Introduction

Introduction

The Progress Classification and Language Service (CLS) identifies topics within documents by analyzing the content against a series of rules and applying a metadata tag denoting classification. In this document, we will talk about several components of Semaphore:

• Knowledge Model Management (KMM) – Where you store and work on your model (taxonomy, ontology or whatever)
• Publisher (PUB) – Creates Rulebases from taxonomy and Rulebase Templates
• Document Analyser (DA) – Tests all of the Rulebases generated for the ontology against sample or reference documents to see how they behave
• Classification Server Test Interface (CSTI) - Where you can look at individual documents to see how specific terms score
• Classification & Language Services Client (CLS Client) and the Precision & Recall Tool - allows you test batches of documents and generate statistics, make comparisons between Classification rules, and perform bulk text-mining and entity extraction to aid in model development

CLS uses templates to help make classification fast, accurate and easy. These templates are highly configurable to meet each organization’s specific classification needs while allowing experts to focus on building a taxonomy, or model, that best reflects the content.

CLS uses the ontology and templates to automatically generate rules, which are then used to classify content. CLS generates up to 25 types of rules, depending on an organization’s needs.

As part of the classification process, documents receive scores for each topic. The score reflects how confident the software is of the classification. A variety of factors combine to achieve the score, and each of those factors is weighted. Both the taxonomy and the templates can affect weightings.

This document describes some of the specifics of this process and aims to help you understand how you can have the taxonomy and Rulebase Templates work together to give you the best classification results.

Audience

This document is intended for anyone that is interested in how automatic classification works in Semaphore and those that are interested in making this process work better.

Additional documentation available

The following documentation provides additional information that may be helpful to the reader:

• Semaphore Classification Server Rulebase Reference
• Classification and Language Service Developer’s Guide