There's always more room for search innovation. In this interview, Bambi Francisco interviews Tim Musgrove, founder and CEO of Digger. The San Jose, California-based startup is developing advanced semantic solutions for the Web, including hosted semantic search, automated content tagging and topic generation, and optimized keyword generation. In this interview, Musgrove and Francisco talk about the launch of Digger's semantic search destination site. Anyone who'd like to test it out here can do so by answersing "vator" in the comments field for the question "How did you find out about Digger."
(Here is an edited version of the interview)
BF: Congratulations on your $4 million dollar raise that you closed in December. It's tough times to raise funds.
TM: It was not easy and we're really happy to have it done.
BF: So clearly investors are interested in what you're doing to solve a problem and help people search and find products. So just to open this up, what is the problem with the way people find stuff on the Web?
TM: I think what's wrong with the way people find stuff on the Web is that we're still locked in the exact keyword-match model, where we were several years ago. Until we can match the concept, or the meaning of text, even when it is expressed in different words, we're not really going to make a big leap.
BF: We started with semantics. Ask Jeeves was all about it. But it really didn't take off. So why is it going to work today?
TM: The problem with the original Ask Jeeves is that it wasn't really scalable. At one point they had a big sweatshop in Berkeley of people hand-coding templates. Unless you entered a query the same way they had already thought of, they couldn't do anything about new queries.
BF: There was also another semantic search company called Powerset and its problem was it didn't index enough pages. They sold for a $100-plus million dollars so they've done a great job in exiting. How are you different from them?
TM: We're different because we completely avoid the indexing problem because we can work with anyone's existing keyword index. We look at what your query means and we can send it into Google or Yahoo and get results. But we really focus on the semantics of what the words means. Powerset focused on the grammar. So, Powerset believed that the grammar of the query was important. They believed people should type in an entire sentence, such as "Where can I park near Market Street." We don't think people want to ask questions in a sentence.
BF: That's because people are used to asking questions using keywords.
TM: Right. Everyone is used to keywords and Powerset said that was not natural and Google made us use keywords against our will. We say that people have been using post-it notes for decades, where you just put in a keyword.
BF: Well in today's generation people are using short hand, like ttyl.
TM: That's right. If you look at the little fragment phrases on Twitter. Short hand comes natural to us. So we think that the query is going to be a query and not normal English. But the semantics of what the word means, say if someone types cab, c-a-b, do they mean Cabernet, the wine, or taxi cab? That's what we want to get at.
BF: It sounds like you are cooperating with Google as opposed to Powerset, which tried to be a Google Killer. So you're leveraging what they've already searched and what they have been indexing. So let's pop up the search destination site Digger.com, which is not live. If anyone wants to play around with it, all you have to do is type in "vator" in the comments field for the question, "How did you find out about Digger?" You'll have access in a day. This is going to be launched the second half of 2009 and this is for the power user. So, let's take a look at a query, "Cape Cod golf course with ocean view." Please explain how you came up with these relevant results.
TM: Well you notice our system is grouping this pretty long query into phrases. It assumes that Cape Cod goes together. It's a place. You're not talking about the fish, cod. And "ocean view" brings up a "beach front" result, so the technology sees beach front as a semantic variation of ocean view. So that's the kind of thing that makes it stand out a little bit.
BF: So let's pop up Google and see the result there. So the result is showing ocean edge results is the top result. By looking at the Digger result, what do you think is different in terms of what you are trying to provide.
TM: In this case, there are so many golf courses that if you don't care about completeness of results and you just want the first one, maybe Google is fine. But if you're used to sitting down and saying what was that particular golf course resort, and you don't want to think of how many different ways to type a query. You don't have to write "beach front," because this tool automatically expands the contextually appropriate variation of ocean front. So you're getting a more complete result set.
BF: What kind of searches make sense for Digger vs Google?
TM: We think that the queries that tend to be longer queries that are particularly more specific are things where Digger is going to be a more powerful tool than just the more default kind of Google search. We think that Google is going to have the biggest index and be fastest search engine and is going to be the simplest use for a long time, but that's not always what you want.
BF: And what is going to be your business model here? Is it going to be paid search?
TM: If we do paid search, you can expect us to do Digger style which is semantic as well. You won't have vacation rental, a handheld company, and a fortune teller all bidding for the word palm. But instead, that will be disambiguated on what kind of palm the advertiser is talking about.
BF: Ok, Tim Musgrove on innovative search experience. We wish you the best of luck with Digger.