What Can You Do with Saffron? Get Creative!

Our customers define themselves as creative entrepreneurs, whether in startups or large corporations.  The SaffronMemoryBase is a new-new thing.  It is a universal form of intelligence for sense making, decision making, prediction, and more.  We provide the “engine” while our customers and partners make the planes, trains, and automobiles.  Or perhaps a new “vehicle” we’ve never even thought about.

To help explain and demonstrate the power of associative memories, we will be providing more and more open source examples and method packaging.  We will keep the REST APIs lean and mean, like a RISC (Reduced Instruction Set Computer) architecture, but we will add methods and examples to support more and more applications.

Otherwise, get creative about how to compose these methods and use the RESTful object model for almost anything you can think of.   For example, we’ve got a Twitter demo on deck!  See Jared Peterson’s posting on “Observing and Querying Tweets (Part 1)”, with more coming soon from him.  We’ll also start sharing more about the business problems we’ve solved with our customers/partners.  Remember that memories can be applied to sense making by “reading” massive data sources as well as decision making by learning human behaviors.

Let me start with one example.  See Sociocast, www.sociocast.com, and the cofounders, Ari Golberg and Albert Azout, as incredible examples of creative entrepreneurs (with Dan Gilbert for visionary financial backing).    Also see Stylecaster, www.stylecaster.com, for how Sociocast is being applied to personalization and marketing in the fashion space.  I warn you: if you are a fashionista, the site has been called “addictive”!

You can imagine that a Saffron-based personalization system starts by representing each customer with a “memory”.  You might also imagine how positive and negative feedback can be modeled by two opponent “matrices” with each memory.  Such customer memories move far beyond “collaborative filtering”, which is really a form of clustering, as in course-grained market segmentation that stereotypes customers into one or another group.  Instead, a memory for each customer allows true one-to-one marketing, acting toward each customer as an individual.  Moreover, beyond linear modeling of each customer, the matrices learn the interactions of product attributes, the environment, and product ensembles.  Who cares how much you like the color red?  A memory knows when you like red and what goes together with it, depending on all the other context too.  For fashion, a memory knows what matches (at least the matches you like).

Great!  But the genius of Albert and Ari is to also see how social influence is fundamental to marketing.  Beyond social friendship networks that represent who knows whom, it is more valuable to know who influences whom.  Who do you listen too?  Who listens to you?    Who do you look to regarding specific products and product attributes?   As Ari says, “Who cares who your friends are?  Influences are what drive behavior.”  They developed Sociocast on top of Saffron to also store the sources of product recommendations, the directional links of social influence, as part of the customer memory.

Better than even I could articulate in this case, Albert describes SaffronMemoryBase as a “manifold of near infinite subnets”.    A memory base is not just a single connection graph or even a hypergraph.    Depending on the nonlinear combinations of product’s attributes, environment, and sources, the memory base expresses a contextually weighted sub-graph of the influence “flow” for each specific product pattern.  Over all the possible combinations of product and context attributes, there are near infinite such expressions.   Albert and Ari’s genius is to understand and use this power for marketing.

I’ve not described even half of what Sociocast then does with these memories, computing not just how much you might like or dislike a product and who might be influenced by your adoption of it.  Sociocast also adds some pretty sophisticated Network Science about social “pressures” and network “visibility” for how consumers act within markets at large.  Sociocast also makes all this continuously “alive”, in the sense that memories update and the information propagates over the network all the time, with every new event.    Socicast is a memory base with dynamic thinking.

So what can you do with Saffron?  Get creative!   Let us know what you are thinking and how we can help.

Tags: applications, personalization, Sociocast

This entry was posted on Saturday, December 12th, 2009 at 9:58 am and is filed under Natural Intelligence. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.