[1K1-GS-6] Language media processing: evaluation / anaysis

[1K1-GS-6-02] Examination of a Method to Evaluate Actions to Be Taken from Sentences by Using BERT

〇Satoru Kai¹, Jiacheng Li¹, Masato Noto¹ (1. Kanagawa University)

Keywords:NLP, BERT, Word-Embedding

Recently, while 5G is becoming more widespread and the use of smartphones outdoors is accelerating, there is a concern that it will be difficult to grasp the situation immediately due to the large amount of information that can be obtained. In addition, it is necessary to consider the credibility of the content. BERT is a pre-training model that uses the Encoder part of a bidirectional Transformer to convert an input sentence into a word-embedding expression string. With BERT, the performance of various natural language processing systems has been greatly improved, but it is necessary to consider how to use BERT for actual tasks according to each task. In this research, we propose a method to replace the network information of modern society with the world of TRPG and judge the action to be taken at that time from the word-embedding expression string by using BERT. The use of this method is expected to be useful in reality rather than on a scaled stage.