Abstract
In order to improve a discussion management for local public transportation policy, topic extraction from discussion records should be elaborated enough to clarify the topic transition in a discussion. This study applied topic model developed in machine learning to the discussion records in local public transportation management, which is collected from 5 cities in local Japan from 2012 to 2015. In our application, 25 topics were obtained from 104,696 words. The obtained topics were well decomposed into different set of vocabularies, and enables to demonstrate topic transition.
