Will Loud3R.com Speak to Your Interests?

By SemanticWeb.com on July 9, 2008 11:35 AM

Jennifer Zaino
SemanticWeb.com Contributor

When Microsoft announced last week its plans to buy natural-language semantic search vendor Powerset for an undisclosed sum, the software giant said it wanted Powerset’s technology to complement its own research and take search to the next level.

As it happens, Lowell Goss thinks his company has already moved in that direction. The founder and CEO of Loud3R.com, which merges human editorial and user input and machine algorithms to find and publish the best content across any given subject matter across web, says that in the first four weeks of operation, 150,000 unique users have visited the network.

Loud3R.com is a semantic publishing platform, albeit not a Semantic Web application.
“Semantic just means you add meaning to the content you find. We take unstructured content across the web, add metadata to it and structure to that content. That is how we solve that particular problem,” he says.

To be specific, Loud3R works like this: Editors, hired because they are subject matter experts and enthusiasts in their own right, man the frontlines, handling a setup process that involves identifying hundreds or thousands of sites they think are the most interesting and most relevant for a particular topic. Currently, those topics number 25, ranging from venture capital to video games to cars, dogs and gossip, with ten more launching this month.

“All the sites we published so far are about passions or enthusiasms, but we can do just about anything,” says Goss.

Semantic Glossary

An editor of a particular topic trolls through blogs, news sites, personal web sites, video sites, and so on, for the best on the web, as well as interact with bloggers and other publications in a particular space to get their recommendations and feedback. They then assemble a semantic glossary description of the content’s concepts, consisting of key words, synonyms, antonyms, spelling variations, “all you need to understand the relationships between concepts and describe those concepts themselves,” says Goss.

That information is fed into the proprietary publishing platform, and then the content discovery engine takes over, leveraging the human intelligence and the semantic information to constantly find and rank the best content, refreshing it every few minutes.

Every couple of weeks the editor may add a new tweak, perhaps, for example, adding a new band to the music site so the system has additional topics for which it can effectively screen. Users can also contribute sources, articles, web sites, and so on, and the discovery engine screens and filters this content to make sure it is relevant to topic at hand.

The content for each site is tuned in a way to make it appropriate to a particular topic, Goss says. For example, when it comes to content about sailing, the focus is on community and source quality, whereas content about politics focuses on user interactions, freshness and buzziness.

“So we vary how we look at a topic for editorial tone, even though the editorial decisions are made by computers, not people,” he says.

Goss, who in his past life worked at companies, including Yahoo, as the leader for product and user experience groups, says the idea for the site came out of his own personal frustrations. As a self-described motorcycle nut, he says, “I found that I had this dual frustration – my RSS box was filling up with lots of dupes and junk and it was just another inbox to keep up with, and at the same time I knew I wasn’t discovering the best stuff across the web.

“So there was this paradox – I had information overload and also was missing out. Our intent is to try to help consumers for any topic to find lots of great stuff and discover the best stuff from across the entire Internet, while making that manageable,” he says.

He sees this as a far better approach to keeping up with the topic you are particularly enthusiastic about, compared to the blank slate of a standard search page. “We give them a place to start for that each and every day,” he says.

Context to Content

Another goal was also to enable users to follow their passions in a spam-free environment – or at least pretty darn close to one.

“Our technology does more than just find and publish information,” Goss says, “but it also filters that information so we eliminate 90-plus percent of spam, splogs, and junk to bring you the best information.”

Goss says that parsing out the semantics behind this is definitely a challenge. Some of its approach is automated, via a natural language clustering system that looks at the text of every item that comes through and parses it in a number of different ways to recognize not just individual terms, but also particular phrases and rearrangements of those phrases so it can tease out items that belong together and create context rules around that. Editorial intervention also supplements this, with rules sets surrounding every term in the glossary to understand, for example, that Bob Smith within the context of politics is a person in Congress and not your local plumber.

“A lot of that is about constructing knowledge and information around each term,” Goss says. “In some ways this is a simulation of what an experienced researcher would do without requiring the user to know how to do that.”

The service is free to consumers, but Goss is all ears to hear from companies who would pay Loud3R money to service them. For example, he says, Loud3R has seen interest from media companies, publishers and others interested in how they might use its technology to aggregate and add context to their own sites.

“We’ve solved a pretty difficult aggregation and semantic meaning problem here, and others can take advantage of that,” he says.

Comments