Escapeling Chatbot - Adaptive Language Learning with AI-assisted Conversational Agents
Telegram chatbot game for English learners.
|Supervisor||Dr. phil. Tobias Thelen|
|Project initiated and lead by||Students|
|Participants||Esther Chevalier, Melissa Donati, Vera Klütz, Tobias Oberbeck, Kristina Sigetova, and more|
|Group size per semester||10-15 people|
|Timeframe||October 2020 - April 2022|
|Research areas||Artificial Intelligence, Linguistics, Natural Language Programming, Human-Computer-Interaction|
Aim and output of this study project:
The aim of this study project was to use conversational agents to help people learn English as a second language (ESL). Therefore, the idea was to create a game with a telegram bot that students can play with their friends to practice their English skills. In the game, the players have to collaboratively solve different English tasks to escape from an alien spaceship.
In the end, we had developed a working prototype with lots of code on our github repository (https://github.com/ALLUOS). We even have our own website (https://alluos.github.io) and will publish a paper about our bot.
Steps that brought us to our goal:
1. conducting market research to see what kinds of products are already available for ESL learners
2. deciding for a framework and relevant methods of how to realize the idea
3. designing English learning tasks and a storyline for a good user experience
4. programming the bot
5. testing the bot, also conducting user tests with external students
6. repeat steps 3-5 many times
7. writing a paper about our case study
Skills that we acquired:
- use of GitHub with a bigger group
- design of human-computer interactions
- planning and conducting user tests
- dividing work in a big team and putting it back together
- working within agile project management methods like scrum and kanban
Prerequisites and student backgrounds:
There were no prerequisites to participate in this study project. Many students taking part in this study project had a background in programming, but also students with no background in informatics were welcome. Other students had a background within UX, human-computer interactions or even ESL teaching.
We were working in smaller sub teams, every sub team had their own plan to accomplish a task of the overall project. All team members were meeting every Monday, where all subgroups presented what they had accomplished during the previous week. The meetings were conducted by the students themselves to foster an independent workstyle. For this project we used parts of the agile scrum framework, where each working period (sprint) lasted for two weeks. At the end of each sprint, we decided together on which tasks we want to work on during the next sprint and the team members assigned themselves to the tasks they want to be part of.
Create a concept which possibilities the bot has to adapt to the users. If possible, create a flow diagram of these processes to make it easier for the rest of the team to understand and discuss the concepts.
Lobby platform creation:
Come up with a concept how the users can find other players to play the game with. Plan how the users will get into the lobby, what functionalities it should have and how the users should interact with each other.
Grammatical Error Correction with NLP:
Think about and come up with a concept of how a free text input can be evaluated with the help of AI. Which models are working best and how can we integrate them into our app?
Here is the link to our website where you can access our telegram bot as well as find out much more about this project in detail: https://alluos.github.io/
We also have a GitHub page, to view our code results just visit: https://github.com/ALLUOS/ALLUOS_PUBLIC_APP