Did You Mean Katia? – Part 2

This is a sequel to my post “Did You Mean Katia?” which I published five months ago. I seriously hope that I do not have to do a Part 3 further down the road. A search for “CATIA” on the Exalead World Wide Web Search Engine still has no link to 3ds.com on the first page of search results. Not even on subsequent pages. I stopped at page 5. And Exalead still asks me if I meant “Katia”?

Click image for larger view

I was in the audience at DSCC 2010 in Orlando last week when Bernard Charles said that his search engine could index every word of text in a DWG file. He also said that now he had the technology to be able to pull all the 16 billion or so web pages of the entire internet into a company’s intranet and index it. This is all nice to hear. But the capability to do something does not imply that you can do it properly.

I have to assume that the search technology used in Exalead’s world wide search engine is the same that is used in CloudView, their enterprise search engine product. And in my opinion, that search technology is plain and simple horrible. The first search result for “CATIA” points to some page at jobvertise.com talking about jobs in IT/Telecom. The word “CATIA” does not even figure anywhere on that page. The first search result for “AutoCAD” takes me to the web site of someone offering AutoCAD courses in the UK, and not autodesk.com. In fact, autodesk.com does not figure anywhere on the first five pages of search results. I just didn’t look further. I mean, WTF?! Is this a search engine or some kind of a freaking joke?

This is shameful, to put it mildly. I tried searching for “Exalead” and it returned exalead.com as the first result. Phew! What a relief.

  • Anonymous

    Deelip, I think Exalead is messing with you πŸ˜‰ My results from this URL:

    http://www.exalead.com/search/web/results/?q=catia

    are:

    1) http://www.3ds.com
    2) http://www.tgstech.com
    3) en.wikipedia.org/wiki/CATIA

    I still got the “Did you mean: katia?” question, but my results were much better.

    Something interesting to note is that your result set was 1.5 million and my result set is 2.4 million.

    Not that I think this is the problem, but I experienced something weird with Google a while back and learned that I had “personalized” search turned on, meaning that Google factored my search history into what results it returned in my current searches. That made my results very different from other people. I wrote a quick bit about it here:

    http://www.razorleaf.com/2010/06/when-search-gets-personal/

  • Neil

    They only paid $195m for it…another Bernard Charles impulse buy perhaps?

  • AlexS

    Deelip, my results (in Germany) are the Same like Jonathan’s…. Greetings from Munich…

  • John

    My results were also the same as Jonathan and AlexS.

    FYI – searching for “Deelip”, your site came up #2. Not sure why “www.nintynine.org” ranks #1 πŸ™‚

    • For me it shows deelip.com at the first result. ninetynive.org appears on the second page. I have absolutely no idea what that site has to do with my name. Some search engine this.

  • David Levin

    Deelip, I’ve got a strange feeling that Exalead somehow takes into account who is asking:). For example, when I ask “geometric solver”, it gives nothing but related to LEDAS – obviously to please me:). To see the name of a competitor, I have to enter it explicitly:)…

    • For me a search for “geometric solver” shows your press release at tenlinks as the first result. πŸ˜‰

  • Dominique

    Maybe it’s learning. I just search catia and 3ds.com came up first. Taking the deelip challenge and deelip.com is now the number one result.

    I remember working with the customer support colleagues at my previous employer, who used to say and probably still do, “if we cannot recreate the bug, then it doesn’t exist” πŸ™‚

    Keep up these great honest posts, thanks.

  • beyondplm

    I think, Exalead improved since the acquisition. In content processing, devil is in details. Bloated volume queries are meaningless if a system has no precise context. Google set a special page rank to few sites like Wikipedia. My hunch, Exalead is not doing it. I’d expect CATIA Wikipedia page should come before http://www.tgstech.com . In addition, their link on 3ds.com is incorrect. Search Google and you will see – #1 is the link on http://www.3ds.com/products/catia (best match) and #2 is Wikipedia – en.wikipedia.org/wiki/CATIA. #3 is IBM (:)) – make sense. #4 is image collection. I vote for Google results. In parallel, the precision of web queries is actually has no relation to the ability of the system to work inside of enterprise organization, in my view.

  • Mohan

    My search results coincide with that of Deelip. Searching for “3ds” http://www.3ds.com is nowhere. But when I search for PTC the results are perfect. the first result is http://www.ptc.com πŸ™‚

  • I too getting same results like Deelip, Katia!!! But when searching ‘SolidWorks’ I get SolidWorks official site at the top. think something wrong with Exalead when searching CATIA.

  • Anonymous

    Sachin, In my view, the problem with CATIA is that the level of noise (compared to DS CATIA meaning) is much higher. If you monitor “CATIA” on Google alert, you will notice the level of noise is more than 50%. The level of noise for CATIA on twitter is even higher.

    • But isn’t the job of a search engine to filter signal from noise? If that job is delegated to the user then what’s the point of the search engine? The link to jobvertise.com as the first result for a search for “CATIA” shows that something is horribly wrong here. I won’t even consider the jobvertise.com page to be noise. Its something more than that because CATIA does not figure anywhere on that page.

      • Anonymous

        Deelip, you are right, of course. However, it is not a simple job :). It took few years utill Google became as good as they are today.

        • Exalead is not a high school project created by amateurs. It was founded in 2000 by FranΓ§ois Bourdoncle and Patrice Bertin, both of whom were involved in the development of the Altavista search engine. How many more years do they need to learn?

          For whatever its worth, I searched for “catia” in Altavista and got the wikipedia page first and 3ds.com second. I’m telling you something is broken in Exalead. Seriously broken.