Scientists from the Osaka University have developed a method for digital dialogue systems. The new technical is called lexical acquisition through implicit confirmation; and it is a method to allow a computer to acquire the category of an unknown word, achieved over multiple dialogues. This is process achieved by confirming if the computer’s predictions are correct in the flow of conversation.
The research is not simply of academic interest, it has a practical application for automation and digital transformation. Many types of conversation robots, chatbots, and voice assistant applications are central to progression in the digital age. However, these systems are limited by virtue of the computers often simply answering questions based on whatever information has been pre-programmed.
The downside of this is that a computer learns from humans by asking simple repetitive questions. Yet if the computer asks only questions such as “What is this?” in order to acquire knowledge, users will not sustain interest in communicating with the computer.
Instead, a more advanced from of communication can be used. The new method intends for a computer to predict the category of an unknown word, gathered from user inputs, during conversation. Moreover, the system allows a computer to make implicit confirmation requests of the user; and to request that the user responds back to these requests. In doing so, the artificial intelligence gains knowledge about words by engaging in dialogues in real-time.
This means chatbots could begin to speak by learning from a conversational partner according to a given situation. This will lead to more interactive services and dialogue systems being developed, a situation where the computer becomes smarter through meaningful conversation with humans.
