|  | Search result ranking |  | |
| | | Joe |  |
| Posted: Wed Jun 11, 2008 10:39 am Post subject: Search result ranking |  |
I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following:
1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order
This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas? |
| |
| | | Jerry Stuckle |  |
| Posted: Wed Jun 11, 2008 10:39 am Post subject: Re: Search result ranking |  |
| |  | |
Joe wrote:
| Quote: | I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following:
1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order
This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas?
|
And your PHP question is?
I'd suggest you find a more appropriate newsgroup - one that knows about search engines. You'll get a lot better answers.
Most people here don't write search engines nor do they have any inkling about how to prioritize results. It's a lot more complicated than just counting the number of times a word occurs.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstucklex@attglobal.net ================== |
| |
| | | Jerry Stuckle |  |
| Posted: Wed Jun 11, 2008 10:39 am Post subject: Re: Search result ranking |  |
| |  | |
Joe wrote:
| Quote: | On Jun 11, 12:06 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote: Joe wrote: I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following: 1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas? And your PHP question is?
I'd suggest you find a more appropriate newsgroup - one that knows about search engines. You'll get a lot better answers.
Most people here don't write search engines nor do they have any inkling about how to prioritize results. It's a lot more complicated than just counting the number of times a word occurs.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ==================
Gordon - thanks, I will look in to that. I didn't realise that FULLTEXT ranked the results, thats why I always used LIKE as it seemed to give more results. I will look in to that.
Jerry - appreciate your point, but implementing search functionality on a website is a pretty common task, and I would imagine a lot of people here have some experience with it. I think the people here are smarter than you give them credit for :)
|
No, it is not a pretty common task. And very few people here have any experience with it. And those who do use something like Google Search - unless they use a CMS.
This isn't about them being smart. It's about not wasting time on doing things which have already been done.
And in case you didn't notice - this is comp.lang.PHP. You aren't asking a PHP question - you're asking a search engine question. You will get much better answers in the appropriate newsgroup.
Or maybe you just figure you're one of those idiots who can ask any question he wants in any newsgroup - despite the agreed upon charter for that newsgroup.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstucklex@attglobal.net ================== |
| |
| | | Willem Bogaerts |  |
| Posted: Wed Jun 11, 2008 10:39 am Post subject: Re: Search result ranking |  |
| Quote: | ... - despite the agreed upon charter for that newsgroup.
|
Sorry Jerry, Google is not very helpful here. Can you tell me where I can find it?
Best regards, -- Willem Bogaerts
Application smith Kratz B.V. LINK |
| |
| | | Gordon |  |
| Posted: Wed Jun 11, 2008 11:07 am Post subject: Re: Search result ranking |  |
On Jun 11, 11:39 am, Joe <j...@faceh.com> wrote:
| Quote: | I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following:
1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order
This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas?
|
several database engines provide a fulltext search facility that will do all the ranking on the DB itself. MyAQL and Postgres (as of 8.3) both have this facility, and I'm sure many others do as well. Check the documentation for your chosen database. |
| |
| | | Joe |  |
| Posted: Wed Jun 11, 2008 11:25 am Post subject: Re: Search result ranking |  |
| |  | |
On Jun 11, 12:06 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote:
| Quote: | Joe wrote: I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following:
1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order
This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas?
And your PHP question is?
I'd suggest you find a more appropriate newsgroup - one that knows about search engines. You'll get a lot better answers.
Most people here don't write search engines nor do they have any inkling about how to prioritize results. It's a lot more complicated than just counting the number of times a word occurs.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ==================
|
Gordon - thanks, I will look in to that. I didn't realise that FULLTEXT ranked the results, thats why I always used LIKE as it seemed to give more results. I will look in to that.
Jerry - appreciate your point, but implementing search functionality on a website is a pretty common task, and I would imagine a lot of people here have some experience with it. I think the people here are smarter than you give them credit for  |
| |
| | | Joe |  |
| Posted: Wed Jun 11, 2008 11:41 am Post subject: Re: Search result ranking |  |
| |  | |
On Jun 11, 12:29 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote:
| Quote: | Joe wrote: On Jun 11, 12:06 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote: Joe wrote: I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following: 1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas? And your PHP question is?
I'd suggest you find a more appropriate newsgroup - one that knows about search engines. You'll get a lot better answers.
Most people here don't write search engines nor do they have any inkling about how to prioritize results. It's a lot more complicated than just counting the number of times a word occurs.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ==================
Gordon - thanks, I will look in to that. I didn't realise that FULLTEXT ranked the results, thats why I always used LIKE as it seemed to give more results. I will look in to that.
