slad3/socialmediadataanalyzer

API for analysis on user downloadable social media user data. This currently includes messaging, search history, device logins. Yes, social media sites keep all this data.

Social Media User Data Analyzer

Server for analyzing downloadable Social Media user data.

You can download your Social Media user data here:

Facebook: https://www.facebook.com/dyi/?referrer=ayi

Instagram: https://www.instagram.com/download/request/

To run for yourself

Installing required packages

pip install -r requirements.txt

From here you can either run App.py to run the server in debugging mode.

Overall Process

When the user uploads the zip file provided by given social media platform, the server unzips the file into a temporary directory on the server.

Once the process finishes and is analyzed, the temporary directory is deleted along with any user data that was provided or processed during that session. No user data will be saved, everything is deleted immediately.

Analysis

The end goal of each analysis is to provide a json format that can be easily readable by a frontend service, whether that's the website frontend or anther service designed to analyze user data over a period of time.

Searches

The end result of analyzing the searches will be a JSON format of getting the frequency of each search, and providing a comprehensive histogram of each search.

Messages

No or content of the user messages will be collected, analyzed, or stored. That's creepy. Only what type of message (text, photo, video etc.), sender, and time will be analyzed.

Analysis of the messages revolves around the response time of each message.

This includes how fast each of you respond, the ability to keep a conversation going, if the relationship between you and the recipiant is one sided, who usually initiates the conversation, etc. This also includes the grouping of messages to find somewhat of conversations.

• Response Times
• Amount of double messaging
• Most initiations of conversations
• Least self initiations conversations

Technical

All compilable code is written in python. Server is written with Flask/Flask-Cors and analysis utilizes NumPy, Pandas, and Scikit-learn.