Skip to main content

Classification Steps to Take

Abstract

Classification Steps to Take

This section contains a short overview and summary of the individual steps you need to take in order to be able to classify documents, emails, user accounts and more.

The Semantic Classifier is based on a number of background functionalities that have been designed to perform reliably and fast when the classification of large numbers of documents in an enterprise environment is aimed at.

To enable the classification and make results as accurate as possible, you need to train classifiers first that you later can use to classify more documents.

The following steps should be executed in this order:

  1. Create one or more Training Boxes (optional, but recommended, since documents can be reused).

  2. Add documents to the Training Box or boxes.

  3. Create one or more Train Classifiers .

  4. Add categories to the Train Classifiers .

  5. Add documents to the Train Classifiers .

  6. Train the classifiers and refine the training result to reach a score of 70+ % for f1, return and precision.

  7. Create a Test Classifier , based on the trained classifier or classifiers and test it using new documents.

  8. Use PoolParty's API to classify documents in large quantities efficiently and fast.

The topics listed here will contain the details and guide you through the necessary steps.