Military Friendly Employers Search Site Latest To Leverage Technology That Culls Meaning From The Crowd
With Veteran’s Day upon us, Victory Media has unveiled its new Military Friendly Employers search site, which provides a way for service members to search, sort, compare, and find a Top 100 Military Employer from nationally ranked companies. Some 16,000 facts – percentage of military that makes up new hires at a firm, support for part-time spouse employment during deployment, assurance of having the same position upon return, full salary for duration of Guard and Reserve duty, and more – feed these searches to point military personnel to the companies whose practices and policies are a fit for them.
Victory Media has been publishing the list since 2003, but this year searching through it is powered by WebKite’s content management platform. Military Friendly Employers is one of about a dozen sites using the company’s technology. It’s not semantic web search, but it is about adding contextual meaning to a data owner’s information to transform it into a vertical search engine site, which it calls a kite. The idea is that data owners can import their data into its management system, and work with its developer and integration team to customize it to their domain needs, including how they think users want to search the decision space. Users can then search, sort, and interact with the content through decision tools, top 10 lists, user reviews, and ratings.
That’s the start, but on any WebKite site, there’s a customize link that adds crowd-sourcing to the picture for search refinement. Fast, for example, doesn’t always equate to plain old quick, says founder Eric Silver. It’s important to get a clear, user-specific definition of the use case, “of what they mean language to mean,” he says.
Think of the automotive sector. “When I think of fast car, I think top speed, torque, track times, so we build a canonical understanding of qualitative search terms by people using sliders to say how important different facts are for a use case,” he says, and to automatically publish text on the site that puts this information in context.
“When you see a description of an item or read comparison text generated in an ad-lib style, it’s based on deciding whether an item is fit for a specific purpose.” On WebKite-powered Vehicle Picks, for example, when a user clicks “great gas mileage” the text in the item description updates to emphasize just that.
WebKite hosts a site’s vertical search engine, and the preferences and weights that people customize can be saved as lists for their specific purpose. “As those lists get saved the engine learns not just that specific saved use case, but also overall what people are looking for,” he says. “Our system is one where anyone who has data or access to it can put together a site where it starts learning from users how they make decisions and how they think about decisions.” Site traffic can go up significantly the more informed the visitors who save lists are about the site’s topic, he says.
Silver actually began WebKite with semantic web technologies in mind, but he says it didn’t work out for what the company wanted to do. “The vision that got me excited was the idea of a deep graph and an understanding of words that you could categorize and understand the information, and it just hasn’t worked,” he believes. The core search problem, he adds, “is that people are trying to make decisions and they don’t formulate the words right or express what is important to them…[and] there’s no way to have one set of relationships for everyone” to be able to respond to wide-ranging needs.
Another site powered by WebKite, he notes, is DrinkStreet, and there’s a big difference in a search of the best drinking games for two people vs. the best drinking games for couples. But, “in the semantic web these are very related [terms],” which creates a problem. Those looking into drinking games for two buddies may wind up being served a concoction of kissing games,
Is there a risk of giving up the idea of an interconnected web with this approach? Silver doesn’t think so. “We publish as a platform so, though we don’t do this yet, we can put together an API,” he says, such that a Webkite restaurant site that wants to show its most vegan-friendly dishes, for example, can connect with another WebKite site like Yummyplants to tie into its data and understanding about ingredients or healthfulness.
- Jinni Launches "Watch Together" Social Recommendations
- Facebook's Steps to Improve Graph Search
- Nara Offers Restaurant Recommendations Anywhere in the U.S.
- Twitter, TV, and Semantic Technology