Hi all, does anyone know how the Google News algorithm work and can point me to a written paper on the subject? I am mostly interested in finding out if it's a different indexing method and different searching method, or is it the same methods used for the regular search engines with small adjustment to searching news? Thanks to anyone who can answer, and/or point to a written paper.
it's interesting. I was under the impression that every search engine work that way. but still, is the way google indexs news and search them the same way as regular internet search, or did they do a special news algorithm adjustment?
I think it's safe to say it's almost impossible to discover the algorithm. I don't need it. I just need to know if there's any written paper on it and if they just took the regular crawler and algorithm and adjust it to news or did they build somthing from skratch. btw in case anyone wonders - I'm writing a thesis about the subject
I would have to assume it is similar. The LSA angle was purchased from Applied Semantics (thus LSI) for use with AdSense/AdWords, is it in the core Search algo? Likely I would think. I could see 'possibly' having a unique indexing and retrieval algo for G News, but it's not a leap to assume much of the common underlying technologies are in place. If NOT, then we can also assume they merely 'focus' the core algo only on News Data documents. That sounds more likely for me as it is more resource friendly and maintenance effective... So regardless of it being a seperate algo (which I would personally doubt) the underlying technologies (G search patents et al) are likely in play What's the paper about?
do you know where maybe I can find a paper about it? my paper is about specific search engines. I want to prove (or disprove) that a specific type search engine that was built for a specific porpuse witch an algo aspecially for it, is better then a general search engine that was twicked for a specified field. I thought maybe I could find some papers that says google (or some other search engine) was twicking.
For me it is unlikely that any are necessarily ‘targeting’ a certain sector over the other. The nature of document indexing and retrieval for the web is too topic multi-dimensional to effectively ‘build’ an algo that way. Now could one ‘tweak’ the algo/filters to skew the results as such? Certainly. It is just not logical to BUILD the algo that way. Unfortunately most of my studies are in Google search technologies, though I do run around some MSN/Yahoo patents from time to time. As far as PROVING such an aspect, that is a lot of A/B ( multi-variant may be faster?) testing that certainly goes beyond the scope of this post…lol
If you're interested in research papers on page rank and search engine algos, searchengineland just posted a few from AIRweb (Adversarial Information Retrieval on the Web). Check it out here -- http://searchengineland.com/070415-133537.php most of it's over my head but interesting nonetheless... EDIT: sorry, didn't realize this was such an old thread, not sure why it was bumped...you've probably finished your paper by now, I would hope!