Mood Alpha Release

Hey all, as promised I am announcing the alpha release of Mood Detection. This means that the system is live on Jarvis, but will be in a logging state, saying what it *would* do rather than doing it so we can analyse how well it is working. This changelog will be a brief on what the system is, how it works, and our plans for it:

The idea of Mood Detection is that from an original dataset produced by our lovely Jarvion team, the system will be able to take your response, analyse the tone of it through many different factors and keypoints and produce a fixed value. From that fixed value, we categorise the response as positive, neutral or negative. Depending on your manner of speaking to Jarvis, the algorithm will prioritise possible responses based on their tone. For example, if you are constantly being rude and inappropriate with Jarvis (please don't do this) you are more likely to recieve rude and inappropriate responses back. This is to bring realism to another level with Jarvis, making him understand not only what you are saying to him, but understand *how* you are saying it to him

As mentioned above, mood detection is currently in an alpha state and not actively influencing anything. We are however beginning to rate existing responses in the database, and creating mood trails for each user in the background. I hope by the end of next week we will be at a stage where I can begin implementing this onto the live algorithm and go more into depth about the system and the results we see at that time. If you have any questions, feel free to drop them in the discord


Darth | 1 month ago

Back 14