A research group consisting of Tatsuro Kawamoto, a researcher at the Artificial Intelligence Research Center of the Industrial Technology Research Institute, and Takaaki Aoki, an associate professor at the Faculty of Education, Kagawa University, has developed a system that can efficiently collect opinions from large-scale free-form answers using machine learning. bottom.
According to AIST, this system was named "voting clustering".Respondents are randomly sampled and presented with a certain number of other people's answers.Respondents determine whether those answers match their opinions, and if they do not, write down their opinions.
By doing this, a behavior graph of the respondents who judged whether or not the extracted opinions agree is created, and by processing this with a graph analysis algorithm which is a kind of machine learning, answers with similar meanings are summarized.The research group says that it will be possible to gather opinions even in a questionnaire with tens of thousands of people.
Traditional open-ended questionnaires have been difficult for analysts to read and process on a large scale.With this method, you only have to read a few representative opinions after automatically grouping the opinions, which can dramatically improve the efficiency of your work time.
In the future, the research group plans to provide a wide range of opportunities for large-scale free-form questionnaires through the site for demonstration experiments, and to repeat the demonstration experiments.The collected data will be analyzed by AIST and the results will be provided to questionnaire users.