Re: People who searched this also searched this!

Well, that would work for a *very* limited scope only. Just imagine the data you´d need to store & analyse even for the simplest kind of solution: For each item, you would have an associated table / file / whatever with the names / IDs of all other items and number of hits (it really depends on the exact circumstances what you would consider a hit. In a search engine like setup, you could say that for each search that actually results in the user selecting one of the results rather than searching again, each of the top 10 results gets a hit from every other -- > very complex. In an online shop, you could do this with less data: You could say that every item a user buys during one session (every item in the basket when it´s submitted) is a hit). Some kind of SQL implementation of this might work just fine for an online shop with, say, 20 items and about 1.000 customers a month that wants to have people-who-bought-this-also-bought-that statistics. But if you want to have a Google-like ranking system, PHP is not the way to go. For that, you need neuronal nets (which can´t be coded efficiently in a scripting language) and years or decades of experience are for. "Dwalker" <dwalker@healthyproductsplus.com> schrieb im Newsbeitrag news:01c275e9$21451340$e066d3c6@b... > Wouldn't the solution really depending on the size/complexity, popularity > and scope of the site?? > > Seems like you should be able to log session id information along with each > of the search parameters/words issued during the session straight into a > database separate from the database(s) being searched. After the user > searched a database(s), you might search your 'word and sessionID' database > using the original search parameter as a means of locating an attached > session id to create a daisy chained link to other words. The question > becomes: how far into the WORD database do you want to dig? You might > implement some kind of recursive snoop for other words if, for example, the > returned list of words is too short. > > > > > THIS E-MAIL MESSAGE AND ALL ATTACHMENTS TRANSMITTED HEREWITH ARE TRADE > SECRET AND/OR CONFIDENTIAL INFORMATION INTENDED ONLY FOR THE VIEWING AND > USE OF ADDRESSEE. IF THE READER OF THIS MESSAGE IS NOT THE INTENDED > RECIPIENT, YOU ARE HEREBY NOTIFIED THAT ANY REVIEW, USE, COMMUNICATION, > DISSEMINATION, DISTRIBUTION OR COPYING OF THIS COMMUNICATION IS PROHIBITED. > IF YOU HAVE RECEIVED THIS COMMUNICATION IN ERROR, PLEASE NOTIFY THE SENDER > IMMEDIATELY BY TELEPHONE OR ELECTRONIC MAIL, AND DELETE THIS MESSAGE AND > ALL COPIES AND BACKUPS THEREOF. THANK YOU FOR YOUR COOPERATION. > > -----Original Message----- > From: Krzysztof Dziekiewicz <krzydzie@wp.pl> > To: php-general@lists.php.net <php-general@lists.php.net> > Date: Thursday, October 17, 2002 2:52 AM > Subject: Re: [PHP] People who searched this also searched this! > > > > Hi guys, > > > Does anybody have a working example of doing the lists of 'People who > > searched this also searched this!' that I see on Amazon and other > > websites. > > That is a very complex problem. It is not for PHP. PHP may only insert > queries > and get results and another system computers all data. Such system is > based on > statistic or artificial neuron nets. For example the company "Net > Perceptions" > makes such systems as NetP based on statistic but they are not very fast. If > you > would like to buy their system you would have to sell some good cars. > > -- > Krzysztof Dziekiewicz > > > -- > PHP General Mailing List (http://www.php.net/) > To unsubscribe, visit: http://www.php.net/unsub.php > > > >