Content area
Full Text
INTRODUCTION
The Financial Industry Business Ontology (FIBO) was born out of a requirement for common, shared meaning across data sources and message feeds in the financial services industry. This followed a number of industry initiatives in the creation of common message formats and common logical data models. Although these message and data model standards each had their uses, it became increasingly clear that the industry suffered from a problem of 'reconciliation hell' in data management, and that what was needed was some way of dealing with common semantics.
These challenges are set to become more pervasive in the future, both as a result of regulatory initiatives to combat systemic risk, and in terms of the architectures now available to financial firms and to regulators in the form of 'big data'. The FIBO standard, having been developed to address the difficult problems of less data, is well placed to address the bigger challenges in the emerging realm of 'Big Data'.
This article describes the development of FIBO as a standard for unambiguous shared meaning, the way in which meaning was dealt with in developing the standard, and the ways in which this may be deployed using the emerging architectures of the 'Semantic Web'.
We aim to show that FIBO is set to deliver two benefits in this context:
The provision of common semantics, so that data from diverse sources may be reconciled and compared.
The use of semantic technology applications to analyse data within large data sets in new ways.
DATA MANAGEMENT CHALLENGES
The FIBO was developed within the context of formal information systems management theory.
There are challenges to managing data in any context. Data quality, provenance of the information contained in data feeds and so on are well known and there are tools to deal with these. A less clearly understood question is the relationship between data and its meaning.
Considerations of meaning in data is important in many contexts - data feeds, integration, analysis and so on. Common language is needed wherever any one system interacts or interfaces with any other system.
In isolated systems, consideration of the meanings of data items (fields in databases, message elements in data feeds) can largely be ignored as these have been dealt with at the...