Firm uses ‘causality’ to improve Internet searches

An Indianapolis firm has developed software designed to sift through massive troves of professional and scientific journal articles more efficiently than traditional Internet search engines.

Allegient LLC and subcontracted IUPUI informatics experts wrote algorithms that go beyond word searches to look for “causality”—relationships between words suggesting one thing caused another.

“It’s a significant improvement in how quickly a researcher can get through material,” said Eric Tinsley, vice president of delivery for the Carmel-based company. “We can advance the cause of knowledge-gathering.”

Allegient is trying to get a foothold in the fast-growing corner of “big data” called text mining, the practice of trolling through text written in plain language in order to find patterns and ultimately important nuggets of information.

The market is growing quickly. Market research firm Frost & Sullivan Inc. estimates annual global sales of $400 million and reaching $1 billion in five years, driven by companies’ rising appetite for searching social media.

Mukul Krishna, senior director for digital media at the Mountain View, Calif.-based firm, said companies in most industries demand greater efficiency with search.

Only about 30 established companies operate in text mining, he said, creating favorable odds for companies that can show a unique value proposition within the next year or two.

“It’s a very sexy place to be in right now,” Krishna said. “It’s attracting a whole lot of people who want a piece of that pie.”

Allegient took over a project launched by Castleton-based My Health Care Manager Inc. Funded by a $1.3 million federal Small Business Innovation and Research grant, the project—Semantic Understanding Research in the Automatic Acquisition of Knowledge—was intended to find a way for hospitals and companies to learn how to better care for the elderly.

However, the startup, which had attracted funding from Indiana University Health venture capital arm CHV Capital, went under in summer 2011.

Tinsley, who was principal investigator at My Health Care Manager, took the project to Allegient, a software and computer consulting firm, and finished in March this year.

Even the most sophisticated researchers in settings ranging from universities to private companies are reduced to pecking words into search engines and then winnowing thousands of articles to find useful information.

Allegient’s software searches the same articles, but with the twist of attempting to think like the researcher.

For example, a hospital or nursing home operator might want to know more about how drinking too much alcohol causes greater risks of falling. The software searches full sentences looking for structures suggesting that one thing leads to another, such as “it” followed by “has propensity toward.”

Users then see a set of articles ranked by relevance.

Mathew Palakal, executive associate dean of IUPUI’s School of Informatics, predicted the software eventually will be refined to a simplicity that will allow even elementary schoolchildren to use it.

“It is very, very unique,” Palakal said. “It has to be properly marketed.”

Indeed, Allegient is still figuring out how to sell it.

For now, the company is letting a few academicians use it free. A couple of other projects are generating small fees from the software.

The original plan of targeting geriatric businesses, such as nursing homes, is still moving forward, Tinsley said.

But Allegient also thinks engineers, paralegals and attorneys are likely candidates for the software, and plans to move the software to the cloud and charge a “seat fee” based on usage. Those fees are likely to range from $10,000 to $250,000.

One hurdle to overcome is access to thousands of publications. Allegient is getting by with Pubnet, a government-operated database of psychological and medical journals that allows analyzing abstracts without paying a fee.

Go beyond the abstract, though, and fees kick in—something Allegient must negotiate.•