🐛 Bug Report

Remove empty fields from raw_review as elastic search does not allow empty fields.

Currently fawkes only has classification for user reviews, which requires either static weights if we are using text-match or already labelled data to use the LSTM based classification model.

It would be nice to have an automatic clustering of user reviews to get an initial sense of all the topics within the data set.

The important thing is to have a model which can be saved and reused in the future for further clustering when new user reviews are added to the dataset.

Given any search query, add the support to fetch data and save it in specified format (json, csv)

Resource: https://www.elastic.co/guide/en/elasticsearch/reference/current/rest-apis.html

Going though a lot of user reviews takes time. Having a summary view of all the issues would be a great feature to have.

Currently the App. Config. app/sample-mint-config.json is converted directly to a class object. Need to add a schema validation using JSON schema for it to be more robust and catch errors before running the code.

🚀 Feature Proposal

Currently fawkes steps are exposed through the CLI as separate commands which directly take in configuration from fawkes's config file. This needs to be changed to a plugin based architecture where each plugin has it's own config and a person running fawkes can specify which plugins to run etc.

• Ideas:
  • Each runnable plugin can be automatically exposed as a CLI command ?

Motivation

• Modernization in general

🚀 Feature Proposal

A lot of user reviews may go into splunk. Splunk exposes an API which can directly be used to achieve this.

Motivation

To track those user reviews which go into splunk.

Example

Splunk can now become a data source.

requirements.txt should ideally have fixed versions for the packages listed in it to avoid any breakages in future.
Eg.
A installs B (1.5)
C installs B (3.5)
'A' could possibly break in future because it is incompatible with B (3.5)