SemanticWeb.com continues its coverage of semantic web-based offerings that help enterprises leverage the news for better business intelligence. Today, we look at Silobreaker.com. (The series begins here.)

Silobreaker.com

Silobreaker has two goals: To be the leading destination site for general user news searches and to become an equally important resource for business, using its public web site as a showcase of and marketing tool for its enterprise offerings. A confirmed information junkie, CEO Kristofer Mansson brings an understanding of how important it is for users to get the information they need fast, and the site aims to enable them to get from one system the answers they require across structured and non-structured information types, as well as make better sense of that data.

“We feel two things are needed with search—to change the user experience and move away from keyword queries generating long lists of search hits,” he says. “And we believe the graphic representation of search has to come into the text arena as it always has been with numerical data.” To get there, Silobreaker saw the need to move away behind the scenes from traditional keyword search indexing to more intelligent analysis of the content set. “We felt that even though many regard search and content analytics as two different things, they go well hand in hand and must converge to present something smarter than search hits in the user interface,” Mansson says.

Semantic technologies and statistical algorithms combine to enable Silobreaker to achieve its ends. An entity-centric approach—an understanding of how people, companies, organizations, topics, and so on—relate to each other and are disambiguated from each other, and the ability to extract entities to connect unstructured with structured data is of course important to Silobreaker. Its entity database includes 50,000 fact sheets on people, companies, and the like, with millions of identified entities. But, says Mannson, “we focus on the whole rather than different technology parts.” He sees a fragmented tecnology landscape where small companies are working away on their piece of the puzzle, such as extracting algorithms or clustering. What the enterprise especially needs, however, is a single system for dealing with the entire workflow around media-monitoring; from back-end content aggregation, intelligent indexing, classification and storage to front-end search, analysis, quality assurance, user collaboration and decision support. “For customers the cost of recreating those parts by relying on multiple vendors was just very expensive,” he says. “We think the glue and the technology behind the glue are equally important.”

Both the enterprise and governments are spending big money on media monitoring of brands, competitors, and so on, usually with a mix of internal resources and software investments and external services such as clipping agencies, Mannson notes. The company’s enterprise software suite provides a local Silobreaker behind the customer firewall that aggregates external and internal content, such as Factiva feeds and Microsoft Office documents. Companies can apply their own taxonomies to the solution. It analyses the content, visualizes it and delivers administrative and collaborative tools around it.

Among the visualization tools to accompany its more traditional search results views are network searches, hot spot searches, and trends searches. Hot spots provides geomapping of user events by extracting location data around topics – for instance, you can get a quick overview of where in the world are the big news events that might affect your company’s strategy. It might be helpful to know, for instance, if your product is being written about extensively in a certain region, so that the company can ensure that availability is strong there. Trends searches try to give users a quick idea of what the aggregate press corps is writing about at any point in time, so you can make comparisons – for example, when it comes to brand intelligence, is your company’s product being written about more or less frequently than a competitor’s at any point in time and over time.

Network search helps users find structure to unstructured content.

Network search finds and connects data in search results on the fly and displays the connections in an interactive network; by creating structure from unstructured data, it helps users find relationships within media flows that extend discovery beyond their original search focus. Technically, the idea is to find structure in unstructured information without relying on manually configured connections or human editing. “So when you make a query there’s a live retrieval of what the network will look like at that time, changing dynamically as new content comes in,” he says. “The closest relationships might be obvious but peripheral ones might be those you didn’t realize or remember, so it’s really a method of getting you to the nitty gritty of what you’re looking for or more importantly to maybe discover new things without using a search box.”

Its Enterprise Software Suite will soon be joined by an Enterprise Online Service, a hosted customizable service for handling work flows around their monitoring and analysis needs. It also offers embeddable widgits and an API that lets users, bloggers, partners, publishers and site owners extend their own services with Silobreaker's content and functionality. The future for Silobreaker on the consumer side will remain free, with the potential for some premium content at a price. It’s focused on serious news around business, politics, and global issues, with ancillary sites on sports and entertainment.

It also wants to deliver a way to put users in complete control of what Silobreaker can tackle, throwing in data from their hard disk or anywhere on the web for indexing to share with colleagues or for setting up social networks around an entity sphere.