Won Bronze Medal (Absolute Rank 3) 🥉
This repository contains the final submission for the Adobe Challenge. The challenge involves behavior simulation and content generation, divided into two tasks. This README file provides an overview of the project, instructions for setup, and details about the implementation.
This task focuses on behavior simulation, where the goal is to give the user, an estimated number of likes that a particular tweet will recieve on twitter if the tweet is posted on the platfrom from the community. Therefore, the outline of this task was to make a machine learning model that will simulate the people present on twitter and give a numerical value that will correspond to the number of likes it may recieve on the platform.
This task involves content generation, aiming to generate the content of the tweet, given the metadata of the tweet, i.e. given the number of likes, content attached, time, usernames etc. predict what would be the tweet. This is beneficial if a user is not able to think about the tweet content, then it can just give the model the data and the target likes it wants, and the mdel will generate the content of the tweet.
Training Data for task 1 - here
Training Data for task 2 - here
Predicted content for the test dataset on all the different metric are provided in this folder.
A detailed description for these tasks, their approcahes and pros-cons of the different techniques used is provided in this report.