Lippy951

I will help if I can. I am new to the 944 turbo scene and been going through my trials as we speak.

333pg333

I like the idea. Wonder if some of the information will be more anecdotal than empirical...but enter at your own risk. Maybe like a Wiki system with extra drill downs?

gregeast

I love it and would be happy to help.

This is essentially what us old timers who read the list nearly every day have been doing, cataloging problems and solutions in our (sometimes) feeble brains. While I too get frustrated at times with new owners that ask the same questions over and over I also realize they haven't been reading this thing for the past 10 years, nor are they likely interested in doing so.

There's an incredible wealth of information on Rennlist, it would be fantastic to see it made more accessible to more people.

Good stuff!

Van

Is there a way to have your system peruse through old rennlist threads and do a keyword association to start putting together a web of related events? Kinds of a fuzzy logic scoring match?

Building something that could do that with an existing dataset would be fantastic.

Tsmith84

The idea is cool, but I think you will have a hard time finding scenarios where people have done a thorough diagnosis and documented it to come up with a conclusion and then properly verifying the repair, not to mention following up with the correct fix. I have done a lot of tech support (not really 944) and nobody ever follows up, even when you specifically tell them to. That hurts the system and really drains the "give-a damn" out of the person trying to help out (you in this case). The problem I for see is that this is generally a place for enthusiasts, that don't necessarily have the background, education or tools to properly work through this. But I don't want to be negative Nancy, so by all means, do what you can. I don't think it can hurt. And while i don't know how much I can help, I will try. I just don't work on these cars much, but have a few of my own. Good luck.

seattle951

Yes, there are techniques for working with unstructured text. However, I am not willing to invest the amount of time required to build the solution. It would be a much more difficult project.

seattle951

To make this project work, we would need to focus on a few scenarios supported by data entry templates. We would need Rennlist members to use the website to record the data. We would need hundreds or thousands of records for the machine learning algorithms. Getting sufficient high-quality data is a large question mark. I am using this discussion to determine how much interest there is.

chrenan

Very cool. I'd love to see something like this come to fruitition. Is there an awareness risk? Could the system become self-aware given adequate time and access to the web?

seattle951

I hope not. I was at a conference last year that covered artificial creativity. One team had developed software that could write humorous cartoons. The system was evolving over time as it was developing its own sense of humor based on human interaction with the cartoons.

We are not in the Battlestar Galactica or Terminator era yet, but maybe this is a first step. We should keep our 944s disconnected from the web and under scrutiny.