Bella 3.0
Chat bots and other interactive conversational agents are making everyday life easier, efficient, engaging, and interesting. While searching for a simple piece of information on the web or completing a task becomes boring and uninteresting at times. But the invention of interactive conversational agents, which engages and guides the user to complete the task, has positively impacted and enhanced the boring experience of searching the web or completing a particular task, making it more fun and engaging. The very famous examples like Cortana, Siri, Google Home and Alexa have added an interesting element to our lives.
Design Intention:
Social and cultural events provide a platform for people with similar thinking to connect, share interests, discuss on various topics and gain knowledge. This in turn helps in holistic development of individuals. Students of Srishti Institute of Art, Design and Technology get to know about the events through posters in the campus, social media, web searches and word of mouth. There are a lot of physical posters in the campus but many times the posters are not well organised on the walls, there are outdated posters mixed with the upcoming event posters, which creates confusions in the minds of the students. The presently available mediums do help the students to look for events and programs but do not provide an engaging and an encouraging experience for the user. Sometimes students do want to go for events but they just can’t find the right event at the right time. Hence to keep the students updated about the upcoming events and programs and to encourage them to participate in the events and workshops, we aim to augment the experience of looking up for events using posters in the campus by designing an interactive conversational agent which understands the user and suggests events accordingly to the students based on the user’s likes and interests.
Initial experiment:
In Interaction design, it becomes extremely important to work on Human feedback. Feedback helps us define our product in a better way, sometimes change the target audience, sometimes specify them, sometimes change our product to make it more “Humane” but sometimes to change motive of the product. Initially we were quite satisfied about our prototype but as we displayed it to the other users, we got different kinds of reactions and various feedbacks and critics.
Our initial brainstorming and experimentation started with a problem statement, where we further realised that flow towards the solution was not in the right direction. Our initial statement was — “Missing out on interesting events creates an unpleasant experience and an unsatisfying experience for the user. The core reason for the problem of unnoticed posters is that the posters lack proper organisation, timely updates, detailed information and effective user engagement. To overcome the problem of unnoticed and unorganised posters in the campus we are planning to design an Audio Visual Conversational Interface named ‘Bella’.
During feedback user testing sessions we got interrogated about — ‘are we questioning the beauty of posters? Or the way they are laid down? Or the fact that few people pay attention to them? Or the way they are not updated and taken down when event is over?’. Through these questions we were able to think critically about ‘are we forcing a problem or it actually exists?’ While designing the solution we realised that we were focusing only on the solution and were trying to include more and more features and forgot the core of our problem. During the designing process initially, we came across various other problems and kept adding different features to our prototype to enhance it and made a solution for multiple problems. Hence our focus was lost. Our flow from the design intention was taking turns as it reached the solution. As a result the solution was not able to create a great impact on the audience. Feedback and user tests helped us shape our product in a better way, to define it with constraints, to think about our final concept and prototype from various perspective and considering different scenarios.
Final Concept and Idea (Working):
Bella provides a platform for the user to interact with his convenience and preference to get to know about events in and around Bangalore. User can converse (through the medium of text/audio speech) with Bella in 2 different ways:
Working of part A- “Specific event”
When user is aware about his own preferences about event, what he is looking for and where he wants to go, then he can choose to select the “Specific event ” option. Here the Bella converses with the user by asking him only specific questions directly asking him about the ‘preferred location’ and ‘category of event’. After the user has entered his specifications, Bella searches and displays only the filtered events that the user is looking for. Based on the search of events in the nearby dates, user selects the desired events and Bella sends the event details to the user’s email id.
Working of part B — “Suggest event”
When the user is looking for events and not exactly sure where he wants to go or what he wants to do. Then he can select the “Suggest event” option. Here Bella converses with the user by asking him ‘This or that’ fun quizzes/questions to understand user likings and based on the user responses Bella intelligently suggests events to the user. Bella smartly frames ‘This and that’ questions based on its intelligent event search engine. In such a scenario, when the event is fed, it can be tagged with various tags available. Tags are part of questions. For example, for a trekking event tag would be adventure which fits into question: Adventurous or Cautious? Hence if user chooses ‘adventurous’, and his other answers are also related tags to trek, he will be displayed event of Trekking.
