Do you need to:

• Establish trust in your information?
• Ensure accountability for the producers and users of your information?
• Discover sources of error in your information?
• Understand the impact of sources on your information?
• Unlock the metadata stored in your information by multiple applications?

If so, PMAF™ is your solution!

Modern defense strategy and execution is increasingly net-centric and distributed, allowing more information to be made available in a timely manner. In this environment, the commander or warfighter must distinguish decision-quality information from potentially inaccurate, or even conflicting, pieces of information collected from multiple sources.

PMAF enables the publisher of information to record standard provenance metadata (pedigree), such as the source, manner of collection, and the chain of modification of that information. In addition, the publisher can define and include other related information relevant to quality assessment, such as domain-specific information about sensor accuracy or organizational structure of agencies. PMAF stores this potentially enormous amount of information in a volume-efficient manner and presents the information to the user in an intuitive graphical format, together with PMAF-generated assessments that allow the user to quickly estimate information quality.

At a Glance

• PMAF is a customizable framework for writing, retrieving and assessing information pedigree and other metadata that determines the quality of your information.
• PMAF works in concert with both new and existing applications.
• PMAF generates and stores the history of information objects, and queries that pedigree metadata to evaluate the level of trust that can be placed in the information.

What is Information Pedigree?

• Pedigree is metadata regarding the provenance of information: its source, manner of collection, and chain of modification
• Pedigree also refers to any metadata relevant to assessing information quality, such as domain- and source-specific metadata including qualifications of authors, spatial location, sensor accuracy, etc.
• Analogy: publishing
  • Papers, articles, books have bibliographies = provenance
  • Each document can be cited by other documents = impact

What Kinds of Information and Metadata?

• PMAF is agnostic to object types, whether structured or unstructured
• Document metadata: author, date of creation, IC-ISM, DDMS, etc.
• Web page metadata: web address, date of last modification, etc.
• Image metadata: resolution, time of capture, etc.
• Sensor metadata: accuracy, location, etc.
• Web service metadata: resources used, algorithm used, etc.
• Software agent metadata: creator, data format, etc.
• All: "source", "contains", "replaces", modifications, feedback, conflicts, etc.

PMAF Features

• General-purpose, extensible framework
• Writes and queries pedigree metadata in a volume-efficient distributed format
• Writes and queries pedigree metadata at a fine-grained level
• Supports customized pedigree-based assessments to establish quality and trust in information
• Supports domain-specific terminology
• Provides interactive browsing of pedigree through a graphical user interface
• Provides web service and Java APIs
• Operates in a variety of information management environments: open system

PMAF Assessments

• Impact/error propogation - what actions have been taken based on this information, and what other information was affected and may need correction?
• Timeliness - is this information outdated or superseded?
• Conflict - was this information processed from conflicting data?
• Feedback aggregation - what have other people said about the quality of this information?
• Corroboration analysis - how many unique sources?

PMAF Architecture

• One or more information repositories - each object has a unique identifier
• One or more client applications that read and write information objects and the corresponding pedigree data
• One or more pedigree repositories, which can be in different location(s) than the information repositories
• The PMAF system, which includes a server to manage the pedigree repositories and provides APIs for querying and publishing pedigree data
• The PMAF GUI can be used independently or embedded in the client application