Research areas and people
For computers to interact with humans it is necessary to both understand how people interact with each other or with computers (pragmatics) and to model it as well as possible. We are dealing with:
- Intelligent and Virtual Agents
- Dialogue Models and Human Computer/Robot Interaction in Natural Language
- Non-Verbal Interaction: gestures, etc
- Audiovisual speech synthesis
- Dialogue Corpora
Estonian as a morhologically rich language with free word order needs an appropriate approach for machine translation. We are researching formalisms and methods for translationg to and from such a language. There are people working on both rule-based and statistical MT, neural translation and automatic post-editing of translation results.
The people working on machine translation are Mark Fišel, Heiki-Jaan Kaalep, Kairit Sirts and students.
Our goal is for the computer to be able to construct, inflect and analyze words as well as a human can.
According to the speller or Keeleveeb queries it could be said that we are already quite close to that goal.
The researchers working on morphology are Heiki-Jaan Kaalep, Jaak Pruulmann-Vengerfeldt, Heli Uibo and Tarmo Vaino.
As a part of language technology research in syntax aims to find a good formalism to represent the deep structure of sentences in Estonian, as well as effective methods for morphological disambiguation and transfer from the shallow representation used so far to a deeper tree structure. Our people are also researching adaptating existing morphological disambiguation and parsers to spoken language and automatic detection of disfluencies to exclude ungrammatical phrases from syntactic analysis.
The people in our chair researching syntax are Kadri Muischnek, Kaili Müürisep, Krista Liin and Kristiina Vaik.
Semantics deals with the meaning of texts.
In the field of computational semantics we have done most work on temporal expressions and named entity recognition. Our main researcher in this field is Siim Orasmaa
Lexical semantics looks at word level representations of meaning, but also smaller units and combining such meanings. This includes textual semantics and WordNet - a network of lexical-level meanings. The researchers working in this field are Heili Orav, Neeme Kahusk, Indrek Jentson and Kadri Vare.
Formal semantics deals with ontologies and logical models of language. Our people working on these topics are Haldur Õim and Erkki Luuk.
Eduard Barbu has been working in various fields of Natural Language Processing both in Academia and Industry settings. His main interests are related to semantics: the nature and structure of concepts, Ontology learning from text and Ontology building.
Our chair hosts several corpora and other language datasets that our research group has collected and annotated to study language and NLP methods. We are also part of the Center of Estonian Language Resources that aims to collect and make accessible in one place all available resources for Estonian language.
Our language resources are managed by Kadri Muischnek, Kadri Vider, Neeme Kahusk, Krista Liin and Kadri Vare.