Case Study of Conversational UI ①

Child-rearing Support  Service “Metis”

May 2018 - January 2019  |  Design Research & Case Study
Designer & Researcher: Liang Yicheng

A Research of Design Method for Conversational User Interface.
This study has been published in The 66th Annual Conference Of JSSD with poster presentation & paper “Research of Design Method for Conversational User Interface” (会話型ユーザーインタフェースのためのデザイン方法に関する考察).

Conversational User Interface is a kind of interface that interacts between human and machines by voice interaction, and the information in conversational interaction is assisted by other technologies such as GUI, IoT, and HMI.
This study is intended to explore the design of conversational UI in the context of maternal and child health and parenting, and the possibilities of new Technologies such as AI assistants.

"METIS" aims to make "more convenient parenting" and "guard the health of infants". The prototype of the AI assistant "METIS" based on AI speakers is designed using Google Action platform.

Psychological Care
Relieve the pressure of parenting.
Give the mother and father the hard work to recognize.

Behavioral and life Care
Make childcare more convenient.

Physical and health Care
Provide health-related information.
Record data on parenting.

From the interview results of maternal and child care nurses, the troubles of parenting are mainly in the two aspects of "physical problems" and "psychological problems."

Physical aspects, such as the baby's physical growth problems, health problems such as illness prevention, and language problem. On the psychological side, the mother sometimes feels that the child is not cute at all, and the pressures from life & work.

The user persona in the parenting scene will tell us what specific needs they have and can be solved using the AI assistants.

Future Expectations
She hopes that children can grow up safely and happy.

Daily Goal
Take care of child's diet and life.
Find own interests and reduce the psychological pressure in her life.

Context
Sato will get up every morning to prepare breakfast for his family. Occasionally take a walk with her children in the morning, and then buy some daily necessities. In the afternoon, she will play with her children at home and look after them.

Scenario

Build a complete, smooth multi-touchpoint story.

METIS is an experienced maternal and child health nurse and a mother. Her advices is useful, and the communication with user are always smooth.

For questions & problems, METIS provide a lot of advice to choose, and work with many outside parties (hospitals, etc.) to solve the problem together.

From Scenario & Needs to Specific Dialogue Design

Two modes of dialogue between AI assistant and users

Built Prototype

with Google Home

I made a prototype of VUI by Dialogflow, and performed user tests and evaluations.

This time, we made a qualitative question in the form of five steps from “very agreeable” to “not at all agree”, and it was evaluated by 11 people.

As a result, the evaluations of “Feature,” “User preference,” and “User expectation” were high. The average value was exceeded 80 points.

In addition, the scores for “Userbility”, “providing advice”, and “Adaptability to talking with machine” exceeded 70 points. So there is room for improvement.

And, it was found that the score for "conversation" is 63.3, which is the lowest item.

User tests & Evaluations

Connect with Device

such as phone, car

Use AI assistant during navigation

Demo animation

In this research, we experimented with some design methods of conversational UI through the case study of AI Assistant "Metis" to support parenting.

Specifically, there are steps such as specifying user research needs, setting problem solving routes by scenario, extending dialogue samples to VUI flow, improving experience through user testing and evaluation. Among them, I thought that the system functionality and the rationality of conversation were improved by using methods such as "AI’s Persona", "Dialogue Samples", "VUI Flow", etc., to establish a conversation between human and machine.

We expect new technologies such as artificial intelligence and voice assistants to be used well in maternal and child health care and other fields.