MQM Features

Last update: 2014 November 21

This document provides a high-level overview of the features of the Multidimensional Quality Metrics.


In order to meet the principles described in the Background and Principles document, the Multidimensional Quality Metrics (MQM) have the following features:

  • Open system. MQM is open and available at no charge for all users.
  • A flexible list of issue types. Drawing on the examples of leading QA systems and tools, MQM provides a comprehensive and extensible list of quality issue types that can be used in QA tasks. Because the definitions are open and publicly defined, anyone can use these issue types.
  • Separation of different aspects of translation. MQM supports analysis of source and target fluency to help users identify the source of problems. It separately evaluates translation accuracy and formatting issues. In addition, it supports assessment of “end user adequacy” (ratings of whether the translation meets its requirements). By separating these aspects, users gain insight into the source of issues and how to resolve them.
  • Multiple levels of granularity. Issues are categorized in a hierarchy, allowing users to use high-level issue types for quick assessments or fine-grained categories to support issue diagnosis and resolution.
  • Specifications. For cases where an existing metric is not suitable for a specific task, MQM uses 12 parameters, based on ASTM F2575, to aid users in selecting relevant issues for QA tasks and to decide how to check the relevant dimensions. By answering questions about these parameters, users will be guided to select appropriate metrics.
  • Support for legacy QA models. And for those wishing to use an existing metric, such as the LISA QA Model or SAE J2450, MQM supports older systems. Any system that uses categories that can be mapped to MQM issue types can be defined within the MQM system.
  • Local marking and audit trails. MQM can interact with the ITS 2.0 standard to provide a way to mark issues up in XML and HTML documents, thus providing a way to verify resolution and provide audit trails.
  • Extensibility. If users need to add additional, or more fine-grained, issue types, there is an easy mechanism to do so. In addition, the results of other QA tasks and models can easily be declared in MQM when they are needed.