-
Response to negative message
-
Response to positive message
-
Overall negative conversation
Inspiration
In modern day society, any kind of phrase could be deemed as offensive, even if you don't realize it. Thus, TriggerBot was created, providing an objective evaluation of dialogue between you and an AI.
What it does
TriggerBot takes in messages from a user, analyzes the sentiment of the message, and responds with an appropriate reply. The message can be typed in or spoken, which is detected and turned into text.
How we built it
TriggerBot is modularized into several components:
- IBM Watson Analytics: Speech to Text service uses speech recognition capabilities to convert speech from the user into text for TriggerBot
- Lexalytics' Semantria: Sentiment analysis to detect mood from the user's message, and also TriggerBot's response
- Cleverbot API: Produces realistic replies to user messages to maintain a conversation
Challenges we ran into
Some of the APIs that went into TriggerBot lacked in-depth documentation. In order to understand and use those tools, we had to reverse engineer the provided wrapper functions. Implementing these components took some effort, but in the end, it was a great learning experience all of us.
What's next for TriggerBot
Since TriggerBot was created with the use of various APIs that use machine learning to produce results, it can be expected that the accuracy of TriggerBot's analysis will improve with time. Future features could include output with more detailed feedback on what part of the input was deemed as "offensive" or "courteous".
Log in or sign up for Devpost to join the conversation.