More than 2.2 million LinkedIn users have the word “expert” in their profiles. Their self-proclaimed expertise ranges from parenting to “humanitarian reform assessment” to “Las Vegas Nevada Short Sale and Foreclosure.”
But how much we say we know isn’t necessarily how much we actually know, according to a new startup, WhoKnows.
The company, founded by Alex Daly, former CEO and founder of major enterprise software provider ArcSight (now part of Hewlett-Packard Co.), and Toufi Saliba, a software developer and entrepreneur, recently raised $1.75 million in Series A funds and joined the accelerator program run by Citrix Systems Inc.
WhoKnows has devised software that tries to tease out an accurate profile of workers’ expertise and skillset by collecting and analyzing data about the websites employees visit, the questions they post on message boards, and how often others in the industry refer to their work. It’s beginning to sell the software to large corporations.
The software quickly builds a profile of an employee, and the profile improves over time, as the algorithm improves by using the concept of machine learning.
“We are going to see machine learning everywhere,” said Tim Connors, who led the investment in WhoKnows through his fund PivotNorth Capital. His other startup, search engine Blekko, also uses machine learning to improve its search algorithm over time.
WhoKnows claims that its software is 93% accurate. It tests the accuracy by going back to the people whose profiles it analyzes and asking them to confirm that they do indeed have the skills the algorithm determined they possess.
The end result, WhoKnows says, is that its software aggregates and maps out an entire organization’s skillset. A large corporation can use the profiles when the time comes for promotions and new hiring, for integrating a newly acquired company, and for layoffs and early retirement offers.
For employees, the benefit is the ability to quickly discover experts in other departments, making work more efficient. As they search for a term on Google, for example, the WhoKnows software automatically displays profiles of colleagues who might be knowledgeable about the topic. The algorithm drills down and displays granular knowledge that someone might not think to list on LinkedIn, but that may be useful to others researching a topic.
To use the system, workers don’t need to learn new software or launch a new app–the software plugs into existing open software used by employees.
Plus, employees who don’t often get credit for what they know can also shine with this system, said JT Sison, the company’s vice president of marketing and alliances. Employees approve what information, including skills and expertise level, get shown to the employer.
WhoKnows spent four and a half years honing in on the algorithm before launching the business. Those who try to cheat the system and make it seem as if they know more than they actually do are unlikely to succeed, the company says. It had more than 10,000 academics use it, test it and try to fool it.
“We gave them a lot of incentives to game the system,” said Mr. Saliba of the testers. “The more they gamed it, the more they earned.” WhoKnows, previously called Xperscore, took note of their methods and says it made the algorithm nearly foolproof.
To find its testers, the startup used an unusual method. It placed ads seeking out Ph.D students in a variety of fields who would be interested in playing with the program for a chance to win an iPod. It took about six months to find the first hundred subjects, said Mr. Saliba. Then, the company asked each of them to find another hundred.
“It took two weeks to get from 100 to 10,000,” he said. The company ended up giving away thousands of iPods, Mr. Saliba said.
For the article, visit WSJ here: http://blogs.wsj.com/venturecapital/2013/11/25/whoknows-might-know-more-about-your-workplace-expertise-than-you-do/