Blekko's 'topspam' category initially includes MoneySavingExpert.com as 'spam' A new search engine called Blekko which promises only to show results from "trustworthy" sites has run into problems with two big UK websites, MoneySavingExpert.com and CompareTheMarket.com, which it thinks are spam-laden.
Rich Skrenta, Blekko's co-founder and chief executive, told the New York Times that since Google introduced its "reputation" system - in which the more links a site has, and the better the reputation of those sites, the higher it figures in search rankings - there has been a proliferation of sites that carry little relevant information, but rank highly because they use links and keywords to game Google.
"The goal is to clean up Web search and get all the spam out of it," Skrenta said.
Blekko searches throughout the "worthwhile web", but only shows the top results for any topic in what it calls "slashtags", and aims to eliminate so-called "content farms" which generate content based simply on what people appear to be searching for.
It also uses "vertical" search engines which focus on specific niches, rather than trying to cover the whole of the web as Google does.
"Google has a hard time telling whether two articles on the same topic are written by [content farm] Demand Media, which paid 50 cents for it, or whether a doctor wrote it," Tim Connors, founder of PivotNorth Capital and an investor in Blekko, told the New York Times. "Humans are pretty good at that."
But some humans in the UK are upset at the way that Blekko is describing their site. Its "topspam" list - in which it records the sites that it views as being spam, with the reasons why - starts with Yahoo's now-discontinued Geocities domain (which ranks top because of its size, and the fact that it cannot be crawled). Those take up the top two spots.
But at No.7 on the list - making it fifth in the ranking of real sites - is MoneySavingExpert.com, which Blekko's system says counts among the "top spam" because a Bayesian analysis of the content - probably the links to other sites or paid links within it - puts it above the threshold.
CompareTheMarket.com, the insurance comparison site known for its meerkat advertising and "Simples!" slogan, is ranked No.17 - fifteenth on the ranking if Geocities is ignored - also falls foul of Blekko's Bayesian spam analysis.
When contacted by The Guardian, MoneySavingExpert's web editor Dan Plant was not pleased by the classification: "Our site has 10 million unique users a month, no banner ads, is totally free to use, only sends the weekly email to users who have opted in (you can't do it by accident) and we never data mine or sell data. I think we're text book on how to operate the right way and not spam our users," he told the Guardian.
"Based on all that, we think it's farcical to include us on list of spam sites and quite possibly defamatory, though having said no one here had heard of this tiny search engine until you pointed it out to us. I'd say that if it continues to make such bizarre judgements about us and other sites, I doubt it will catch on!"
Skrentra responded: "Our automated systems have flagged those two sites as having a lot of "spammy" keywords. No doubt this is because they are in fact about money and personal finance! That is what the "Bayes" score on the right means.
"If these are in fact good sites - and I'm sure they may be - we could add them to our /money tag and use them to benefit all personal finance searches on blekko.
"But this kind of report from you is exactly why we are opening up all of the details about how blekko works to the public. Helpful reports that we are mis-labeling important sites with borderline spam scores as spammers. Now we can fix them!"
Skrentra added that this approach was a benefit of his site's willingness to open its workings up to users: "If the spam-score was kept secret, as it is in Google and [Microsoft's] Bing, you would just not see those sites come up as often as they otherwise might. By opening the details of how blekko works, we can steadily improve our relevance and results."
Plant said this could be positive: "If they are willing to change things based on experience that's good news. It sounds like the scoring system they are currently using is a bit blunt and arbitrary, but then what spam-scoring system isn't!"
CompareTheMarket had no comment at the time of writing.
Ahead of the launch of Blekko, Skentra was interviewed about how Blekko was aiming to be different from rivals: he suggested that it will be due to its social nature.
Rich Skrentra, chief executive and co-founder of Blekko, explains its rationale
Ingen kommentarer:
Legg inn en kommentar