AIBot

A basic express.js & MongoDB based bot, Dockerized for your convenience. It can provide answers based on a dataset and learn from user feedback.

Installation

Ensure Git, Node, Docker and Docker-compose are installed on your machine. Then: Git clone https://github.com/up730418/Bot Move to the Bot folder then npm install

Starting the bot

Simply run docker-compose up

Once you have started the server navigate to localhost:8080 and the bot should welcome you!

Loading Data to the bot

To add questions and answers to the bot uncomment services.js line 18. If this line is uncommented it will add a new data to the Bot when the container is started. You can then add/remove data to the object on services.js line 146 to populate the Bot.

Alternatively to this you can load data directly into the database in following way:

• Load questions in to the faq collection in following format: [{ "question": "Do you like Cheese", "answer": 1}]
• Load answers in to the answers collection in following format: [{ "answer": "No, no one likes Cheese", "id": 1}]

Done you should now have a fully functional Bot!! Happy Questioning!

Persisting Data

Data will remain within the mongo container until it is removed. To persist it beyond that uncomment the volumes section of the db service in docker-compose.yml. You may want to modify the location to suit your needs.

To Do

• Restructure database. The current implementation works but could be vastly improved.
• Update mongo and the JS mongo driver to current versions.
• Make it smarter. Currently only one no reply is needed to make the bot not serve an answer for a question. This could lead to the bot being abused fairly easily.
• Add a web interface for adding questions and answers.
• Improve the bots UI.
• Split up the getResponse function into smaller functions for easier maintainability.
• Create functions that will allow the bot to check the database for updates periodically/onchange so multiple instances can be created.
• Add Tests!