Johannes Matiasek

Office:

Freyung 6/3/1a
A-1010 Wien, Austria

Tel:

(+43-1)532 46 21-8

Fax:

(+43-1)532 46 21-9

Email:

	@
johannes.matiasek		ofai.at

Research Interests

I'm interested in almost all aspects of processing natural language. My work in natural language processing started with programming a natural language frontend to a relational database back in 1985. We developed a case-frame based grammar already using a bit of unification and employed generalized quantifiers in the logical form.

Later I got more more involved into unification-based formalisms, especially HPSG, and for my thesis I employed constraint logic programming for processing unification-based grammars. I also developed a tactical generator for German, that used a grammar integrating the syntactic and morphological level.

Recently, I used statistical models for text prediction and machine learning methods for text categorization and filtering. I'm now involved in shallow semantic processing of large parallel corpora of raw speech transcripts and formatted final reports for the purpose of reconstructing the originally spoken text.

Publications

• Up-to-date list of my publications. Many of them are available online, links are provided with the list.

Projects

Current and recent projects I am/was working on include:

• ExpertSeek, a project in which we develop core technology for a new kind of semantic search system that is able to automatically identify experts for clearly defined project tasks making use of textual similarities in expert and task profiles, and of models of explicit expert knowledge.
• Inspiration, a project in the framework of the Knet COAST, aiming at modelling and supporting document creation processes aided by speech recognition. The first application area is the production of medical reports.
• SPARC (Semantic Phonetic Automatic Reconstruction of Dictations). This project aims at laying the foundation for a radical improvement in the quality of automated dictation by making explicit use of semantic knowledge.
• An Automaton for the Moderation of Internet-based Discussion Fora is concerned with the detection of inappropriate postings in online fora of a leading Austrian online newspaper (derStandard.at).
• FASTY (Faster Typing for Disabled Persons) deals with the creation of a system for increasing the text generation rate of disabled persons by Predictive Typing and dedicated advanced input devices.

Less recent projects I was involved in were:

• InQuaC (Innovative Quality Control Methods for Rotating Machines Using Artificial Intelligence Methods) deals with the development of a trainable, artificial intelligence based inspection system to be introduced into the assembly line for the purpose of realtime in-line quality control.
• TAMIC - P (Transparent Access to Multiple Information for the Citizen - Pensions) is developing an access system to very large multiple databases in the pension and social insurance field by combining natural language technology with menus and graphics.
• FACILE (Fast and Accurate Categorisation of Information by Language Engineering) aims at the development of a system able to categorise texts from the area of finance and business news in an exact and specific way.
• VieCtoS is a Concept-to-Speech system focussing on how to produce correct prosody and intonation.
• LS-GRAM: I (together with Harald Trost) made minor contributions wrt. the derivational morphology of German.
• GIST (Generating InStructional Text) is an EC funded project dealing with the multilingual generation of administrative texts. The part I have been responsible for was the tactical generater for German.
• Implementing principle based unification grammar formalisms within the CLP paradigm, this work has resulted in my PhD thesis.
• VIE-DU, a unification-based dialog system.
• Datenbank-Dialog, a natural language frontend to relational databases.