[4Q1-GS-10] AI application:

[4Q1-GS-10-05] A case report on the application of document similarity to the analysis of survey responses

〇Toshio Tamamuro¹, Yutaka Takahashi¹ (1. Senshu University )

Keywords:Document Similarity, Qualitative intelligence analysis, analysis of survey responses

In analyzing qualitative information, it is extremely important to minimize the subjectivity of the analyst and to obtain objective and reliable results. To this end, methods are required to reduce the influence of the analyst's biases and preconceptions on the results. To address this issue, inter-document similarity (hereafter referred to as “similarity”) has attracted much attention. With the rapid progress of Natural Language Processing (NLP) technology in recent years, the accuracy of similarity has improved and its range of applications has expanded Similarity is particularly effective in the analysis of survey responses and writing test results. For example, when a teacher evaluates a written test, it is possible to minimize the subjectivity of the evaluator by checking consistency and deviation from the correct answers. Similarly, when a marketer analyzes the results of a survey, it is possible to suppress subjective judgments by verifying consistency and deviation from the hypothesis. In this study, we introduce a specific case in which such similarity was applied to the analysis of survey responses, and report the results of comparing the similarity between the correct answers and the responses. In addition, this research will report not only the advantages obtained through this method, but also the issues that need to be solved in the future.