Companies marching through the paces of detailed XBRL tagging don’t like to hear it, but we’re still just on the cusp of seeing real utility in the repository of data that is starting to build via XBRL. Investors have a long way to go to learn how to use the data to make investment decisions.

Regulators, on the other hand, have had a bigger head start, and the Securities and Exchange Commission is rolling up its sleeves to determine how it can use the data to better regulate capital markets. The SEC has said it is gradually getting more of its accountants up to speed on using XBRL to perform its routine filing reviews.

A recent academic study out of the University of Virginia demonstrates how XBRL data can be mined and managed in a way that makes it much easier to spot outlier elements in balance sheets and income statements. Those outliers can be indicative of any number of things – from investment opportunities to possible fraud, the authors conclude.

Randy Cogill, a professor in systems and information engineering at the University of Virginia, and his assistant, Steve Yang, applied graph similarity measurement to XBRL-rendered financial statements to measure structural differences in those financial statements. Graph similarity measurement has been used in applications such as text mining, pattern recognition, and computer vision, among others.

The authors selected a sample of 44 public companies in the retail and energy sectors for their analysis. They formulated the financial statement similarity measurement problem as a generalized graph similarity problem and constructed a graph similarity metric to measure the similarity of financial statements. They confirmed that the graph similarity metric is sensitive to structural changes in balance sheets, laying a foundation for discovering more important monetary movement patterns in financial statements, the authors say.

Yang said one of the benefits of XBRL data is that it lends itself to such modeling and measurement so that users of the information can extract any information they need. With so much information to digest, it’s the best way to glean meaningful insights. He envisions the findings could be useful for investors, who might build automated tools to help select financial information that is most important to them.

He also sees the benefit for regulators, who can use the tools to identify outliers that command extra attention. “It could be that they will find a deviation is because someone is doing something wrong," he said, whether intentionally or unintentionally. "They can extract very valuable information from this complex structure.