USEUROPEAFRICAASIA 中文双语Français
China
Home / China / Innovation

Chinese researchers create mood-based chatbot

CGTN | Updated: 2017-05-12 09:28

Chinese researchers create mood-based chatbot

Human-like chatbot. [Photo/CGTN]

How do you tell if the people you're chatting with online are real people, rather than text-generating robots? The struggle for bots to mimic idiomatic language means it can be pretty easy to spot. That shows how hard it is for researchers to develop a human-like chatbot.

But now, some researchers in China are moving one step closer, after creating a chatbot with five different mood types.

The moods include happiness, sadness, anger, disgust and loving. The bot can reply to people's words differently, according to the mood it has been set on.

A leader of the project, Professor Huang Minlie at Tsinghua University, calls the bot an "Emotional Chatting Machine" (ECM).

He typed a critical comment about Valentine's Day into the machine as a test. The machine replied "Happy Valentines Day!" when set on a happy mood, and complained "you're just showing off!" when set to angry mode.

"It's still an initial exploration in the area, as ECMs can only talk in text. It doesn't have a voice output or facial expressions," Huang told Xinhua.

"But researches are moving rapidly on creating human emotions," he added.

So how do ECMs work? The research team created an emotion classification algorithm, and used it to classify all kinds of comments on social media websites.

So words from an ECM could be lifted from something you said online last week, and that's why their reactions sound so much more human-like than other bots.

Editor's picks
BACK TO THE TOP
Copyright 1995 - . All rights reserved. The content (including but not limited to text, photo, multimedia information, etc) published in this site belongs to China Daily Information Co (CDIC). Without written authorization from CDIC, such content shall not be republished or used in any form. Note: Browsers with 1024*768 or higher resolution are suggested for this site.
License for publishing multimedia online 0108263

Registration Number: 130349
FOLLOW US