Improvement of Estimation of Objective Scores of Answer Statements Posted at Q&A Sites
Yuya Yokoyama, Teruhisa Hochin, Hiroki Nomiya · International Journal of Software Innovation · 2013
To eliminate mismatches between the intentions of questioners and respondents of Question and Answer (Q&A) sites, the authors have clarified the characteristics of the question and answer statements. It has been shown that the impression of the statements could be captured by nine factors, and the factor scores could be estimated from the feature values of the statements. Here, the objective scores of answer statements are provided. So far the authors have tried to estimate the objective scores of answer statements through multiple regression analysis. They are estimated from the factor scores estimated by using multiple regression formulas already obtained. As a result, with the consideration of natural logarithm, objective scores of the statements on love counseling & human relationships could be estimated good. The estimation accuracy of the statements on Yahoo! Auction and PC, however, is required to be improved. Therefore, this paper tries to improve the estimation accuracy of objective scores of Yahoo! Auction and PC. Here, the authors adopt Steven's power law as well as Fechner's law as new explanatory variables. In considering multicollinearity, the authors selected as many explanatory variables as possible. As a result, objective scores of the statements of PC are estimated fairly good.