The multilingualization of terminology is an essential step in the translation pipeline, to ensure the correct transfer of domain-specific concepts. Many institutions and language service providers construct and maintain multilingual terminologies, which constitute important assets. However, the curation of such multilingual resources requires significant human effort; though automatic multilingual term extraction methods have been proposed so far, they are of limited success as term translation cannot be satisfied by simply conveying meaning, but requires the terminologists and domain experts’ knowledge to fit the term within the existing terminology. Here we propose a method to encode the structural property of a term by aligning their embeddings using graph convolutional networks trained from separate languages. We observe that the structural information can augment the semantic methods also explored in this work, and recognize the unique nature of terminologies allows our method to fully take advantage and produce superior results.

Abstract The authors propose a method for automatically generating Japanese–English bilingual thesauri based on bilingual corpora. The term bilingual thesaurus refers to a set of bilingual equivalent words and their synonyms. Most of the methods proposed so far for extracting bilingual equivalent word clusters from bilingual corpora depend heavily on word frequency and are not effective for dealing with low‐frequency clusters. These low‐frequency bilingual clusters are worth extracting because they contain many newly coined terms that are in demand but are not listed in existing bilingual thesauri. Assuming that single language‐pair‐independent methods such as frequency‐based ones have reached their limitations and that a language‐pair‐dependent method used in combination with other methods shows promise, the authors propose the following approach: (a) Extract translation pairs based on transliteration patterns; (b) remove the pairs from among the candidate words; (c) extract translation pairs based on word frequency from the remaining candidate words; and (d) generate bilingual clusters based on the extracted pairs using a graph‐theoretic method. The proposed method has been found to be significantly more effective than other methods.

Abstract In this article, we propose a method for the comparative analysis of concentration in author productivity distributions. We define the notion of concentration on the basis of two viewpoints (absolute and relative concentration) and select G (Gini's index) and V (the number of authors) as suitable measures for these two viewpoints. We then discuss the statistical peculiarity of author productivity data (i.e., most of the statistical measures change systematically according to changes in the sample size) and we explain our method using developmental profiles, which takes into account the sample size dependency of statistical measures. Finally, by applying it to actual data, we demonstrate the usefulness of the proposed method.

The paper proposes a method to describe conceptual meanings of Japanese deverbal nouns for the purpose of analysing Japanese compound nouns. A lexical meaning of each constituent word is very important since it deeply influences a word formation of compound words. Therefore, we intend to use the lexical meanings of constituent words to build a Japanese compound noun analyser. As a first step to describe lexical meanings of words, we have established a simple, semantic structure based on lexical conceptual structure (we refer to the structure we establish as TLCS). After the analysing 250 deverbal nouns that appeared in technical terms, we found that the only 14 types of TLCS cover various deverbal nouns. We explain the types of TLCS and also show the procedure of assigning TLCSs to deverbals. In order to evaluate the efficacy of TLCS, we made use of the accuracy of simple compound noun analyser for analysing technical terms. The result was very good, which shows that our proposed method is very promising.