This paper illustrates the application of a developed global fund reporting ontology (GFRO) for efficient financial reporting. The GFRO extends Financial Industry Business Ontology (FIBO). Existing reporting financial information systems lack the ability to integrate data from heterogenic sources and provide unified and consistent financial reports that will comply with regulations. This study reveals that by integrating the power of XSLT and Semantic Web technologies, operationalised through the development of a scalable working prototype, allows financial services industry experts to build more flexible and consistent reports. Our research shows that the consistency of financial reports can be dramatically improved by using an appropriate inference engine.
Introduction
Data required for reporting to regulatory bodies is open to interpretation, from teams of
legal experts to data analysts and senior management, interpretation of legal documents
can leave experts arguing ad infinitum about minor nuances of language and
terminology. These challenges have spurred the need for the development of a
standardised language across financial instruments and institutions, a terms sets such that
there is little room for interpretation and the regulator receives transparent and
comparable data from all institutions for aggregation, and to be able to prove that the
report is consistent with regulations which are constantly changing and developing [
Towards addressing this shortcoming, we have developed GFRO and implemented a framework that uses GFRO to provide consistent and unified financial reporting across heterogenic data from different sources. Our research aims to advance research on ontologies by illustrating their application for improved reporting capabilities over a broad subset of financial instruments; specifically, bonds and equities, and to perform reasoning over source data to infer new observations.
Bonds are a debt investment instrument in which an investor loans money to an
entity, the entity borrows the money for a fixed period of time and repays moneys to the
investor as incremental interest payments with a bullet payment of the principle at a set
date. Many variations of bonds coupon and principle payments also exist. Equities are a
piece of ownership and generally control of an entity, generally referred to as shares or
stocks.
1. Main Ontology-based Features of the Framework for Financial Reporting
The most important task in adapting the FIBO [
We generate three types of rdf graphs. The first one covers point in time representations of the financial instruments i.e. month end data. The second transformation covers transactional data of financial instruments i.e. buying and selling of positions. The last one covers accrued interest to certain financial instruments such as bonds.
The second component of the implementation establishes communication between
the end user and the triple store. It uses parametrized SPARQL queries to deliver data
needed to generate financial reports, or conversely, return exceptions in case of
inconsistencies. On top of this, is the GUI layer, which implements a graphical user
interface that allows subject matter experts to create custom intelligent reports over
financial data including regulatory reports, automatically populating regulatory reporting
templates. The end user can produce reports across many funds or fund types over
features of that fund. We use the JavaFX [
Figure 2 illustrates a UML class diagram of the implementation of all layers. The central class is GenerateReport. Method getTransformation runs the XSLT transformation of financial data stored in a folder and generates appropriate RDF file, and exports that file into the triple store, also loading GFRO into the triple store. The system runs SPARQL queries necessary for reporting of financial instruments. Some of these queries are parametrized, and some of them are not. For example, parametrized queries for bond reporting are reusable for any type of bond. All queries are stored in one folder.
When an end user submits a request for generating a report, then method
runQueries() in Query class runs all queries over Stardog [
Clockwise from bottom left; we represent the equity holdings of the fund, by country, and next by sectoral allocation of the equities as pie charts. The bar charts are used to represent currency allocation of the equities within the fund and the physical cash holding of the fund. Data in the tables’ represent overall summary data. The representation of the data is standard in its presentation; however, the method of retrieval and querying is unique in its flexibility.
Databases find it difficult to query from a data end point and must be queried more
generally and data collated and validated [
Although we tested the process described in the paper over a large amount of complex and varied data representative of fund level data, a limitation of our research is that we did not test beyond the scope of bonds and equities. The next stage of our research is to explore the utility of our framework in a bid data setting.
This application of regulatory and risk reporting over a complex set of data and the
ability to automatically verify results over reasoning should allow further research into
the benefits of ontologies. The adoption of a shared ontology could potentially
dramatically reduce the timeframe for transaction processing. The financial ontology
standards (FIBO) are at early stages of development, these standards are conceptually
strong but lack the testing to allow for robust implementation. All findings were reported
to the EDM Council and Object Management Group to help to improve the FIBO
standard wherever possible [