Most web-pages comprise images, text and tables. We have specialist searches for images and text but no search targets tables.
Tabular data is handled by text search engines but only as discrete entries - it cannot be queried in terms of the underlying meaning.
A search engine capable of indexing tables with their column headings, captions and general page context should allow for on-the-fly collation of data with matching descriptors from varying sources with a ranking for how related the data appears to be.
For example - a search for "Names of Kings and Queens of England by Year" might be able to combine data from a number of sources, with matching records building faith in the existing records, and misfitting records shown as 'dubious' but linked to their source.
Such a system would enable journalists, students and even scientists to quickly pull in numerical and other tabulated data on a topic, saving hours of search. The inclusion of tables from scientific papers would be the icing on the cake.
Edit - this is nothing to do with meta tags - it is to do with crawling and indexing content in <table>s-- vincevincevince, Aug 06 2007 MarkLogic Demo http://paycheck.dem...ecs.xqy?ticker=HYSLExecutive Pay Demo from MarkLogic [craigts, Aug 06 2007] So you are asking web site designers to make better use of meta tags? Not much of an invention here.-- Galbinus_Caeli, Aug 06 2007 "Names of Kings and Queens of England by Year" - is a poor example search choice. That information is well known and widely disseminated, and not, frankly, very tabular (it's more of a list).-- DrCurry, Aug 06 2007 A company, MarkLogic, does some stuff like this. Although I'm guessing it requires customization for each type of table you plan on indexing. See demo link.-- craigts, Aug 06 2007 I shall have to submit my carefully collated and tabulated dataset of famous shims of the world, presented according to whether they're a shim or not.-- Ian Tindale, Aug 06 2007 halfbakery