SlackInspector

This is a slack bot which will add monitoring on the required slack channels and report the messages with details if they are found offensive.

Overview

This is implemented using

1. Hubot - a scriptable chat bot
2. MongoDB
3. IBM Watson AlchemyLanguage API

The dashboard is implemented using

1. AngularJs
2. same MongoDB instance

Installation Details

1. Node 6.x and MongoDB have to be installed.

2. Create a bot integration in Slack web page as mentioned below:

3. Save the integration by filling the details and save the Slack token generated.

4. Run the following commands for building required node modules and starting the bot:

   npm install
   HUBOT_SLACK_TOKEN=xoxb-************-************ ./bin/hubot --adapter slack &
   

5. Add the bot as a normal user in the slack channel where monitoring has to be done.

Functionality

1. The bot will monitor every message in the channel and checks whether it has any foul words in it. This would be first level of filtering^[1].
2. If any bad words are found, then the message will be sent to IBM Watson AlchemyLanguage API to get the emotion levels of the message^[2].

1. If the emotion results of Anger/Disgust/Fear turned out to be greater than 30%, then the users in the Slack channel will be notified with a link to report this. (The user will be unaware of the entire process)

1. Upon clicking the link, the report will be submitted. (Stored in MongoDB)

Admin Dashboard

Admin can login to a web application to see the submitted reports and other statistics.

User Dashboard

User can see the usage statistics of the words entered with timely displayed graphs.

References

[1] . Foul Words Dataset - #link#

[2] . IBM Watson Developer Cloud - AlchemyLanguage API - https://www.ibm.com/watson/developercloud/alchemy-language.html

Built With

• batchfile
• coffeescript
• css
• html
• javascript
• shell
• typescript