Predicting Degrees of Technicality in Automatic Terminology Extraction

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2883–2889 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 2883 Predicting Degrees of Technicality in Automatic Terminology Extraction Anna H¨atty1,2, Dominik Schlechtweg2, Michael Dorna1, Sabine Schulte im Walde2 1Robert Bosch GmbH, Corporate Research, Renningen, Germany 2Institute for Natural Language Processing, University of Stuttgart, Germany {anna.haetty, michael.dorna}@de.bosch.com, {schlecdk,schulte}@ims.uni-stuttgart.de Abstract While automatic term extraction is a well- researched area, computational approaches to distinguish between degrees of technicality are still understudied. We semi-automatically cre- ate a German gold standard of technicality across four domains, and illustrate the im- pact of a web-crawled general-language cor- pus on predicting technicality. When defin- ing a classification approach that combines general-language and domain-specific word embeddings, we go beyond previous work and align vector spaces to gain comparative em- beddings. We suggest two novel models to exploit general- vs. domain-specific compar- isons: a simple neural network model with pre-computed comparative-embedding infor- mation as input, and a multi-channel model computing the comparison internally. Both models outperform previous approaches, with the multi-channel model performing best. 1 Introduction Automatic term extraction, i.e. the task of extract- ing linguistic expressions characteristic to a spe- cialized domain, is a long-researched field within natural language processing. Assessing the tech- nicality of the extracted terms, however, is still a niche within this area: technicality refers to the de- gree to which a term is specialized and exclusively used by experts in a domain. Up to date, studies on term technicality are mostly restricted to medical terminology and relate to the communication be- tween doctors and patients. Especially in times of growing amounts of domain-specific websites with both lay and expert users (e.g. DIY ‘do-it-yourself’ communities, such as 1-2-do.com), the communi- cation between experts and lays becomes increas- ingly important across all specialized domains. Fur- thermore, term technicality prediction is important for a range of tasks such as automatic thesaurus creation, assessing text specialization, and domain knowledge acquisition. Above all, predicting tech- nicality can be considered a more fine-grained and expressive form of terminology extraction. In this work, we first semi-automatically col- lect German specialized domain corpora to cre- ate a gold standard of term technicality across four domains: automotive, cooking, hunting and DIY. Based on a qualitative analysis of terminolog- ical phenomena and variants of ambiguity across domain-specific and general-language corpora, we then suggest two methods to explicitly integrate not only vector space model representations derived from the corpora, but also comparisons across the vector spaces. In a first approach, we enrich the combined general-language and domain-specific word embeddings with a difference vector as input for a classification system. In a second approach we design a multi-channel feed-forward neural net- work with a Siamese network component to repre- sent the vector comparison internally. 2 Related Work Existing studies on technicality predominantly fo- cus on levels of familiarity or difficulty of termi- nology in medical, biomedical or health domains. Term familiarity refers to a user’s subjective un- derstanding of term technicality. …

Content truncated to 500 words