On my way!

Dependabot

Description

 

Depenadbot is an SMS bot built using Twilio, Flask, and Google NLP.

 

Design

Every smoke session is attended by the cell phone. So why doesn’t it try to help you?

 

Vernacular & Tone

Dependabot’s vernacular is crucial in helping the quitting process by speaking with the user as joyful, not serious, and at the end of the day, a friend. Part of addiction treatment is giving the smoker a placebo by diluting the addiction to something that can be laughed about. In doing so, the user’s cost to quitting is minimized over time.

Dependabot’s tone is based on David Hasselhoff’s Knight Rider, where Hasselhoff’s vehicle, KITT, is an artificially intelligent electronic computer module in the body of a highly advanced, very mobile, robotic automobile with a not-so-serious tone.

Features

“I’M FINALLY FREE” Account Deletion Feature

Users must text Dependabot the phrase “I’M FINALLY FREE” to stop receiving text messages. This signal’s a subtle cost of lying to quit the bot, and most smoker’s who are on the path to quitting tend to be authentic with their actions in regards to efforts on quitting.

CRAVING Feature 

Text Dependabot “CRAVING” to trigger an instant crowdsourced tip on how to address nicotine cravings.

TALK Empathy Feature

The TALK command is followed by any free-form text the user would like to discuss. This activates the sentiment analysis capabilities.

SESH Feature

The SESH feature notifies Dependabot on the moments you’re smoking, allowing the system to track the times your cravings trigger in the future to better deliver tips.

SET DATE Feature

The SET DATE command instructs the bot to set a quit date the user has determined, and work to help the user meet that deadline.

Empathy and Google NLP

The main advantage to talk therapy in addiction therapy is the effectiveness of empathy. Dependable use’s Google’s NLP API to analyze the sentiment of a user’s response given any response, calculate the given sentiments, and respond empathetically by responding according to the predicted user emotion.

 

Categories: DATA SCIENCEPRODUCT