A keyword search system that ranks articles by degree of match. For an example of a highly successful and FAST! keyword search, please see the corporate site - www.seco-larm.com. The site has several product search systems on board but the one I am focusing on is the keyword search, which returns pages (over 1500 pages on the site) almost instantly ranked precisely by degree of match to multiple keywords, and displays a little sumary of what is on each page, as well as letting you go directly there in a flash. The search engine itself was developed by someone who posted it for free use by the JavaScript Source. It was then modified and somewhat improved by yours truly.
To implement this search system, the keywords for each page that a JoeUser creates would have to be entered by the user and then - presumeably usine PHP or something equivalent - added to the search database. The database is included as part of the search system. I.e., it is downloaded when you call up the Search System, so that the search itself is not using the internet at all, reducing bandwidth demand for you as well as enormously speeding up the search process.
To make the system even faster and more compact, a dictionary table could also be referenced. The separate searchable pages in the four product category databases for seco-larm.com only run to a few hundred at most, while I hesitate to imagine how many pages total are encompassed in JoeUser.com. If the PHP engine replaced the keyword entries with smaller pointers to a list of all keywords, then it is possible that both search time and space could be drastically reduced. And of course, most of a page address is boilerplate, which you repeat ONCE if you're coding smart.
The remaining problem, which I am working on, would be to turn the search query into an "aquarium," which all former Amiga owners will understand instantly. The "Aquarium" was the massive database of all Amiga software that presented the user with the option of entering keywords manually or selecting as many keywords as desired from a grid. The system would then return all the matching pieces of software - very quickly. (The name came from the most prolific compiler and distributor of public domain software for the Amiga - Fred Fish.)
The advantage of an Aquarium is that the keywords do not have to be guessed at, and you don't have to deal with dozens of synonyms. You can easilly put a thousand keywords on a screen table. The user clicks on every keyword they want to match and then hits a GO button or Return. The option should be for the creator of an article to select from the Aquarium table first, and then, if necessary, add unique search terms not in the table.
Further refinements might include letting the creator weight search terms - at least those from the Aquarium table, and, of course, allowing the user to save searches and refine them without having to try to select the same keywords again.
