nessmahm/DeepLearning-For-Influencer-Classification

Classify Instagram influencers using machine learning. Includes automated data collection, preprocessing, NLP and CV models training, and LLM-based RAG for insights into Tunisian influencers. # Influencers Classification Project ## Table of Contents - Introduction - Project Description - Features - Dataset - Installation ## Introduction This project aims to classify social media influencers into different categories based on various their posts. The goal is to provide a tool for identifying key influencers in specific areas for marketing and collaboration purposes. It's part of our end of year project at INSAT. ## Project Description Influencer marketing has become a critical strategy for brands to reach their target audiences. However, finding the right influencers who align with a brand’s values and target demographics can be challenging. This project leverages machine learning techniques to classify influencers into distinct categories to aid marketers in their decision-making process. The project involves several steps: 1. **Data Collection**: Gathering data from Instagram using APIs and web scraping techniques such as Selenium. 2. **Data Processing and analysis**: 1. **Data Preprocessing**: Cleaning and normalizing the data to ensure consistency and accuracy. 2. **Data Exploration**: Conducting statistical analyses on the dataset to uncover insights such as the distribution of post languages, top categories of content, and other relevant statistics. 3. **Modeling and Evaluation** 1. **Model Training**: Using NLP and Computer Vision algorithms to train models on the extracted features. 2. **Model Fusion**: Combining the outputs of two different models to improve classification accuracy. 3. **Model Evaluation**: Evaluating the performance of the models using metrics such as accuracy, precision, recall, and F1-score. 4. **Visualization**: Creating visualizations to represent the results and insights derived from the analysis. 4. **RAG and GEMMA Insights**: Utilizing RAG (Retrieval-Augmented Generation) and GEMMA to create a chatbot that can answer questions about Tunisian influencers. This chatbot leverages the data and …