forward look, inc.'s research focuses on the management of transactions and their underlying data. Our goal is to address Michael Stonebraker's challenge of 'integration on demand', and supplant the prevailing but more brittle approach of 'integration in advance'. By tracking how data is assembled by various counterparties to structure the transactions that flow between them, we determine what dynamically enables data exchanges in support of STP. We focus on how differing terminologies (eg, variants of XML or in-house data dictionaries) and varied application contexts (eg, front vs back office processes) result in syntactic and semantic variances, and more importantly, how these differences can be programatically bridged by spanning the implicit ontological gaps.

We continually discover interesting facets about transactions and data that act as catalysts for our client's implementation strategies. For example, we observed how reference data depth (eg, the number of codified data values for "Settlement Location") exhibits significant correlations with the costs of trade fails - but not necessarily with STP rates. On the other hand, low STP rates exhibit strong associations with reference data breadth (eg, the number of codified data categories such as traded market, counterparty identifiers, security types, and so on).

We actively explore how evolving schema and data mapping techniques can be used to automatically harmonize information contexts. We regularly examine how semantics (data meanings and usage) and ontologies (contextual maps between related terms across the industry) capture the industry's processes and workflows, with particular attention to the differing information requirements of various trading counterparties across the life cycles of cross-border transactions.

An overview of our approaches and results can be found in our 2003 Research Note (and accompanying Diagram) presented at the ACM Electronic Commerce Conference in San Diego CA (US). Additional outcomes were published in our 2004 Industrial Track paper presented at the IEEE Conference on Data Engineering in Boston MA (US).