Wednesday, January 25, 2006

Who owns your personal information?

There are increasing concerns about who has access to information about consumers, whether it be marketing firms with their "targeting" based on your credit card usage, selling of cell-phone call records, or what keywords you may have used in a search engine. Just today there was an article in the New York Times by Katie Hafner entitled "After Subpoenas, Internet Searches Give Some Pause" concerning government efforts to force Google to hand over records of user search queries. Beyond the usual privacy concerns, my big question is this:

Who owns information about a consumer?

And I would suggest that the answer should simply be: The consumer owns all information about the consumer. It should not be Google's or any other vendor's property to do with as they please.

I've been working on a white paper that concerns how to use software agent technology to support the development of a quantum-leap knowledge-based web for consumers. It's still very rough with lots of work needed, but one of my tenets is that consumers own all information (knowledge) about themselves. Another tenet is that my envisioned consumer knowledge web is consumer-centric. Not merely consumer-oriented, like many services, but consumer-centric in a way that forces people and vendors and governments to accept the the consumer is the center of it all, not proprietary business interests.

You can find my draft white paper here: The Consumer-Centric Knowledge Web - A Vision of Consumer Applications of Software Agent Technology - Enabling Consumer-Centric Knowledge-Based Computing.

-- Jack Krupansky

7 Comments:

At 7:04 AM MST , Anonymous John said...

Jack,

I applaud your investment in distributing agent information. I am examining your "white paper". It is broad, but how can I examine a topic deeper? For instance:
1) What is a practicle agent framework (java) to deploy today?
2) My agents may not be "intelligent". My ontology is constructed on category, subcategory, item {list of list of attributes} category mgmnt. My agents simple capture buying preferences.

You can see my interests are near-term and limited. John

 
At 9:19 AM MST , Anonymous Jack Krupansky said...

Unfortunately, my primary message is that much more deep research is needed. Further, that attempting to deploy solutions using today's technologies will create and cause far more problems than will be solved. Sure, you can do some limited, point-solutions that address some narrow, focused, near-term concerns, but my interest is in deep solutions to bigger, broader, longer-term concerns.

Sorry I don't have a "quick fix" for you.

I sincerely believe that we really need to kick off a massive research program ASAP, or so many of the issues will overwhelm our patchwork of lame, partial solutions in the years and decades ahead.

-- Jack Krupansky

 
At 7:59 AM MST , Anonymous John said...

Jack,

I value your observations and the information gained from your posts - (for instance- The book Service Oriented Computing) however; I fundamentally disagree with your ideas on "deep" research.

I believe that elegantly addressing specific problems (polishing gems) is the road to progress. Using an example from theoretical physics: In 1905 Einstein published short papers which proved fundamental in Kinetic Theory (random walk), relativity (no absolute space/time) and Quantum Theory (light quanta). A beautiful summary of this work:
http://www.symmetrymagazine.org/pdfs/200502/
voices_einsteinpapers.pdf

He did not approach these as he did Unified Theory but addressed specific problems. That said -- can you point me in a direction on my problem?

 
At 9:51 AM MST , Anonymous Jack Krupansky said...

I'm in a rush this morning, but hopefully this weekend I'll respond to the overall issue of the role of deep research to progress, maybe as a separate post.

Rest assured that my ultimate goal is the practical application of research. I just want a deep foundation so that we can build truly robust structures.

That said, I personally think Einstein's uncomplete work towards a Unified Field Theory was a far greater contribution than his earlier work.

I'm no Einstein, but if I were I'd rather work on unified theories that have dramatic long-term applications than sit around polishing mere "gems".

Back to your original problem/question, I think you need to elaborate your question for me.

As far as a practical agent framework, there are a variety of patchwork attempts, but for any given application you may be better off rolling your own, based on existing work of course, but heading off in a deeper sense that helps leverage your specific applications. And if your application is so simple that such an approach is not warranted, maybe the whole application can be implemented without the extra complexity of software agents. Ultimately software agent technology will become easier, but for now it's a pain.

On the issue of intelligence, there is one school that wants each agent to be intelligent and another school that wants intelligence to be "emergent" from the collaboration of larger numbers of agents (e.g., swarms).

On ontology, I'd first ask whether you have a true ontology or simply a database schema. Your list seemed doable with an SQL schema.

-- Jack Krupansky

 
At 11:34 AM MST , Anonymous John said...

Discussions can be powerful learning experiences, despite fundamental disagreements. I, and I believe most physicists would side on the fundamental contributions of Einstein's early work.

Elaboration of my limited expectations:

In the future, I wish to adapt an ontology for retail items. For concreteness assume ski items (boots, skis, clothing). I have not found an established ontology, so I am prepared to make one with protege. However; my preference is to adapt an established ontology. My understanding is that the ontology gives me a "communication language" with future ski sites.

You are correct, I currently handle this with Hibernate, an ORM mapping tool to a database. Jade has an example of wrapping java in an ontology -- I would probably follow that approach when integrating the ontology.

Agents I do not require an "intelligent" agent. I want to construct a simple buyer/seller agent constrained to my ski ontology -- to prevent this from being an academic exercise -- I would like to adapt an approach with a possiblity of having future server agents. Something as simple as the Jade bank example would meet my needs.

Motivations:
The site functions without agents. I wish to make architectural decisions that will accomodate agents in the future. Presently a buyer (agent) can use drop down boxes to select subcategory attributes (Saloman skis, 180cm, etc)and search my database. In the future this agent would use an agent server to broaden the search. I would not do financial transactions so leave security concerns aside.

Constraints

The framework must be in java.

It will be a component in J2EE.

I am interested in Rules engines.

My limited review leads me to Jade and a Jar that includes Jess (rules) engine. Inference is presently not an interest.

My concern is that I will adapt technologies that knowledgable people would shun. There are many abandoned frameworks.

Thus I seek guidance.

 
At 12:09 PM MST , Anonymous Jack Krupansky said...

I don't have any guidance as far as commercial deployment of software agent technology today other than simply: Don't do it.

Software agent technology is still a very nascent field of research and is simply *not* ready for prime-time deployment, other than in limited, narrow niches.

On the other hand, if you want to take one of the existing packages and enhance it to suit the needs of your application, that's possible, but only with significant work, and still involves significant risk.

At least from your cursory description, it certainly doesn't seem that your application can be decomposed easily into a large number of agents. In fact, your application seems like a traditional client-server application - multiple clients, but basically a single server.

Multi-agent technology makes sense when you have dozens, hundreds, or even thousands of agents running on the server side (not the client side).

A lot of proponents of agent technology go to extraordinary lengths to justify their hype, but they really can't escape the fact that they don't have the research results to back their extravagent claims, yet.

-- Jack Krupansky

 
At 5:01 PM MDT , Anonymous Anonymous said...

Hi,
I doubted anybody could 'force' Google into a consumer centric internet platform. Look at the numerous reverse phone lookup service provides out there, nobody even can stop them from "own" all your personal information, in this case your telephone and cell phone records. To find out more about reverse phone lookup service, you can visit this website at www.fonerecords.com.

Regards,

Ray

 

Post a Comment

Subscribe to Post Comments [Atom]

Links to this post:

Create a Link

<< Home