A New Technique for Identifying Scientific/Technical Terms and Describing Science Texts: (An Interim Report)
Y. HUIZHONG · Literary and Linguistic Computing · 1986
A new technique is described in the paper for identifying scientific/technical terms It can also be used to characterize text types At the speed at which science and technology develops today, there appear every year a great number of scientific/technical terms As type-setting tapes and other machine-readable texts are becoming readily available, it becomes attractive to consider the possibility of automatic term collection There are broadly two basic types of terms single-worded terms (e g ‘hardware’, ‘software’, ‘terminal’, etc) and multi-worded terms (e g. ‘programming language’, ‘radio wave transmitter’, ‘hydraulic power transmission system’, etc) Since terms are highly subject matter specific, it is possible to identify single-worded terms on the basis of their frequencies of occurrence and distribution Multi-worded terms are identified on the basis of their collocational behaviour An algorithm is given for the identification procedure. The 2-word combinations recovered from the texts have shown remarkable difference between science texts and general English texts such as a novel, which can be used to characterize text types. The possibility is also discussed to develop a long term automatic monitoring and collecting system of scientific/technical terms so that the term bank can be updated on a yearly basis Another possible use of the present technique is automatic abstracting or indexing of science texts