Unusual searching behaviour (web) - can anyone explain this?

First of all, this isn’t critical. I’m just curious as to why Lotus Notes is doing this and whether it is designed to work this way.

If I do a search for part of a word on the web (for example, ‘axa’ to show up all instances of the word ‘taxation’) I will sometimes get a result (document found) when I should get none (no instances of ‘axa’ in the returned document).

I’m using a ‘$$SearchTemplate for blah, blah view’ form for searching. The full-text index is up-to-date (deleted, then recreated).

When I open the returned document and view the page source, I find that the displayed page contains the following text:

function _doClick(v, o, t) {

var url=“/domino/website/website.nsf/docid!OpenDocument&Highlight=0,axa&Click=” + v;

if (o.href != null)

o.href = url;

else {

if (t == null)

  t = document._domino_target;

window.open(url, t);

}

}

It appears to me that the code responsible for highlighting search keywords in the returned documents is somehow being detected as a hit. Alternatively, the document is mistakenly being detected as a match (for some reason) and then the source of the ‘$$SearchTemplate etc…’ form which adds the ‘&Highlight=0,searchkeyword’ code to the destination URL for the links to each returned document is adding this to the page source after the fact (the highlight code may be incidental, rather than the source of the problem). Which do you think it is?

Either way, it’s weird. Any idea whether this is normal behaviour?

Paul.

Subject: Re: Unusual searching behaviour (web) - can anyone explain this?

The code you posted is dynamically generated by Domino when the page is rendered to HTML – it is not actually contained in the document. It therefore should not “cause a hit”. The only thing that should matter is your full-text index, which contains only the information stored in the fields (and possibly attachments, if you have that enabled) on the documents themselves.

It sounds like your index is out-of-date. Are you sure it’s up-to-date when you execute the actual search? Just because you rebuilt it 10 minutes beforehand doesn’t mean it’s fully accurate at the time of the search itself.

Subject: RE: Re: Unusual searching behaviour (web) - can anyone explain this?

Hi Erik,

I thought it would be odd if the highlight code was able to cause a hit.

I’ll check the full-text index again.

Thanks,

Paul.