An advertisement targeting system for VK, the largest Russian social network. The project comprised three main parts: a Web interface to specify the target to advertise, an engine to predict best online groups for placing the advertisement, and a bot to push messages to selected groups. The engine used word2vec word embeddings to measure semantic similarity between advertisement keywords and group descriptions.
Server.py - Web-server to handle interface communication.
Index.html - Web-Interface itself.
FinalCode(copy).py, FinalSend(copy).py - VK-bot.
Just_a_good(copy).py - Engine.