Indexing RoboHelp 7 webhelp output - terms matched

[fchapman]

We recently purchased Zoom to provide a better index for RoboHelp 7 webhelp output. In general everything looks as if it's working well, but I'm puzzled about the indexing and worried about the extent of the problem.

If I search for "copy" the results tell me that the "Save report" topic has 1 matched term. However, if I open the topic I can see copy 3 times in the main text and a further 2 times in DHTML drop-down hotspots. The "copy and paste" topic is reported with 2 matched terms but clearly has it 13 times in the body of the topic. Any ideas about what's happening here?

The unfortunate result is that important topics are shown low on the results, and I wonder how many other places this is happening. I need to get this working properly on the topic text because RoboHelp strips out all Zoom meta data on compiling the output.

[David]

Matched terms does not refer to the number of instances a word appears on a page. It refers to the number of search terms matched.

So a search for, dog cat mouse, has 3 terms. And can match up to 3 terms with any particaular result page. Obviously a page with all 3 terms is going to be more relevant than page with just 1 of the 3 terms.

Now is the case of a single word search, like copy, it would normally be impossible to get more than 1 term matched. But if you have turned on auto wild card substring matching (which we don't recomend) then it is possible many more terms matched.

If you then search for, cat, then this becomes *cat*, which will match words like catalogue, location, truncated, etc.. and you have get a high number for terms matched.

In your case, copy, might be matching Copyright and other similar words.

See also this FAQ
Q. How do I make some pages appear higher up in my search results? How does Zoom's page score system work?

[fchapman]

Thanks for the explanation

Thanks for your clear reply. I had turned on the substring matching because I thought that words like "copy" should really be matching "copying" too. However, your example shows how this can skew results so I've turned it off again.

Thanks also for the link. I'll see what I can implement to improve things. It's a real pity that RoboHelp 7 is so keen to strip out meta tags (I can't upgrade to RH8 which is better about this because we have mid-term plans to move to Madcap Flare).