This repository is for web mining project: Music Artist PersonalRank Recommender based on social network and tag system In this project, we implemented 2 PersonalRank Recommenders, one with social network , one with social network and tag system. In the meantime, we implemented 3 basic recommenders as our baseline: Content-based RS, Iterm-based CF RS and User-based CF RS.

this project implemented the following featrues. We put every section in separate notebook in data/notebook directory. you can run in this order.

• Section 1: Download and Split data set
• Section 2: Tag Clustering
  • Section 2.1: Tag preprocessing
  • Section 2.2: Tag clustering: BERT
  • Section 2.3: Tag clustering: Levenshtein distance
  • Section 2.4: Tag clustering: user artist correlation
  • Section 2.5: Tag clustering Result
• Section 3: PersonalRank-based Recomender
  • Section 3.1: User similarity computation
  • Section 3.2: Graph Construction
  • Section 3.3: PersonalRank implementation: Based on Iteration & Matrix
  • Section 3.4: Example
• Section 4: User-based CF & Item-based CF Recommenders
  • Section 4.1: Calculate similarity matrix for user & item
  • Section 4.2: Predict rating based on item-based or user-based methods
  • Section 4.3: Recommend artists based on user preference
• Section 5: Content-based Recommender
• Section 6: Evaluation
  • Section 6.1: User-based metrics
  • Section 6.2: Item-based metrics
  • Section 6.3: Content-based
  • Section 6.4: PersonalRank without tags
  • Section 6.5: PersonalRank with tags

The project is organized as below. we put all the data in data directory. if you want to run the code from scratch , you do not need to download the data directory.

we use the data set hetrec2011-lastfm-2k. you can download the data set from this link: https://files.grouplens.org/datasets/hetrec2011/hetrec2011-lastfm-2k.zip or download from data/dataset. we also contain a code to download and unzip automatically.

• notebook
  • 1_splitting_dataset.ipynb
  • 2_tag_clustering.ipynb
  • 3_personalrank.ipynb
  • 4_user_item_basedCF.ipynb
  • 5_content_based_rank.ipynb
  • 6_evalution.ipynb
• data
  • dataset
  • split
  • tags
  • interim
  • external
  • result
• readme.md