-specific dataset is available. We research tools which can enable non-expert users fine-tuning their applications even in the lack of big labelled datasets.
