Terms are something more than ordinary words. They are the linguistic representations of specific standard concepts in particular subject fields. Words form the vocabulary of a text; terms form its terminology.

Terms are always related to a special language, a "jargon" for a specific subject area - for example, medicine, law, electronics engineering, etc. The difficulties encountered during a term recognition task are due to the fact that distinguishing terms from single words is not an easy matter. Moreover, terms can be single words or multi-word constructs; again, though, the identification of the boundaries of a term is not a simple task.

Term-o-nizer is an automatic term recognition tool that can locate and extract existing terminology in a text collection. The terms that Term-o-nizer can retrieve may comprise single or multiple words.

Besides automatic term recognition, obviously very helpful for understanding what a document is about especially during a translation session, further possible applications of Term-o-nizer include:

• specialised dictionary construction and maintenance
• human and machine translation applications
• hypertext linking
• text categorisation
• information retrieval and extraction
• text abstracting and summarization

Term recognition gives the potential to work with large amounts of real data that would be very difficult to handle manually. This way, one can be consistent and accurate on the terminology used and can therefore have better access to the information that lies in multilingual documents. And last, automatic term recognition is faster and a lot cheaper than manual term recognition.

Contact person: Stelios Piperidis