This web application provides music recommendations based on your favorite artists or particular songs. It uses songs lyrics to find the most similar compositions

Musicmatch uses pre-trained word2vec word embedding model which results in a matrix with coordinates of songs in n-dimensional space

• Each word in songs’ lyrics were presented as a vector in n-dimensional space
• It uses distributive method - constructing a vector space based on information about the contexts of words
• It models semantic proximity - only the distance has the meaning in the model, not dimensions of space

Web application can provide recommendations based on the semantic similarity of song’s lyrics

It calculates cosine similarity between song vectors and prints TOP one with highest coefficients

##                        0071kWo4vzvKWe6dsxKeWj 00GnqoRAlvhz6JqwLiIflb
## 0071kWo4vzvKWe6dsxKeWj              0.0000000              0.7690408
## 00GnqoRAlvhz6JqwLiIflb              0.7690408              0.0000000
## 00HMDIr0RvLZ0BnzUfQ6Cb              0.4842672              0.6298551
## 00IZzVN1ipCGbR5nuNZ2RM              0.9372999              0.7379970
## 00KAoyuXPgEIw0x3yNMBY5              0.8736227              0.7941738