MARC is an extremely flexible data structure. Relational databases tend to view the world as a set of linked tables with data elements. Of course as anyone who has ever survived a CS data structures course can tell you, you can use almost any data structure to store any other data structure. Its mostly a matter of efficiency. Yes, people have stored MARC data in a relational database. It worked, sorta of.
Think of a table where any column could be repeated any number of times, or omitted, or be of any length, or.... A book can have no authors, or 1 author or lots of authors, or some authors are really corporations, and other authors are really pseudonyms, and some authors have different names in different languages.
Not an insurmountable problem if all you have
is a relational database. The usual method is breaking the MARC record up into lots of itty-bitty pieces. You have the author table, the title table, the uniform title table, the serial title table, etc. The problem with breaking it up is you have to put it all back together again. Think about doing a join across 900 different tables. At this point CJ Date starts in with denormalizing your data....
Its a case of when you have a hammer everything looks like a nail. Relational databases have their place, but they aren't always the best way to store and retrieve all types of data. There are databases which aren't relational, and sometimes they outperform a relational database on specific types of data.
All Library of Congress MARC records created after 1968 by the Library are not copyrighted within the United States of America. Anyone in the USA can use LC's post 1968 MARC records for any purpose whatsoever. I used to publish the NEWBOOKS list on Usenet listing all the new computer science books cataloged each month by LC. And my former employer made the entire post-1968 catalog available on the Internet years before LC got an Internet connection.
The Library does maintain the copyright on its records outside the United States (you didn't pay US taxes did you:-). Any bulk use of LOC MARC records by people outside the USA must be licensed, with the money going to the US Treasury, thank you for your contribution to paying our national debt. If you want your
own copy of all the complete post-1968 catalog, you can pay LC a tape duplication fee which is supposedly only enough to cover their costs. Of course, its the government the home of $700 toilet seats, so their costs are very high.
Library of Congress MARC records prior to 1968 are a bit more complicated. In exchange for the taxpayers receiving a discount for the work, the Library agreeded a private company could convert LC's paper catalog cards into electronic records and sell the records to other people. Of course, you are still free to go to the Library and copy all the cards you want. Nor is there anything stopping another company, or just a group of concerned citizens, from making their own copies of the paper library cards and distributing them for free. You just won't get the electronic files from the private company, you'll have to type them or scan them yourself.
Slashdot Mirror
MARC is an extremely flexible data structure. Relational databases tend to view the world as a set of linked tables with data elements. Of course as anyone who has ever survived a CS data structures course can tell you, you can use almost any data structure to store any other data structure. Its mostly a matter of efficiency. Yes, people have stored MARC data in a relational database. It worked, sorta of. Think of a table where any column could be repeated any number of times, or omitted, or be of any length, or .... A book can have no authors, or 1 author or lots of authors, or some authors are really corporations, and other authors are really pseudonyms, and some authors have different names in different languages.
Not an insurmountable problem if all you have
is a relational database. The usual method is breaking the MARC record up into lots of itty-bitty pieces. You have the author table, the title table, the uniform title table, the serial title table, etc. The problem with breaking it up is you have to put it all back together again. Think about doing a join across 900 different tables. At this point CJ Date starts in with denormalizing your data....
Its a case of when you have a hammer everything looks like a nail. Relational databases have their place, but they aren't always the best way to store and retrieve all types of data. There are databases which aren't relational, and sometimes they outperform a relational database on specific types of data.
All Library of Congress MARC records created after 1968 by the Library are not copyrighted within the United States of America. Anyone in the USA can use LC's post 1968 MARC records for any purpose whatsoever. I used to publish the NEWBOOKS list on Usenet listing all the new computer science books cataloged each month by LC. And my former employer made the entire post-1968 catalog available on the Internet years before LC got an Internet connection.
:-). Any bulk use of LOC MARC records by people outside the USA must be licensed, with the money going to the US Treasury, thank you for your contribution to paying our national debt. If you want your
The Library does maintain the copyright on its records outside the United States (you didn't pay US taxes did you
own copy of all the complete post-1968 catalog, you can pay LC a tape duplication fee which is supposedly only enough to cover their costs. Of course, its the government the home of $700 toilet seats, so their costs are very high.
Library of Congress MARC records prior to 1968 are a bit more complicated. In exchange for the taxpayers receiving a discount for the work, the Library agreeded a private company could convert LC's paper catalog cards into electronic records and sell the records to other people. Of course, you are still free to go to the Library and copy all the cards you want. Nor is there anything stopping another company, or just a group of concerned citizens, from making their own copies of the paper library cards and distributing them for free. You just won't get the electronic files from the private company, you'll have to type them or scan them yourself.