Toward natural language understanding by machine reading comprehension

Abstract

The ability of artificial intelligence (AI) to comprehend text is becoming a major topic of discussion. While natural language understanding by AI poses a difficult problem, a significant improvement in AI reading comprehension has been achieved in recent years through the application of deep learning techniques. This article introduces machine reading comprehension technology now under research and development at NTT with the aim of achieving an agent that can understand business knowledge written in manuals, understand the language used by customers, and provide appropriate answers to questions.