Objective:

• Categorize hip-hop songs and artists by lyrical content and prosodic style.

Notebook Order:

• data_collection -> preprocessing -> topic_modeling -> clustering
• webapp folder contains files for flask deployment including Tableau dashboard embedding

Data Sources:

• Artist List: “List of hip hop musicians”, Wikipedia https://en.wikipedia.org/wiki/List_of_hip_hop_musicians

• Artist and Song Metadata: Genius API https://api.genius.com/

• Song Lyrics: Genius.com web scrape (API doesn’t support lyric requests)

Methodology:

• Data Collection
  • Scrape wikipedia for artist names
  • Genius API requests for top N songs by each artist
  • Genius.com scrape for lyrics to each song
• Preprocessing
  • Clean and tokenize text
  • Generate TF-IDF matrix
  • Calculate unique word and syllable rates
• Analysis
  • Topic Modeling
  • NMF Topic Modeling to create semantic categories
  • Combine with unique word and syllabic information and apply KMeans clustering
  • Aggregate artists’ song categorizations to characterize their style
  • Deployment
    • Recommender flask application
    • Tableau visualization

Link to Tableau Public Workbook:

https://public.tableau.com/views/Metis_Proj4_Viz/Dashboard1?:language=en&:display_count=y&publish=yes&:origin=viz_share_link