Jerry - appreciate your point, but implementing search functionality on a website is a pretty common task, and I would imagine a lot of people here have some experience with it. I think the people here are smarter than you give them credit for :)
No, it is not a pretty common task. And very few people here have any experience with it. And those who do use something like Google Search - unless they use a CMS.
This isn't about them being smart. It's about not wasting time on doing things which have already been done.
And in case you didn't notice - this is comp.lang.PHP. You aren't asking a PHP question - you're asking a search engine question. You will get much better answers in the appropriate newsgroup.
Or maybe you just figure you're one of those idiots who can ask any question he wants in any newsgroup - despite the agreed upon charter for that newsgroup.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ==================
|
Perhaps you are misunderstanding what I mean by search engine... I only mean for searching the content of a single website, for example search for the word 'bacon' in all pages on my website. Pretty much every PHP website I've ever made contains this functionality. And just for the record, a google search embedded in your website is a cop- out.
But now we are definitely off topic, so unless anyone else has anything 'constructive' to add, I'll ask the question elsewhere as my good man Jerry suggests. |
| |
| | | Jerry Stuckle |  |
| Posted: Wed Jun 11, 2008 12:00 pm Post subject: Re: Search result ranking |  |
| |  | |
Joe wrote:
| Quote: | On Jun 11, 12:29 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote: Joe wrote: On Jun 11, 12:06 pm, Jerry Stuckle <jstuck...@attglobal.net> wrote: Joe wrote: I am trying to improve the quality of search results when a visitor enters a search term, and I am interested in any algorithms you guys use and have success with. Currently I do the following: 1) Break the term in to individual words. 2) Perform SQL DB search on individual words using LIKE '%term%' 3) Loop through all results counting how many occurances of the terms 4) order the results from high to low count. 5) display the results in that order This is better than just searching SQL and dumping, however its still pretty basic. Any thoughts or ideas? And your PHP question is? I'd suggest you find a more appropriate newsgroup - one that knows about search engines. You'll get a lot better answers. Most people here don't write search engines nor do they have any inkling about how to prioritize results. It's a lot more complicated than just counting the number of times a word occurs. -- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ================== Gordon - thanks, I will look in to that. I didn't realise that FULLTEXT ranked the results, thats why I always used LIKE as it seemed to give more results. I will look in to that. Jerry - appreciate your point, but implementing search functionality on a website is a pretty common task, and I would imagine a lot of people here have some experience with it. I think the people here are smarter than you give them credit for  No, it is not a pretty common task. And very few people here have any experience with it. And those who do use something like Google Search - unless they use a CMS.
This isn't about them being smart. It's about not wasting time on doing things which have already been done.
And in case you didn't notice - this is comp.lang.PHP. You aren't asking a PHP question - you're asking a search engine question. You will get much better answers in the appropriate newsgroup.
Or maybe you just figure you're one of those idiots who can ask any question he wants in any newsgroup - despite the agreed upon charter for that newsgroup.
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstuck...@attglobal.net ==================
Perhaps you are misunderstanding what I mean by search engine... I only mean for searching the content of a single website, for example search for the word 'bacon' in all pages on my website. Pretty much every PHP website I've ever made contains this functionality. And just for the record, a google search embedded in your website is a cop- out.
|
So am I. And whether you like it or not, google search is a powerful engine and does an excellent job on a site.
| Quote: | But now we are definitely off topic, so unless anyone else has anything 'constructive' to add, I'll ask the question elsewhere as my good man Jerry suggests.
|
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstucklex@attglobal.net ================== |
| |
| | | Jerry Stuckle |  |
| Posted: Wed Jun 11, 2008 12:13 pm Post subject: Re: Search result ranking |  |
Willem Bogaerts wrote:
| Quote: | ... - despite the agreed upon charter for that newsgroup.
Sorry Jerry, Google is not very helpful here. Can you tell me where I can find it?
Best regards,
|
Among other places, one of the "official" sites: LINK
-- ================== Remove the "x" from my email address Jerry Stuckle JDS Computer Training Corp. jstucklex@attglobal.net ================== |
| |
| | | Michael Fesser |  |
| Posted: Wed Jun 11, 2008 2:12 pm Post subject: Re: Search result ranking |  |
..oO(Jerry Stuckle)
| Quote: | So am I. And whether you like it or not, google search is a powerful engine and does an excellent job on a site.
|
I prefer third-party search engines or my own. A simple index-based engine with ranking is not that hard to write. Of course it's not that powerful than a professional engine, but usually enough to start with.
A while ago I used the Google Search via their SOAP API - it was a pain. The results where often not up-to-date, the response from the server was slow and the API allowed only to get up to 10 search results at a time.
Micha |
| |
| Page 1 of 2 .:. Goto page 1, 2 Next | |
|
|