LOL.. what can I say to that post? Priceless.

I'm not sure why you thought you needed to say anything.  It's not like I addressed it to you or anything.

Of course, it may have helped that I wasn't being a pr... err.. not nice person.

Gotta love the veiled slurs...remember it takes one to know one...a nice person that is!

I agree with Dan...he's got it right.  Kalmbach owes it's existence to the masses.  They should pay more attention.

I really don't want this thread to turn into a comment thread on what's wrong with the publication side of the hobby. If you'd like to discuss that, please open up another thread.

Let's keep this thread on topic - the magazine index and its progress.

Thanks ...

OK sorry Joe for going to far! I realize your doing all you can and I thank you.

Dan

Dan:

I don't want to squelch any one's desire to have their say, I'd just prefer it be on a different thread. Feel free to start one if you would like to continue that discussion - I just ask that it not be in this thread.

If Kalmbach's primary concern is about the security of the database, perhaps they (or the NMRA) could be persuaded to sell it as a stand-alone searchable disc at a realistic price. Though it hasn't been updated for close to two years, it remains a useful tool for extracting information from old magazines.

I am Very Happy that I ran searches on some of my favorite subjects several years ago and put the printed results in a binder.

Hi there,

to get the discussion on the application going again, I designed a first draft for a data model. This is mostly derived from my own homegrown index that I use at home.

See http://home.vrweb.de/martin_fischer/articles.html

You'll need Flash to browse the page.

This is my first attempt at drawing an entity relationship diagram, so there might be mistakes. I still need to add the layout for the keyword system and the users, especially contributors.

Any comments?

Regards

Martin

Why can't you just capture this as a JPG or PDF?  I tried your link and, after a lot of overhead for Flash and then something else to load, I got "license validation error."  

dave

I am waiting to see what is in the entity called "keyword" or in the  "keyword system". After working on indexing for a number of years (see my post on 8/10/2010), I have come to the realization that most of the problems in indexing natural language items lie within this area. For instance, how does one handle variations in spelling, such as:

tank car
tank cars
tankcar
tankcars
tank-car
tank-cars

Will "tank car" be indexed as one word or two? If two, how does one keep it from being confused with the terms "fuel tank" or "water tank"?

Also, hows does one handle synonyms or near-synonyms such as "barrel", "drum", "keg" and "cask"?

In my online index, all set and ready to go, these things have all been addressed. It even handled the guy who was directed there by Google about 15 or 20 times looking for "homosot", when what he wanted was "homosote".

Why spend hundreds or thousand of hours reinventing the wheel?

Rod Goodwin

Rod, I looked at your "index" and while your theory is sound, IMO your implementation (as in the user interface) needs work.

I believe rebuilding something from the ground up using experts from all the areas of web application design should yield the best final product. If we can get access to the Kalmbach index data, we will most certainly export it into a new data model and put a much more state-of-the-art user interface on the front.

What I've seen so far is not something your typical modeler could use very easily. The original Kalmbach index was difficult to use as well, with more than a few quirks.

There's a lot of things to be considered in building a good, easy-to-use index. Deduping and data quality is only a part of the problem. There's also a good underlying data model, and there's a good search engine parser. Then there's designing the output from searches so that it's easy to understand and use.

Finally, there's a no-brainer, friendly web page interface. The more powerful the search parser, the cleaner the index web page interface will be.

Dave,

sorry for that. I checked from two different browsers without problem. I'll upload a PDF version tonight.

Regards
Martin