Sure, you can spot accounting fraud by digging through hundreds of pages of your company’s financial documents. But who’s got the time?

Supported by a grant from PricewaterhouseCoopers, Virginia Tech professors Patrick Fan and Greg Jenkins are coming up with an automated way to detect corporate fraud.

The catch: They’re sniffing it out using a company’s publications — press releases, letters to shareholders, transcripts of analyst calls, etc. — instead of its financial data.

The reasoning behind it? Looking for fraud based purely on a company’s financial performance can be like searching for the proverbial needle in a haystack — it’s labor intensive, and odds are good you’ll come out of it with sore hands, a headache and nothing else.

But narrowing down the search based on a company’s language, according to the researchers, yields much better results. Their new software will look for patterns and trends in a company’s word usage that don’t mesh up with the same type of data taken from other companies in similar industries.

Those kinds of inconsistencies usually mean something’s fishy with a company’s books.

Apparently, numbers can be misleading, but words always tell the truth — the prototype software is already up to 80% accurate in detecting fraud, and the team still has until summer 2009 to improve on it.

Tags: accounting fraud, plagiarism, Virginia tech