Consumers of financial information come in many guises from personal investors looking for that value for money share, to government regulators investigating corporate fraud, to business executives seeking competitive advantage over their competition. While the particular analysis performed by each of these information consumers will vary, they all have to deal with the explosion of information available from multiple sources including, SEC filings, corporate press releases, market press coverage, and expert commentary. Recent economic events have begun to bring sharp focus on the activities and actions of financial markets, institutions and not least regulatory authorities. Calls for enhanced scrutiny will bring increased regulation and information transparency While extracting information from individual filings is relatively easy to perform when a machine readable format is utilized (for example, using XBRL, the eXtensible Business Reporting Language), cross comparison of extracted financial information can be problematic as descriptions and accounting terms vary across companies and jurisdictions. Across multiple sources the problem becomes the classical data integration problem where a common data abstraction is necessary before functional data use can begin. Within this paper we discuss the challenges in converging financial data from multiple sources. We concentrate on integrating data from multiple sources in terms of the abstraction, linking, and consolidation activities needed to consolidate data before more sophisticated analysis algorithms can examine the data for the objectives of particular information consumers (for e.g. competitive analysis, regulatory compliance, or investor analysis). We base our discussion on several years researching and deploying data integration systems in both the web and enterprise environments.
E. Curry, A. Harth, and S. O’Riain, “Challenges Ahead for Converging Financial Data,” in Proceedings of the XBRL/W3C Workshop on Improving Access to Financial Data on the Web, 2009.
2. Agenda
Digital Enterprise Research Institute www.deri.ie
n Motivation - Financial Data Ecosystem
¨ Data Providers
¨ Data Formats
¨ Data Consumers
n Converging Financial Data from Multiple Sources
¨ Entity Centric Approach
¨ Architecture
¨ Identity Mismatch
¨ Data Query
n Data Integration Challenges
n Recommendations
4. Financial Information Providers
Digital Enterprise Research Institute www.deri.ie
n Individuals: e.g. CEOs reporting equity sale
n Companies: e.g. 10-K filing
n NGOs: e.g. sector-wide lobbying groups
n Government: e.g. regulators, central banks, statistics
offices
n Worldwide organisations: UN, OECD
n Academics: various economists, public policy
n ...
n Publicly available datasets, purchased datasets or in-
house sources
5. Various Data Formats
Digital Enterprise Research Institute www.deri.ie
n Unstructed Text
¨ News articles, press releases, raw transcripts of investor calls
n Hypertext
¨ Coporate websites, goverment websites, ...
n Spreadsheets, et al.
¨ CSV files, word docs, pdf, powerpoint, ...
n Strucutred Data
¨ XML, XBRL, CSV, SDMX, ...
n Graph Structured Data in RDF
¨ DBPedia, CrunchBase, RSS-CB, ...
