Skip to content

Adopting Generative AI for Personalized Endorsements in Tiktok Shop

Notifications You must be signed in to change notification settings

Hemant7499/tiktok-techjam-2024

Repository files navigation

Tiktok-techjam-2024

YouTube video.

Inspiration

Most social-media platforms support product endorsement by displaying advertisements on behalf of certified sellers. These advertisements lack target-audience knowledge and have limited area of creativity for the sellers. Such generic ads have a low probability of reaching the right audience.

TLDR; there are two main areas in tailored discovery to improve ad success:

  • Recommend specifc products to users based on user data
  • Allowing seller to choose an influencer for advertisement.

What it does

Our tailored discovery system is a two part AI service, where

  1. Knowledge Graphs helps sellers by endorsing their product in the form of a personalized advertizement, specifically to a users who are more likely to buy their products based on their activity/interactions on Tiktok.
  2. GenAI helps sellers to choose an influencer whose digital footprints are utilized to generate a personalized ad video, in a controlled manner.

AppWrite Schema

Checkout this schema diagram. Database Name: tiktok-clone

App Setup

git clone https://github.com/Hemant7499/tiktok-techjam-2024

  1. Setup an Appwrite account
  2. rename .env.example to .env
  3. Fill out credentials from Appwrite

Running the react front-end

Run the following commands.

npm i npm run dev

Running neo4j

Download desktop app for neo4j via this link Setup your neo4j knowledge graph by reading the documentation

Following libraries are required to run for generating ad deepfake

  • scipy
  • ffmpeg
  • openai
  • soundfile
  • liberosa
  • torch
  • transformers

Credits for base project to John-Weeks-Dev

About

Adopting Generative AI for Personalized Endorsements in Tiktok Shop

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published