Features:
Bella’s Personality: We have defined Bella’s personality by keeping her a little friendly. We have tried to use a conversational style in which youngsters talk to each other. At the same time, we have kept in mind that the user is aware of Bella being a machine and marked a limit to restrict ourselves from making Bella seem very unnatural and humanly more than required.
User Control: User has a control over Bella. To start a conversation with Bella the user needs to prompt by saying “Yo Bella” and Bella starts responding. Bella replies on the basis of what the user responds and guides the user based on user preferences.
User encouragement: Bella aims to encourage the students by showing specific event. Bella also suggests events based on personality testing questions.
User engagement: Bella tries to engage the user by asking him interesting fun questions in order to suggest events. In order to enhance the conversation, Bella may suggest the user an interesting fun task at the end of the main conversation (For example, send a postcard to a friend, climb on a tree) The fun suggestions encourage the user to engage in positive activities which contributes in his holistic development.
User accessibility: A digital screen of the conversational agent will be placed inside N5 Campus and one each in other campuses. Along with that Srishti students will also have an access to Bella on their laptops through the Srishti Wi-Fi Intranet. Hence conversational agent can be accessed from any place at any campus.
User Flexibility: User can access Bella at any point of day during the college hours.
User privacy: Bella believes in maintain privacy of the user. A set of earphones will be provided with the digital device that will be put up in the campus premise. If user feels uncomfortable in conversing through audio, user can choose “Text only” option through which conversation will take place through text messages. No records of the user entered information in the form of text or audio are stored in the device.
Prototype:
To define our conversational agent, we have created a prototype using ‘Wizard Of Oz’ method. There were three stages involved:
- Building an interface prototype using a prototyping tool, Marvel app. User was made to interact with it and conversation happened with a person hiding nearby.
- Building a basic prototype for conversation in Python leading to suggestion of events by conversational agent. The conversation started with agent asking user “What are you up to?” and with identification of keywords in user’s response, conversation was built up to lead to suggestion of events.
- Building of structured and final prototype for conversation using Python leading to either suggestion of events based on user’s personality or user specifying about his needs and agent responding on those choices and video to show the prototype. The machine voice element added in the video is not a part of the present prototype but is just representation of the final device.
Code:
Code was done in two stages. Stage one had conversation mapped in Mapping- I and was executed using if, elif and else nested loops.
Stage two was much more structured and had two parts- suggest and specify. In specify part, user could specify choices according to options available and program will publish results by matching their choices to dictionaries of all the available choices and will print the corresponding result. Placeholders were used to identify the input and match it with dictionaries.
If for some choices, events are not available, reset function would be activated asking user to provide their choices again.
After user is displayed with list of events and user has selected the events he wishes to go to, user can also choose to send their event details to their own email-id’s. This was made possible by importing SMTP library which connects Python to Google’s mail service. A g-mail account for Bella in the name of srishti.bella@gmail.com which was used and is being used to send event details.
To make conversation interesting, a small code was added where Bella suggests an interesting activity to do to user. Some activities were added and using random, they were made to print randomly.
For suggest part, nested if and else was used to create quiz to know personality of user. There are two quizzes from which one is selected randomly.
Mainly functions and lists were used to enhance program’s functionality.
Demo Video:
Scope and future aspects:
Bella can be developed further by enhancing her features for a better user experience. Following are a few futuristic concept ideas:
1. A camera scanner installed inside the device, which can scan the users expressions/accessories to understand the user in a better way and suggest him events accordingly.
2.An Application of Bella can be developed for users convenience.
We aim to work on the device in detail by focusing on its primary intention by keeping it simple and user friendly, hence creating a path for future innovations and development.
Project by Apurva Ulabhaje, Rachna Bodhe and Rashika Goyal