6. Financial Information Consumers
Digital Enterprise Research Institute www.deri.ie
n Competitive Analysis
¨ Mash-up of financial figures and analyst commentary for
decision support
n Regulatory Compliance
¨ Forensic Economics
¨ Spotting patterns or conditions that support fraud or money
laundering
n Investment Analysis
¨ Individual/Institutional investors
¨ Transparent fund comparisons
¨ Evaluate potential fund return
7. Goal
Digital Enterprise Research Institute www.deri.ie
n Integrate data for:
¨ Central access
¨ Cross document analysis
n Our group works in data integration and have applied
our approach to pilots in the financial services
industry
n Report on experiences and lessons learned
8. Converging Financial Data from Multiple
Sources
Digital Enterprise Research Institute www.deri.ie
n Provide common data platform for search, browsing,
analysis, and interactive visualisations across sources
n Entity centric approach
¨ Single data view allowing information filtering and cross
analysis
¨ Consolidate data into coherent graph 'mashed up' from
potentially thousands of sources
n Key challenge is semantic integration of structured
and unstructured data from the open Web and internal
corporate data sources
9. Converging Financial Data from Multiple
Sources
Digital Enterprise Research Institute www.deri.ie
n Large graph of RDF entities
n Entities typed according to what they describe
¨ People, locations, organizations, publications as well as
documents
¨ Inter-relations and structured descriptions of entities
n Entities have specified relations to other entities
¨ People can work for companies, people know other people,
people author documents, organisations are based in
locations, and so on
10. Data Integration Approach
Digital Enterprise Research Institute www.deri.ie
n Lifting data sources to common format, in our case
RDF (Resource Description Format)
n Integrating the disparate datasets into a holistic
dataset by aligning entities and concepts
n Run domain/task specific analysis algorithms on
integrated data
n Interactive browsing and exploration of integrated
data or results of algorithmic analysis
13. Identity Mismatch
Digital Enterprise Research Institute www.deri.ie
n Need to connect sources that may describe the same
data on a particular entity
n Case studies analyzing connections between people
and organizations
¨ SEC filings (Form 4) identified 69K people connected to 80K
organizations
¨ Same analysis on database describing companies produced
122K people connected to 140K organizations
¨ Data needed to be enrich and interlinked using entity
consolidation (a.k.a. object consolidation) to avoid having the
knowledge split over numerous instances
¨ Ontology-based disambiguation
14. Data Query
Digital Enterprise Research Institute www.deri.ie
n SPARQL, the semantic query language allows queries/
questions to be asked:
¨ What do the companies ‘Microsoft’ and ‘IBM’ have in common?
¨ What competitors of ‘HP’ are in ‘Arlington’?
¨ What’s the relationship between ‘Microsoft’ and ‘IBM’?
15. Data Integration Challenges
Digital Enterprise Research Institute www.deri.ie
n Text/Data Mismatch
¨ Human language often ambiguous
¨ Same company might be referred to in several variations
(e.g. IBM, International Business Machines, Big Blue)
¨ Ambiguity makes cross-linking with structured data difficult
n Object Identity and Separate Schema
¨ Sources differ in how they state the same fact
¨ Differences on level of individual objects and schema
¨ SEC Central Index Key (CIK) to identify people (CEOs, CFOs),
companies, and financial instruments
¨ DBpedia use URIs to identify same entities
¨ Methods have to be in place for reconciling different
representations of objects and schema
16. Data Integration Challenges
Digital Enterprise Research Institute www.deri.ie
n Abstraction Levels (Data Context)
¨ Financial data sources provide data at incompatible levels of
abstraction
¨ Classify data in taxonomies pertinent to a certain sector
¨ Differences in legislation on book-keeping (e.g. Indicators
from Euro regulators may not be directly comparable with
indicators from US-based regulators)
¨ Differences in geographic aggregation (e.g. region data from
one source and country-level data from another, IBM Ireland
Ltd, IBM Europe, IBM Global,…)
17. Data Integration Challenges
Digital Enterprise Research Institute www.deri.ie
n Data Quality
¨ General challenge integrating data from multiple sources
¨ Errors in signage, amounts, labelling, and classification can
seriously impede utility of systems operating on such data
¨ Combining erroneous data aggravates the problem
¨ Within open environment data aggregator has little or no
influence on the data publisher
¨ Challenge for data publishers/consumers to coordinate to fix
problems in data or blacklist sites providing unreliable data
18. Recommendations
Digital Enterprise Research Institute www.deri.ie
n Agree approach to the specification and use of
common identifiers or at least their mappings
n Adhering to common publishing method reduces
integration effort and facilitates data reuse
¨ Linked Data principles
n Convergence between data providers requires
coordination and time
¨ No need for “Big Bang” integration
¨ Follow a pay-as-you-go iterative approach to integration
19. Digital Enterprise Research Institute www.deri.ie
Thank you for listening
E. Curry, A. Harth, and S. O’Riain, “Challenges Ahead for Converging
Financial Data,” in Proceedings of the XBRL/W3C Workshop on
Improving Access to Financial Data on the Web, 2009.