Knowledge Extraction

Knowledge extraction quickly categorizes and tags content. Save time, reduce errors and enhance value by making information easily findable and discoverable.


Can Kairntech’s AI make life easier for information providers?

Extract information with superior accuracy

automated-text-labeling-entity-linking
  • Quickly create datasets & train high-accuracy models
  • Extract, disambiguate and link entities to Knowledge Bases
  • Contextualize data with Wikidata and business vocabularies

Classify documents efficiently

  • Quickly build classifiers including IPTC Media Topics
  • Patented semantic fingerprint technology on Wikidata
  • Use a single AI model for multiple languages
automated-text-labeling-iptc-classification

Embed and industrialize at scale

training-ai-models-and-quality-assessment
  • Deploy on premise and embed with a rich REST API
  • Continuous improvement with human-in-the-loop
  • Scale seamlessly to millions of documents

Want to learn more?

How does it work ?

1

Create entity detection dataset: upload documents, pre-annotate with 0/few-shot learning, validate manually, enrich by using active learning.

2

Create text classification dataset: upload documents, pre-annotate with manually, enrich by using active learning.


3

The AI learns from the annotations. Train models, combine with technical components into AI pipelines. Deploy, embed and industrialize at scale.

100% secured, 100% transparent

GDPR

All our data storage systems take into account the constraints of the GDPR.

Access

Manage fine-grained access rights to facilitate access to multiple stakeholders.

Hosting

In the cloud or on-premise, choose the mode that best suits your organization.

They trust us

  • ” At AFP, we are constantly striving to improve our editorial processes. Thanks to Kairntech named entities recognition and categorization technologies, we can offer our journalists, but also our customers, more powerful and relevant search and navigation tools. “
    Louis-Cyrille Trebuchet
    Chief Technology and Information Officer

  • ” To make the most of our content, it is essential to build high-quality AI models simply and quickly and to combine them with knowledge base. “
    Gaëlle Recource
    Head of Content

Want to learn more ?

We are here to help

Industries like information provider and publishers (content enrichment), customer support (ticket categorization), e-commerce (product review analysis), healthcare (patient record classification), finance (fraud detection in transactions) and homeland security (event & threat detection) benefit significantly from Knowledge Extraction.

Yes, we can integrate business vocabularies into the solution, tightly synced with Wikidata for up-to-date knowledge. We’ll also suggest new candidates (for your validation) to continuously enrich the vocabularies.

We implement pipeline that combines AI models, Entity Linking & Disambiguation leveraging advanced algorithms that allows us to reach the state of the art in this domain.

We use various metrics like fmeasure (precision, recall), ROC-AUC… through interactive dashboards. Confusion matrix are still missing but we’re on it!

Once the system is live, we gather user feedback to:
(i) expand the commercial vocabulary and synchronize the Knowledge Extraction pipeline with new entities in real time, and
(ii) incorporate new texts into the training dataset to retrain the model and update the pipeline.
These operations are performed using Kairntech Studio.