BasicCR

Identifies terms within a text using basic (as opposed to deep-learning) techniques

Installation

1. After installing NeuralCR, copy the files basic_text_matcher.py and basic_text_matcher_flask_loader.py to the NeuralCR directory.

2. Edit the file app.py adding the lines import basic_text_matcher and import basic_text_matcher_flask_loader to immediately below the line import ncrmodel_flask_loader.

3. Register the basic model loader by adding the line MODEL_LOADERS['basic'] = basic_text_matcher_flask_loader.loadfromrequest to immediately below the line MODEL_LOADERS['neural'] = ncrmodel_flask_loader.loadfromrequest.

Usage

• NCRModel interface conforming BasicCR models can be instantiated by calling basic_text_matcher.BasicTextMatcher(id_file, title_file) where id_file is the file path to a JSON file of an object that maps different names to a common identifier, and title_file is the file path to a JSON file of an object that maps these common identifers to their human readable names.

• If app.py is started with the --allow_model_put flag, BasicCR models can be instantiated through the REST API by making a HTTP PUT request to /models/<new model name> with the arguments of:

  • model_type = basic
  • id_file = same as above described id_file argument
  • title_file = same as above described title_file argument