Automatic Terminology Intelligibility Estimation for Readership-oriented Technical Writing

This paper describes automatic terminology intelligibility estimation for readership-oriented technical writing. We assume that the term frequency weighted by the types of documents can be an indicator of the term intelligibility for a certain readership. From this standpoint, we analyzed the relationship between the following: average intelligibility levels of 46 technical terms that were rated by about 120 laymen; numbers of documents that an Internet search engine retrieves using each term as a keyword from various types of websites (i.e. term frequencies). The result of the analysis shows that term intelligibility for a target readership can be estimated by regression analysis of the term frequencies weighed by the type of website. As pilot studies, we developed two regression models for estimating the technical term intelligibility for the target readership. One uses the machine learning method based on ν-SVR, and the other uses multiple regression. In order to evaluate the models, we used the results of a survey on laymen’s intelligibility levels for 50 new technical terms, and then compared the survey results with our estimated results. The results gave a correlation coefficient of 0.66 between the survey results and estimated results.