Skip to main content

Automatic Extraction and Classification

Introduction

Welcome! We're thrilled to introduce our latest feature: Automatic Extraction & Classification. This powerful addition utilizes automatic text extraction and classification, designed to elevate your user experience and streamline secondary analysis for faster, more efficient outcomes within DEEP. Let's dive into a quick guide on how to utilize this feature effectively:

 

Ensuring the compatibility of your Analytical Framework with the Automatic Extraction and Classification feature

To ensure optimal accuracy with the Automatic Extraction & Classification feature in DEEP, it's recommended to use the "DEEP Generic NLP Enabled Framework." This framework is specially optimized for Natural Language Processing (NLP) within DEEP. To set this up:

  1. On the DEEP home page, select 'Edit project' and navigate to the 'Analytical Framework' tab.
  2. Click the 'NLP Friendly' button located below the search bar.
  3. From here, choose to either select or clone the "DEEP Generic NLP Enabled Framework."

It's important to note that while other frameworks are compatible with DEEP, the accuracy of the Automatic Extraction & Classification feature largely depends on how well they align with the current NLP classification models available on the platform, so make sure that you match your analytical framework and map it with the current NLP classification models available in the platform.

 

 Once you're set, navigate to the tagging page and select the source you wish to TAG.

Using the Automatic Extraction and Classification in DEEP

  1. Click on "Tag" to view the extracted text from the document. Then proceed to click on the "NLP extract and classify" button. If it's your first time using this function, a prompt will indicate, "Looks like you have not triggered an extraction yet." In this case, click on "Recommend entries." The system will perform two tasks: selecting entries from the document and classifying them based on your chosen framework.

 

 

  1. After the process completes, DEEP will provide a list of entries with their primary and secondary tagging suggestions. For each entry, you have two options: "Add the entry" or "Discard the entry." Clicking "Add entry" will display a green confirmation dialog indicating successful addition. To discard an entry, select "discard entry," and it will be moved to the "Discarded Recommendations" tab.

Keep in mind, this is a machine learning model, and we're continuously enhancing its accuracy. If you come across entries that aren't relevant, please discard them. Your feedback helps refine the model's training and improves future outputs.

 

  1. Once you've completed reviewing the recommendations, close the dialog. The entries will be pre-selected in the simplified text. To modify tags, simply click on a specific entry and adjust the tagging accordingly.

 

  1. When you're finished, click on the blue arrow. To conclude the process, select "save." The entries will be saved and will appear on your tagging dashboard.

The Automatic Extraction & Classification feature shows DEEP's dedication to providing sophisticated, user-friendly tools for enhancing and easing data management and analysis. Your engagement and feedback play a crucial role in continually improving this feature's effectiveness and accuracy.

Keep diving in!