finding out which searchword is on which page

[wnoordhoek]

in context of the statistics database id like to know how to figure out which searchwords that are searched for are on which pages....

this way i can optimise search statistics and present the documentalist that wants to know this kinda things too, a result for her work


i know its in the files but i havent figured out the structure on how to link one and other together

(remember i want the stats in the database too , im that difficult person )

[Ray]

I'm not sure if this is what you are asking - but do you mean that you want all the search results for each query, also stored? I'm unsure as to why this would be useful - that same query can be fed through the search engine and you can get the results in an instant.

Each query may return many results. For example, if you search for a word that appears on every page of your site, you could have 100,000 URLs returned. So for each search query performed, you would have to save 100,000 URLs. Even if we did not write the actual URLs and only represented them with pointers into our index data - that would still be a big waste. This would also be inaccurate, because when you change your site (and the index data changes), so would your stored search results.

Would you really need all search results ever performed to be saved? Surely you would only be interested in a selected few at a time, and it would make more sense to execute the search for those few words instead? Have you considered adding a link from the statistics reports page to the search page, passing the query along as a request?

[wnoordhoek]

hey, i didnt think of passing a link with the searchwords in it.... nice idea

what i meant was
id like to see on which pages the searchwords are present, however storing them with my stats is getting big indeed, so i figured it would be a search while showing the stats, and then they can put that in the reports to show...

for example...
process the searchlog
show them the searchwords with the countings of how many times there has been searched on that word....
then when the user clicks a link for a specific word then they will see on which dates the words have been searched for and on which pages the word has been found


sounds like im looking for a "plugin" (in my case in php) that returns purely the page urls without anything else (no layout no nothing) just the url itself on the fly so i can put those urls in the stats page in my way i want......

[Ray]

I think it raises too many configuration issues to be a standard feature. What you would want ideally would be a stats page which provides a link to the search page with the search word passed as the query - however, the stats page at the moment can be viewed offline (on your local hard disk) or online (on a web server). By adding links between the stats page and the search page, there will be a co-dependance there and it raises many potential issues (path/link to the script, security with offline/online zones in browsers, etc.), especially considering all the various scenarios/configurations that Zoom can be used in.

However, you are free to implement your own solution using the log file, which is in CSV format for this purpose, and there are many ways to parse and re-interpret the data as required. You can then come up with something that best fits your exact scenario.

[wnoordhoek]

i think indeed it will be the best to have a link to the searchscript with the searchword directly filled in.... for now

i guess im just too demanding for stats the way i like it

right now in your software u have stats generation code....
if that could be put into asp/php (i know there are extra libs needed for php to put graphics on tho, and probably for ASP too, but that is user choice to use then) then it would also be more dynamic in the way that the one that needs the statistics can pull them up live.....


im persistant i know but sometimes i get ideas in my head that i need to have done or denied before i move on

hope to see ur reply soon again

[Ray]

We're not planning on implementing server-side statistics reports. As you mentioned, there are extra requirements for libraries and components to be installed on the server to provide graphical capabilities - and they are not at all common or easy to install (and would further complicate support issues). Because of this, we would have to also keep the current offline statistics report tool in the Indexer (for people who can't install the components required for server-side graphics, etc.), which means quadrupling the development and maintenance required for the Indexer, PHP, ASP, and CGI versions. With these things in mind, we don't think it is practical to do this for the minimal gain (if any) in usefulness.

You can schedule statistic reports to be generated automatically. This would allow you to update regularly in a fairly transparent process. Should also note that statistics should not change enough for real-time analysis to make much difference. A top 10 search word would not differ in the duration of several hours when the entire log file covers a period of weeks, months or years.

In fact, if you think about it, there are more disadvantages to offering real-time statistics generation because you are putting a heavy load on the server CPU unnecessarily - as opposed to doing it just once on your computer, the server would have to generate a report everytime someone requests for a report.