29 September 2026 to 3 October 2026
OeAW Main Seat
Europe/Vienna timezone

From Text to Terms: TermScout, an AI Module for Early Terminology Processing

30 Sept 2026, 14:00
1h 30m
01 | Sitzungssaal (01 | Sitzungssaal, OeAW Main Seat, 1st floor)

01 | Sitzungssaal

01 | Sitzungssaal, OeAW Main Seat, 1st floor

Austrian Academy of Sciences Dr. Ignaz Seipel-Platz 2 1010 Vienna

Speakers

Bruno Nahod (Institute for the Croatian Language) Marijana Tuta Dujmović (Polyclinic for Hearing and Speech Rehabilitation SUVAG)

Description

This paper presents TermScout, a GPT-based tool developed to support the initial stage of terminology management by identifying candidate terms in specialised texts. The study evaluates TermScout using a corpus focused on cochlear implantation, assessing both English term extraction and the generation of Croatian equivalents. TermScout identified 100 English candidate terms from a specialised textbook, while Croatian equivalents were generated either directly by TermScout or through a subsequent process utilising TermAI. An independent term validation tool was used to assess all results. The tool demonstrated strong performance in English term extraction, achieving a validity rate of 92%. In contrast, generating Croatian equivalents proved more challenging, with validity rates of 22% for TermScout and 29% for TermAI. These findings indicate that concept identification and bilingual term generation are distinct tasks that require distinct linguistic and terminological expertise. The study underscores the potential of AI-assisted workflows for terminology management, while emphasising the continued need for expert validation.

Presentation materials

There are no materials yet.