Powered by Zoomin Software. For more details please contactZoomin

How Classification Works

Exploring Rulebase Templates

  • Last Updated: May 13, 2026
  • 3 minute read
    • Semaphore
    • Documentation

After individual concepts are tuned in Knowledge Model Management (KMM) and classes are defined, Publisher looks at the Rulebase Templates to see how these terms should be weighted when creating the Rulebases.

The Rulebase Templates provide another opportunity to define and tune scoring. It is here that you can control issues that affect types and classes of concepts, relationships you’ve created between concepts, how parts of the documents score, and in what order the Publisher finds the words in a concept.

Using document structure for rules

For example, the Publisher can generate rules that look for concept labels in different parts of the document, such as the body, title, headline, abstract, and subject and keywords fields and weight each part differently. In the following diagram the title rule and body rules are weighted differently:

You also can set Preferred Labels to count more than Alternative Labels:

Rulebase fragments showing different scores for different kinds of labels

Combining Rulebase Template elements

You can combine different elements in the Rulebase Templates to make scoring even more precise. In this example, the Rulebase Template looks for a combination of a Preferred Label in the Title to find a score.

The “1”s in the diagram above show the Rulebase Template that tells the Publisher to generate a rule that looks in the title of the document for the Preferred Label and give it a score of 35 for a phrase match. The “2”s show that the Publisher should also generate a that looks for the Preferred Label in the title of the document as a sentence match. Note that it scores significantly lower – 15 instead of 35 – if the Preferred Label is found as a sentence match.

Scoring based on position within the document

Another way to refine scoring is have your Rulebase Templates look for content higher in a document. For example, you may want terms in the first paragraph of your document to score higher than terms found in the bottom of the document.

Classification Server can count terms found in the field start higher than those found elsewhere in documents

Above we see that the Rulebase Template tells the Publisher to generate a rule that looks for the Preferred label as:

  1. a phrase in the Body of the document,
  2. and give a score of 20,
  3. but score it higher if the term is in the start of the Body.

You can define field starts according to your needs. They might be:

  • The first sentence of a document.
  • The first paragraph of a document.
  • The first 100 words of a document.

We’ll make sure you develop the right strategy

Progress provides a number of out-of-the-box classification strategies that can be tuned to the individual characteristics of a client’s content. Templates generally take a few days to develop. Once set up and working, they tend to remain static. Working with Progress, clients develop a Rulebase template that determines the precise classification strategy for their organization.

TitleResults for “How to create a CRG?”Also Available inAlert