PeerLogic Web Services: Peer-reviews-NLP

Summary

The Peer Review NLP Web Service (PRWS) utilizes natural language processing to perform sentiment, volume, and emotion analysis upon student peer reviews.

The service was created with the intention of being set up using Docker and with the HTTP server created using Gunicorn.

The most recent Github featuring the work for this project can be found here.

Components

The project features the following primary directories:

• data: A storage space for tagged comments. Utilized for X.
• documentation: A number of Swagger-supported .yml files which outline the web service's API.
• metrics: A number of files which handle the input and output from the webservice, reading JSON input and returning it using different methods.
• model: Several non-human-readable files representing the compiled models utilized by the web service.
• preprocessing: The primary "runner" files which perform the actual prediction associated with the web service.
• templates: Contains a single, non-functional HTML page apparently meant for testing the NLP.

In addition, it also contains a number of files associated with initial start up. Each of these sections will be discussed on independent pages, all linked here for ease of navigation.

Page Links

Please click one of the following links to read more about PRWS's:

• Set-Up
• Input and Output Handling
• Web Service API
• Prediction Code
• Model Details
• File Breakdown