The following is a special contribution to this blog by Louiqa Raschid, a professor in the School of Business, Institute for Advanced Computer Studies, Department of Computer Science, and Center for Bioinformatics and Computational Biology at the University of Maryland, and H.V. Jagadish, Bernard A. Galler Collegiate Professor of Electrical Engineering and Computer Science at the University of Michigan. Louiqa and Jag co-organized a workshop on next-generational financial cyberinfrastructure on July 18-19.

Earlier this month, experts in computer science as well as finance gathered outside Washington, DC, to consider the need for a new financial cyberinfrastructure, and to elucidate the computing research challenges that are arising in this increasingly interdisciplinary space. Participants were drawn from academia, industry (both tech vendors and financial institutions), and government (regulators) at this Next-Generation Financial Cyberinfrastructure Workshop sponsored by the National Science Foundation’s (NSF) Directorate for Computer and Information Science and Engineering (CISE). An associated planning meeting was funded by the Computing Community Consortium (CCC).

Our challenge arose from the strong interconnections in financial markets in recent years, with interdependence between financial institutions. Traditional financial models are good at dealing with variance as well as with risk, but they typically have strong independence assumptions built in. Yet the scenario of greatest concern is when one financial institution collapses, and many others collapse due to their interrelatedness. Such “systemic risk” is what regulators most want to protect consumers against, but the tools of the past may not be adequate for the task into the future.

The Dodd-Frank Wall Street Reform and Consumer Protection Act created an Office of Financial Research (OFR) within the U.S. Department of Treasury and charged it with collecting (proprietary) data from financial institutions necessary to evaluate and manage systemic risk. The OFR, in conjunction with other regulatory agencies, has begun to take steps in this direction, thereby creating an opportunity to use technology for more effective financial modeling and regulation.

As the workshop participants discussed, the technical challenges that arise in the financial sector push the envelope in many areas of computing, including semantic technologies, data management, information extraction from natural (legal and financial) language text, network analysis, social media analysis, and intelligent agents, to name a few. Some examples:

• A legal entity identifier (LEI) to uniquely identify organizations that can engage in financial interactions is in the process of being adopted as an international standard. This provides a unique opportunity to create networks of organizations, their subsidiaries, partners, etc. One can then overlay financial contracts, streams of payments and fund transfers, etc. Simulation and analysis of these extensive networks, at the granularity of the individual organization and financial contract, will provide economists and researchers with an unprecedented opportunity to understand the behavior and evolution of financial markets and ecosystems.

• An area of potential importance is the use of semantic technologies and ontologies. The Financial Industry Business Ontology (FIBO) has been developed by the Enterprise Data Management Council, an industry group. FIBO is an extensive ontology as well as a source of reference terminology. Efforts are underway to exploit FIBO to create taxonomies, and possibly multiple poly-hierarchies, to represent financial instruments. The hope is machine learning and rule-based formalisms will provide methods to automatically classify a financial contract, or to reason about a Living Will document for a financial entity. At present such contracts are opaque and the unraveling and resolution of financial contracts typically occurs in bankruptcy court.

• Human language technologies including entity extraction and entity resolution from text documents have reached a high level of maturity in the last decade. In contrast, most financial reporting to regulators occurs in the form of paper or PDF files, or using XML-based technologies. This presents many opportunities to create rich document repositories as well as knowledge bases of the extracted entities and their relationships. Further research will be needed to validate the quality of the extracted knowledge.

We encourage you to stay tuned for a report from this workshop, to be published in approximately a month’s time, delineating additional important research problems that we hope the computing research community will be inspired to address.