Extractions from AI models

New Konfuzio feature compares AI results

Maximilian Schneider

When analyzing documents, it is important to work precisely and efficiently. Konfuzio helps you achieve this by extracting valuable insights from large amounts of data and facilitating interaction with AI models for document extraction. The new feature amplifies these benefits by allowing you to monitor any AI in any document. The result is increased transparency and better results, as you now have detailed insights into the functioning and effectiveness of the AI models. This ensures that your document analysis is even more accurate and efficient.

More transparency and more precise results through the comparison of annotations.

Previously, users could only access the extractions of the current model. And this was only possible if the user had uploaded the document from scratch. So until now, the "pure" annotation of the extraction AI was not visible on existing test and training documents. From now on, however, a more comprehensive understanding can be achieved by examining the nuances of multiple models.

Comparing annotations made easy

Konfuzio offers the flexibility to train multiple extraction AI models tailored to each document category. These models evolve over time, improving their performance and providing more accurate annotations. Since the performance of the models can vary depending on the document, Konfuzio offers a new feature that facilitates comparison at the document level. Users can use the Document Viewer seamlessly access previous extractions where annotations from each model are available for comparison.

By navigating to the "Filter" section and selecting the "Version" drop-down menu, users can examine the annotations from different models applied to the same document. This dynamic feature allows you to observe in real time how the annotations change when switching between models. Below you will find a brief run-through of this new function:

Understanding the difference between extraction and annotation

As Annotation is a mark in a document that was created by a person or a machine. A Extraction is the "unchanged" output of an extraction AI, which is stored in the background.

When extractions are displayed on a document, they are converted to annotations.

Please note: Only the term "annotation" is used in the web interface.

Event information - Incubator Start-up Center Passau

Join us on February 29, 2024 at the Passau Start-up Center for an event with the CEO of Konfuzio, Christopher Helm. Explore the topics of document processing and AI with industry experts and network with regional leaders.

Business Incubator Passau Economic Forum Event

-> Secure your place in Passau now.

What comes next?

Stay up to date so as not to miss the upcoming releases and register for our updates:

Your subscription could not be saved. Please try again.
Please confirm your registration by clicking on the link in your email inbox.

Want to stay up to date?

Subscribe to our bi-weekly updates.

We use Brevo for sending our updates. By submitting this form you agree that the personal data you provided will be transferred to Brevo for processing in accordance with Brevo's Privacy Policy.

Soon, users will be able to run models and view results via the Konfuzio API. As you have read, exciting new features and events are just around the corner. So Konfuzio remains at the forefront of document analysis and continues to adapt to ongoing user needs.

-> Also explore our section "In Progress" in the changelog for detailed insights into our ongoing developments.


If you have any questions, our experts will be happy to help and advise you. Please use the following form to contact us:

    About me

    More Articles

    Oracle DB Release Konfuzio

    New at Konfuzio: Oracle Database Integration & further features

    How does the latest Konfuzio Server update change the landscape of Intelligent Document Processing Software? We are very pleased to announce our new...

    Read article
    Software Release Update Tech News

    Performance improvement for all AI models at Konfuzio

    At Helm & Nagel GmbH, we have improved the performance of Konfuzio and introduced a number of new functions. Our...

    Read article
    System news Feature Release Update

    Effective communication with System Messages in Konfuzio

    We are pleased to announce a new feature for our self-hosted Konfuzio users: System Messages. This innovative tool improves the...

    Read article