Using CountingGridsPy Library

CountingGridsPy

CountingGridsPy is the machine learning library that supports BrowseCloud. CountingGridsPy ships with Python scripts that transform raw text data into the files that the BrowseCloud application consumes. At a high level, these scripts tokenize, clean the tokens, vectorize the tokens, run the learning algorithms, and produce model files.

Clone the code from here to get started:

 git clone https://github.com/microsoft/browsecloud.git

Input

Requires:

• Python 3.6.7+ (You can get it from www.python.org)
• Your shell is in the EngineToBrowseCloudPipeline directory.
• Dependencies installed as shown below in the dependencies section.
• engine_type is "numpyEngine".
• inputfile_type is "simpleInput" or "metadataInput".

inputfile_type	Description	Example
"simpleInput"	path_to_output_folder contains a TXT file called inputfile_name with the following schema: {title, content (i.e. abstract), and link}. It does not have a header row. If you do not have a title, then leave it blank, followed by a tab. If you do not have a link, leave it as blank, after a tab.	https://github.com/microsoft/browsecloud/blob/master/CountingGridsPy/ExampleInputs/MuellerReport.txt
"metadataInput"	path_to_output_folder contains a CSV file called inputfile_name with the following schema: {alias, title, abstract, responseId, surveyId, link, image} with a header row. This type allows the user to correlate topics with metadata beyond sentiment analysis.

• inputfile_name must reference a csv or a txt file, as shown in the table.
• PYTHONPATH environment variable setup to search for the CountingGridsPy module.
• extent_size_of_grid_hyperparameter and window_size_of_grid_hyperparameter are integers such that the former is greater than the latter. The quotient of the square of extent_size_of_grid_hyperparameter and the square of window_size_of_grid_hyperparameter is the number of topics the model can carry (i.e the model's capacity).

API:

 python .\dumpCountingGrids.py path_to_output_folder extent_size_of_grid_hyperparameter window_size_of_grid_hyperparameter engine_type inputfile_type inputfile_name

Example:

 python dumpCountingGrids.py data_folder 24 5 numpyEngine metadataInput channeldump.csv