British search and information management vendor Autonomy will later today announce it has a new product coming down the chute, that it says helps to find formerly hidden patterns in databases: IDOL SPE.
In an exclusive interview with CBR yesterday, the firm’s founder and CEO Mike Lynch said that the new technology has been in beta with some of Autonomy’s customers for several months, and will probably hit general availability, “In another month or two.”
Listen to an exclusive CBR podcast with Mike Lynch discussing the news here.
Lynch said that this is the largest software development project the firm has been working on for the past three years; the company spends around $100m a year on R&D.
The SPE in the product’s title stands for Structure Probabilistic Engine. Lynch said that the technology is able to detect patterns in structured data – i.e. databases – that would previously have been hard or impossible to find. “A common problem in databases is when you search for something and it comes back with no matches,” Lynch said. “Instead of only looking for black and white results, our probabilistic engine can handle shades of grey too.”
“We’ve taken our IDOL technology that works so well for searching and analysing unstructured data and turned that around to look at structured data,” Lynch said. “We’re bringing our Meaning Based Computing to the structured world.”
Lynch said IDOL SPE will be able to run standalone on the IDOL platform, but will also be capable of being layered on top of relational databases, such as Oracle, SQL Server or IBM DB2. Lynch said that it is possible that database companies will want to OEM the product from Autonomy, and that, “We would argue that the kind of things it can do might well become standard as part of databases in the next few years.”
To describe how the technology might add value in a real-world setting, Lynch said that flight booking systems that can’t find a direct flight between two destinations might instead be able to find flights to the next nearest destination.
It does this in two ways – by looking for patterns in the data itself, and by looking for patterns in the usage of the information in previous queries or transactions. “What you’re now getting is partial answers from the database, so rather than database currency just being about black and white, it’s now able to handle greys,” Lynch said.
When the technology hits general availability it will not be built into the IDOL Server – it will be considered an additional function and hence with an additional license fee.
Lynch said that the firm has not shown the technology to any database companies just yet, just to a number of beta customers.