Coursework for my Multimedia Systems and Applications class. I was given a dataset of 10,000 images from Flickr. I was tasked with developing a program that could recommend image tags, based on the sample data set. A user inputs a tag and the system will output the top 5 related tags. The system used a co-occurrence matrix technique, where tags are deemed related if they appear together for a particular image. E.g. if many images have "city" and "architecture" together, inputting "city", the system will suggest "architecture". The project processes data using .csv files for the 10,000 image data set, parses and stores it into data structures to create the co-occurrence matrix. It also calculates the IDF (inverse document frequency) to use a weighting factor to get more accurate results. Services like Flickr, Instagram and Youtube use something similar to this when users upload content.