MITB Banner

The Recipe Behind Microsoft’s Beloved Chinese Chatbot XiaoIce

Talking to voice assistants or chatbots are more like asking something to do with no emotions attached which is far different from talking to a human. With the advancements in artificial intelligence and machine learning, the tech giants are always looking to develop something out of the box. In an effort to make these assistants more human-like in nature, Microsoft has come up a chatbot that not only assists you whenever you want but also conveys it emotionally.

AI-based chatbot, XiaoIce, translated to “little Ice” in Chinese is one of the ambitious projects of Microsoft, that was released by researchers in May 2014 at China. It’s uniquely designed as an AI companion with an emotional connection to satisfy the need for human communication and is catering to over 660 million users.

The Making

The design principle is mainly based on three specific steps:

To create IQ and EQ: A sufficient high IQ (Intelligent Quotient) to help users complete specific tasks and high EQ (Emotional Quotient) to meet the user’s emotional needs like affection, social belongings, etc. The aim is to enable it to identify a personality which is a set of characteristics such as behaviour, cognition and emotional patterns that creates a distinctive character of an individual.

The IQ capacities include knowledge, memory modelling, image, and natural language understanding, reasoning, prediction, etc. for the development of dialogue skills. The most sophisticated skill is the Core Chat that can engage long and open-domain conversations with the users. On the other hand, the EQ consists of two components, empathy (capability of understanding what the user is experiencing) and social skills. Social skills include query understanding, user profiling, emotion detection, sentiment recognition, etc.

Social Chatbot Metric: Unlike the performances of other bots that are measured by task success rate, this social bot use of Conversation-turns Per Session (CPS) as social chatbot metric to measure the success of long-term emotional engagement with the users.

Hierarchical Decision Making: A hierarchical decision-making process is cast so that the social bot interacts with the users over a sequence of discrete dialogue turn and at each turn, the bot observes the current dialogue state and chooses a response according to that dialogue policy.

Three Layers Of The System Architecture

  • User Experience Layer to connect the chatbot to different chat platforms. It includes a set of components like speech recognition and synthesis, text normalisation and image understanding that is used to process user inputs and responses of the chatbot.
  • Conversation Engine Layer is composed of dialogue manager, an emphatic computing module, Core Chat, and dialogue skills. The dialogue manager keeps track of the state of dialogues and responds in reference to the particular state.
  • Data Layer consists of the set of databases including the collected human conversational data., non-conversational data, knowledge graphs and the profiles of all the users.

Features

This is more than just a chatbot because it has certain features that are incomparable to other chatbots. Over these years, the chatbot has developed more than 200 different skills. This bot can predict what a user will say next to keep the discussion on and can communicate accordingly to the nature and state of the conversation. Here are some features of this amazing bot listed below:

  • Unlike other virtual assistants, this chatbot gives you a human-like feeling and tries to become your friend no sooner you start chatting.
  • It behaves the character of an 18 years old female persona that is reliable, sympathetic, affectionate, etc. and responds sensibly to sensitive questions by shifting skillfully to a new comfortable topic. It has strong language ability and visual awareness.
  • This awesome bot has the unique ability to create poetry. A book of 139 poems is generated by the chatbot titled as “The Sunlight That Lost The Glass Window”. The book was published in Beijing and is said to be the first-ever poetry collection written by artificial intelligence.
  • This bot is definitely many steps ahead in fashion than many of us since recently it has started designing images and patterns on fabrics for international fashion and garment producers.
  • It can create immaculate designs that are popular among young generations and can paint unique images by itself.

  • It has also been trained in a technique like reciting kid’s stories. The stories are automatically analyzed and it chooses suitable tones and characters to complete the entire process of creating a kid’s audiobook. This free-of-cost audio service is available for the users in Asia and is provided to five countries till now with different names like Rinna in Japan, Facebook messenger in India, etc.

Access all our open Survey & Awards Nomination forms in one place >>

Picture of Ambika Choudhury

Ambika Choudhury

A Technical Journalist who loves writing about Machine Learning and Artificial Intelligence. A lover of music, writing and learning something out of the box.

Download our Mobile App

CORPORATE TRAINING PROGRAMS ON GENERATIVE AI

Generative AI Skilling for Enterprises

Our customized corporate training program on Generative AI provides a unique opportunity to empower, retain, and advance your talent.

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox
Recent Stories