Application of 3-gram to English Statements Posted at Q&A Sites to Obtain Factor Scores

The widespread diffusion of the Internet has enabled us to communicate on Question and Answer (Q&A) sites. While there have been increasing numbers of registered users and posted Q&A statements, the issues of mismatches between the intentions of questioners and respondents are becoming conspicuous. In order to solve these problems, nine factors for Japanese Q&A statements were obtained through impression evaluation experiments. Syntactic information was then extracted through morphological analysis (MA) and employed as the feature values in performing multiple regression analysis to obtain factor scores for any Q&A statements. Factor scores can be calculated by utilizing N-gram applied to Part-of-Speech extracted through MA. Since most of the methodology so far have mainly depended on Japanese, the validity of the method applied to other languages must also be inspected. So far, in a similar fashion as in Japanese, the nine factors were experimentally obtained from English Q&A statements. The factor scores were then calculated utilizing feature values of English statements extracted through MA. It has also been confirmed that N-gram could be applicable to calculate factor scores instead of syntactic information using mere MA. Therefore, in this paper, 3-gram, as well as MA, was utilized as the feature values of English Q&A statements to obtain factor scores. Through multiple regression analysis, it has been shown that applying 3- gram would lead to as good accuracy as mere MA for all nine factors. Hence it could be concluded that 3-gram would also be applicable to the methodology when applied to English materials.