Without getting too far afield in pontificating and math theory, one of the things I learned long ago at the dawn of the personal computing age is that any data set can be searched and sorted in a very short time if the group is arranged in no more than 7 (ideal) logical groups with each organized progressively below that.
To apply this to the real world of any data set which may or may not directly apply to a mathematical sort. Take 1000 pages. If they can be logically split into 7 subsets, that gives you approximately 150 pages to deal with. Dividing those once again into seven subsets gives you only 20 pages at the third level and 7 at the fourth. The challenge is to give careful thought to the top seven categories. If you find that too much is falling into a category, then the structure is too broad, if too few, then the category can probably best be combined with another. For my research on Civil War Soldier records I have 6 million men to deal with. These can be logically split into Union and Confederate, then by state, then by Unit of Service (regiment or battalion), then by Company. So in 4 levels I can drill down from 6 million names to a list of 100 or so names which logically sort on last name. The war itself also breaks down into logical groupings, places (10,000 elements), events (6,000 elements), dates (1500 days). Places break out into states and counties. Events break out into battles, skirmishes, campaigns, and naval actions - and dates break out into weeks and months. Again any given data item can be reached at 3 or 4 levels from the top once you understand the structure. If you're stymied as to how to group your data elements take a look on line at the LOC library classifications. These eggheads spend their entire life putting knowledge into piles. It may (probably will) surprise you that someone has already organized your pile for you. John Rigdon ReseachOnLine http://www.researcholine.net _______________________________________________ pmwiki-users mailing list [email protected] http://www.pmichaud.com/mailman/listinfo/pmwiki-users
