High-performance content classification driven by machine learning
Automatically add tags to all your documents based on their full content,
not just the metadata

Why classification matters


Content agility

Create consistency: tags learned from one source are automatically applied to others.

Transparently file documents in the right location in your ECM or archival system to locate them more easily.


Information discovery

Boost accessibility: tags are as many filters that enable users to search for relevant content.

Find documents contextualized to the user’s profile and generate customized recommendations.


Customer intimacy

Analyze support tickets, emails, forum posts and more to detect topics and react appropriately.

Route inbound requests based on subjects and urgency.

How it works

Classifier uses artificial intelligence and machine learning to automatically detect the characteristics of documents. Each tag is associated with a unique signature that is subsequently used to select which tags to apply to new documents. Ongoing quality control provides a feedback loop that adjusts incorrect tagging, increasing precision over time.

Training phase

In the learning phase, Classifier creates its semantic signature repository. By providing a subset of already tagged documents, you can greatly accelerate this phase. If such a set does not exist, a web-based workbench guides you through the initial tagging. Antidot’s Active Learning technology drastically reduces the effort by guiding you in the selection of a representative sample of documents.


Classification phase

Once trained, you can inject into Classifier your full document corpus for initial, batch tagging. Then, on an ongoing basis, every new document is processed in real-time. Tagging of each document takes only a few milliseconds, making it easy to integrate classification into any business
or IT process.

Feedback and quality control

Classifier’s self-evaluation capability prevents over-fitting and under-training bias. At any time, you can visualize a report on the quality of the training, with metrics for each tag. Ongoing quality control lets you iterate to improve the precision of the tagging, with the modified signature repository being made available in real-time to the industrial classification process.


Efficiency and precision

Implementing Classifier is fast, easy, and requires no technical expertise. Entirely UI based, the product drives you through the implementation process and ensures a short time-to-value. Its use of the most advanced machine learning algorithms guarantees a precise outcome, even when only a limited reference corpus is available.



Classifier applies any number of tags to each document, regardless of how your classification is structured (lists or trees) and no matter how deep or wide it is.



Classifier is language agnostic. It seamlessly handles your multilingual corpus by automatically detecting the language of each document.


Easy to integrate

Through RESTful APIs, Classifier easily integrates with any web application or software package. It handles documents in batch mode or as a flow, tagging a document in just milliseconds.