Resolution_Bot

Resolution Bot is a step above your normal virtual trainer. Talk to it, reason with it and bargain with it to gain the necessary empathetic skillset to face up to any difficult customer.

Comment

Inspiration

Facing a difficult customer must be one of the scariest things one can face. Especially when we do not have enough training, because of a lack of time in our fast-paced world, or because of a overtly stifling and formal training regime. Thus, we decided to come up with a bot that can help fill this training gap, and can change depend on context and actions.

What it does

Resolution Bot generates harassments scenarios for the user to properly manage. The user has to then attempt to calm the bot down through three mechanisms - calm confrontation, reasoning or sympathizing. Response bot takes these responses into account and generates a score for each reply. This reply is a function of the emotions detected and the toxicity detected by the emotions. After the user has calmed the bot down sufficiently, the training scenario ends.

How we built it

There is a speech to text translator, so that the user can articulate their statements as though they were actually in the scenario, and a text to speech translator so that the bot's response is more natural. Both were based off Google's speech services and pyttsx3 for the variation in voice. The AI relies on hugging face's transformer library to get the NLP models up and working.

What's next for Resolution Bot

Resolution bot has a great reach, being relevant to any sector with a customer focused branch. Thus, its long term use would be higher as a result.

We hope to make the app more robust, and branch out to an extensive range of different scenarios depending on a user's occupation. We can work with industries to better understand the nuanced challenges each workplace faces. A pose estimation model to detect a good body posture would also be an interesting path to explore.

Appendix

[1] Korczynski, Marek, and Claire Evans. “Customer Abuse to Service Workers: An Analysis of Its Social Creation within the Service Economy.” Work, Employment and Society, no. 5, SAGE Publications, July 2013, pp. 768–84. Crossref, doi:10.1177/0950017012468501.

[2] Sanh, V., Debut, L., Chaumond, J., & Wolf, T. (2019). DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter. arXiv preprint arXiv:1910.01108.

Share this project:

Updates