Briggs599 / appSHNE

The application of representation learning for semantic associated heterogeneous networks in creating android app embedding layers.

appSHNE: The Application of Representation Learning for Semantic-Associated Heterogeneous Networks in Creating Android App Embedding Layers

Paper: https://briggs599.github.io/

3.8.2021 updates - Alex

• wrote EDA notebook that is callable from command line
  • Run EDA with the following command line parameter: -eda
  • EDA can be run with the following parameters: time and limit
    • python run.py -eda -time will run the EDA and print the time to run it on completion
• Cleaned old code and adding documentation
• To do:
  • Clean up parameters in config/params.json and delete unused parameters
  • Remove unused methods
  • update dockerfile with nbconvert and pandoc to run EDA.ipynb from command line
  • Run EDA on 1000 apps

3.5.2021 updates - Alex

• added argument -log for the <redirect_std_out> (save console output to log file) parameter
• Moved SHNE_code to src directory

3.2.2021 updates - Alex

run.py has been updated to include more command line arguments

• -t, -test, -Test: Run on test set
• -node2vec, -n2v: Run with node2vec instead of word2vec
• --skip-embeddings: Skip the word embeddings stage
• --skip-shne: Skip SHNE model creation final step
• -p, -parse: Only create node dictionaries dict_A.json, dict_B.json, dict_P.json, dict_I.json, api_calls.json, and naming_key.json
• -o, -overwrite: Overwrite previous node dictionaries created when parsing
• --save-out: Save console output to file
• -time: time how long to run main.py

Updated params config file. All parameters used are now found in config/params.json.

• All outputs will be saved under the values for <out_path> and <test_out_path>
  • Subdirectories to save configured in respective dictionary.
    • For instance word2vec embeddings will be saved under the path <save_dir> in the <word2vec-params> dictionary int config/params.json
• All filenames parameterizable