From Innovations, a website published by Ziff-Davis Enterprise from mid-2006 to mid-2009. Reprinted by permission.

Can you bottle innovation?  Conventional wisdom says no; innovation comes from inspiration backed by knowledge.  It can’t be packaged or automated.

However, a Boston-based company is challenging the conventional wisdom. Innovation Machines has developed technology that applies a semantic search engine to the task of mining possibilities for innovative new materials and procedures.

If the phrase “semantic search” means nothing to you, join the club.  I got a telephone briefing on Innovative Machines’ technology and couldn’t quite figure out what it did.  So I stopped in for a visit and got one of the more impressive demos I’ve seen in recent years.

I’m a veteran of thousands of demos, and so have learned to be skeptical, but this was interesting stuff.  Innovative Machines’ customer list would indicate that the company is on to something.

Semantic search involves mining text documents not only for terms but for relationships between terms. Most search engines can’t do this. They can deliver some insight by finding words in close proximity to each other, but they don’t establish a clear relationship.

For example, if you search for “smoking” and “cancer” on Google, the results indicate there’s a relationship between the two, but the search engine won’t explicitly define that relationship. Semantic search goes a step further. It’s intended to deliver a small number of results but with terms that are specifically related to each other. For example, a semantic search engine might infer from its search that smoking and cancer are related and return documents that explain that relationship.

The semantic engine is at the core of what Goldfire does. A host of other features are wrapped around that, including a project workbench and a database of scientific and patent literature.  The demo I saw showed one example of how innovation can be guided, if not packaged.

Suppose your company makes packaged food and you want to figure out a way to substitute artificial sweetener for sugar.  Engineers can use the workbench to deconstruct ingredients in the current product and then test the substitution of various artificial sweeteners.  Goldfire’s scientific database understands the characteristics of alternative ingredients, such as texture, taste, heat tolerance and chemical interactions.  A researcher could model the impact of substituting different artificial sweeteners and determine which ones are good candidates for a new recipe.  By querying on the attributes of potential substitutes, engineers could also discover new ingredients they hadn’t thought of.

The patent database comes into play when attempting to innovate on existing intellectual property.  For example, an automotive engineer could deconstruct the components of a patented turbocharger and test the impact of substituting different metal alloys.  This could lead to an improved design that doesn’t infringe on existing patents.  In fact, Invention Machine says this re-engineering of existing patents is one of the most popular applications of its product.

Goldfire isn’t a simple product and to use.  Customers typically go through several days of training and setup to customize the software to their industry.  It also isn’t cheap; installations run in the six figures. For the kinds of problems Goldfire is meant to solve, however, these costs aren’t surprising.

Goldfire is a difficult product to describe, but an easy one to understand once you see it in action.  The company provides several podcasts and videocasts that demonstrate how customers are applying the technology.  This isn’t innovation in a bottle, but it’s a pretty good start.

Incidentally, I have no financial interest in the company or its product. I just think this is a technology that deserves more attention.