Dynamic online learning for classification

If you enable the "Use dynamic classifiers during classification" option on the Advanced Online Learning Options window, this means that you can make use of the accumulated training documents to improve classification results without training your project. This is very useful if your project is trained at specific intervals. The accumulated training documents benefit your project between training intervals.

Important If you are using document separation, you cannot use dynamic classifiers. So, if you are using document separation, classification online learning can only collect documents automatically. This means that in order to improve classification, the training documents need to be imported and then your project retrained and published before the collected documents can improve classification.

Classification Online Learning Process

The dynamic classifiers perform classification using the documents that were collected and trained by the Knowledge Base Learning Server as well as the documents in the Classification Training Set. This ensures that the classification results are as accurate as possible.

Dynamic classifiers are created at the end of a batch to minimize performance issues, the "Number of documents required to repeat the training of the content classifiers" option is available to configure how often the classifiers are updated.

By default, this option is set to 1. This means that the training of the dynamic classifiers happens after each batch where at least one document is collected. If you set this option to 100, the dynamic classifiers are not updated until there are at least 100 documents. Subsequent batches benefit from those additional 100 training documents, while more are collected. Once an additional 100 documents are collected, the classifiers are updated again. The value for this option is configurable between 1 and 10,000.