Weightings don't seem to work as expected

[hidefkd]

I've been doing a number of experiments to see if I can get words in the page titles to rise to the top of the searched list, to no avail.

For example, I have a page with the title 'Signal strength'. When I search for 'Signal strength', no matter what I do with Weightings, this page always appears second in the search results. This page has 'Signal strength' listed 14 times, while the page which is listed first has 'Signal strength' only once.

Here is my search:

http://www.hauppauge.com/search/search.cgi?zoom_sort=-1&zoom_query=SIGNAL+STRENGTH&zoom_per_page=10&zoom _cat%5B%5D=-1&zoom_and=1

I have done lots of experiments in both Weightings and in the page itself, and the score is always the same amount apart. For example, I change the Weighting of the title to +5, the score on the top listed page is 288 and the page which should be on top is 282. I change the Page Weighting to +2, and the top listed page drops to 276 and the page with 'Signal Strength' in the title drops to 270.

The difference in scores is always 6, no matter what I set in Weightings.

I have also tried adding <h1></h1> around the searched for words in the 'Signal strength' web page. The scores on all pages change in lock step and the difference in scores is still 6.

Any idea what I'm doing wrong? I would have expected an increase in the score of the 'Signal strength' web page, and that the other web pages would maintain their previous score, but the scores seem to track on all pages.

[David]

Maybe you are indexing cached versions of the pages? You can turn off caching from the "general" tab in the Zoom configuration window. So maybe the changes you make on the page are not being taken into account becuase you will be indexing old (cached) copies of the page.

[hidefkd]

I checked and 'Reload all files' is turned on. Is this the caching option?

Do you have any idea why the scores of the pages move together? In other words, if I add a keyword to one page, and then search for that keyword, the scores on ALL pages move up.

[Ray]

Yes, the "Reload all files (do not use cache)" option is what was referred to by David above.

The latter sounds unusual, and I have not seen that happen. Can you give us the URL to these pages so we can see this for ourselves?

[hidefkd]

Take a look at:

http://www.hauppauge.com/search/sear...ignal+strength

The main page on the topic "Signal strength" is always listed second, while a page while casually mentions the topic is listed first. I've added keywords to the 'Signal strength' page, goosed up keywords to +5, made title +5. All I have been able to do is change the score, but there is always a 6 unit difference between the first and second listing.

Thanks for taking a look. This is pretty odd.

[Ray]

I had a look but can't find anything obviously wrong with the files at this point. We will have to investigate further on Monday when I get back to the office, and see if it is a bug.

Could you e-mail us a copy of your ZCFG file containing the configuration you are using with all the weightings etc? This might be of help in narrowing down what triggers the problem.

[Ray]

We were unable to reproduce the behaviour. With the default settings, I indexed some of the pages on your website, but got expected results (searching for 'signal strength', the support_digital_signal_monitor.html page returned a score of 224 and support_hvr950.html a score of 140).

As mentioned before, we will need to see your ZCFG file before we can continue. Perhaps it is due to a setting here, for example, you can give preference to shorter URLs or smaller pages, and that may affect it.

Also, can you let us know if you are using Incremental Indexing on this set of index files, as that would also complicate the scenario and introduce some other factors of concern.

[hidefkd]

The ZCFG file is on its way.

Thanks!

[hidefkd]

Ray,

I built a new ZCFG file from the Zoom default, and now the weightings are working correctly. It would be interesting to find out what I did to the Configuration to cause this odd problem. But for now, I am good.

Thanks for your help!

[Ray]

We tried some tests in Offline Mode but everything behave as expected. The weighting settings did not trigger any problem like the ones described or exhibited on your site before.

Several things of interest:

• You had "Content density" weighting set to "Strong adjustment", which means that a high priority is given to files small in size. While the files we looked at given from your example and your website were not drastically different in filesizes, this would have influenced the factor a bit. Perhaps the files were more different earlier on, and since then, these files have been changed, and no longer reflect the same situation?
• As mentioned in our e-mails, since it turned out that you are indexing in Offline Mode, we were unable to reproduce the exact same indexing scenario, and can only assume your files online are the same as the ones being indexed offline. If the files are different in any way, this may be a contributing factor to the issue.

It should be safe to set your word weighting back to your preference as before. Unless of course, if you are happy with the results returned from the default settings as it is. But do let us know should the problem arise again.