google uses a filesystem... yes, that's no relational database. It's not like GoogleSQL it's more like GoogleFS.
David, First try to learn basics of SQL, you can go to W3Schools Learn SQL page <http://www.w3schools.com/sql/default.asp> then you can try using RevSQL or Valentina to play with SQL commands. What I do in your case was to have three tables Table 1: the data. In this table I'd hold the data, the real data, like in case of reddit or delicious is the URL. Table 2: the keywords In this table I hold metadata, one entry per keyword, all unique. Table 3: relationship between table 1 and 2. This table just maps relationships between keywords and data, this allows me to have a many-to-many relations where I can map a single keyword like 'oranges' to many items such as all the products that are orange-like and to map one product like 'maltese orange' to various keywords such as 'make-nice-soft-drink', 'orange', 'expensive-outside-malta'. I am by no means a DB pro or something like that, but that's the pattern I'd use to hold data and metadata that are so generic that I know nothing about them besides the fact that they might be related to each other. If you know beforehand the keywords and can group them in families or group your data in sets then you probably have a better schema than mine that would speed your searches and use less records. Andre On 6/8/07, william humphrey <[EMAIL PROTECTED]> wrote:
You are asking an interesting question. There is lots of information on database design (which includes schema) but the type of database design you want is more like what they do on the big search engines. When google indexes everything and builds a table of keywords -- I don't even know what they use... Does anyone? On 6/7/07, David Bovill <[EMAIL PROTECTED]> wrote: > Going to have a go at knocking up a database for RevCode handlers - and I > want to create a database schema for it. I know I want del.icio.us style > tagging, and I'll be indexing them with the language terms used as well as > keywords such as "html", "array", "outline" or "view" etc I'm thinking I > need a database as I know I'll want to experiment a lot with searches - > particularly for handlers with the highest hits... and that seems a bit > hairy to do with arrays and memory based indexes? > > So the question is - is there a resource anyone knows of for database > schema's - a sort of design pattern thing for databases for people like me > that know a little but not enough to design one except by trial and error? > Say for folksonomy tagging for a starter? > _______________________________________________ > use-revolution mailing list > [email protected] > Please visit this url to subscribe, unsubscribe and manage your subscription preferences: > http://lists.runrev.com/mailman/listinfo/use-revolution > _______________________________________________ use-revolution mailing list [email protected] Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
_______________________________________________ use-revolution mailing list [email protected] Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
