テキストマイニングを用いた統合報告書の評価モデルの構築法

抄録

The purpose of this study is to propose the method for constructing a model to evaluate the quality of the integrated report, called IRQ. The model, which is constructed by the proposed method, is a logistic regression model, that is trained on the descriptions of integrated reports selected and not selected by “Excellent Integrated Reports” of GPIF (Government Pension Investment Fund in Japan). As a result, first, it is confirmed from both quantitative and qualitative perspectives that the IRQ, which is calculated from the model, represents the quality of the integrated report. Second, two of three explanatory variables in the model are statistically significant, indicating that the model captures the descriptive characteristics of the selected integrated reports. Lastly, the model is shown to have a precision of 0.85, which is approximately 0.16 better than precision of a model, that is developed by the method introduced in the previous study. In conclusion, by using the model developed by the proposed method, it is possible to quantitatively clarify the difference in descriptive characteristics between the selected and non-selected integrated reports.