Environment setting

• Put data in mdp/data inside main directory
  
• Put challenge set in main directory
  

How to build recommendations:

1. run build_structures.py
2. run full_kernel_songs.py
3. run create_Ku.py with parameters 1 and 0.5
4. run create_pl2title.py
5. run words_similarity_builder.py
6. run calc_P.py
7. run titles_similarity_0.py
8. run user_based_MSD_1.py
9. run item_based_MSD_5_10_25.py with parameters 0.7 0.4 5
10. run item_based_MSD_5_10_25.py with parameters 0.7 0.4 10
11. run item_based_MSD_5_10_25.py with parameters 0.7 0.4 25
12. run selected_KOMD_100.py with parameters 100 50000
13. run merge_csv.sh

Estimated memory: 80Gb
Estimated time: 20 hours
Estimated disk space 150Gb

Most of the steps above can be parallelized as shown in the following figure. The estimated time is calculated considering a sequential execution.