Twitter Helper�

Generative AI Hackathon by Lablab AI and Cohere

​

Team Lucifer

TABLE OF CONTENTS

01

02

03

04

05

Background

Problem statement

PROJECT GOALS

The solution: AI

Working demo

01

Background

OUR TEAM

Yasien Essam

Ahmed Hossam

We are 2 Senior biomedical engineering students with an entrepreneurial spirit and a burning desire to make cool stuff with NLP and generative models!

02

Problem Statment

• You are a content creator or a daily user on Twitter, wants to increase your traffic and share your content with people with the same interest.

​

• You find twitter character limit irritating and want to summarize your tweets.

​

• You are a user who wants to know which accounts to follow based on their content.

​

03

Project Goals

PROJECT GOALS

GOAL 1

Increase Engagment through popular generated hashtags.

GOAL 2

Summarize tweets so It can fit twitter character limit

GOAL 3

Analyze twitter accounts to visualize their interest and classify them.

04

The solution: AI

How does it work?

• We trained three Cohere’s X-large models on several training examples to get specific outputs to meet our project requirements.

​

• The hashtag model trained on tweets and their relevant hashtags.

​

• The mention model takes the output from the hashtag and suggest a famous account to mention to get high traffic.

​

• The Summary model takes an input of tweet which exceed 280 character limit and try to compress it to less than 280.

​

• We used Streamlit to build our simple UI to take user’s inputs

​

​

05

Working Demo

06

Future Work

• Increase the training examples to get better results.

​

• Get twitter Approval to use their API to scrape data from the website and get accurate results.

​

• Get the whole data of the user tweets to know his interests.

​

• Optimize the UI and UX

​

​

​

​