ADOTAS – I’m doubtful that an algorithm can judge something as subjective as quality content, but an algorithm can certainly be used to scrape off the obvious crap (perhaps the “low-clinging crap”?).
Through AdSpam, a new machine-learning algorithm that examines multiple spam signals including page content, Blekko has labeled 1.1 million domains spam and banned them from the results of its social search engine. Last month Blekko booted the top 20 content farms determined by their slash-tag concocting user base — these included Expert Village, eHow and ChaCha.
The AdSpam technology will be used to evaluate page before it can appear in Blekko search results. Interestingly, Blekko notes that a solid sign that a page is junk is “aggressive participation in self-service online advertising networks” — pages with little content and a slew of self-service ads get the worst marks from AdSpam’s algorithm.
“With this update, we’re reaching much further down the long tail of spam to improve the quality of our index,” CEO Rich Skrenta said. “We’re combining algorithms with community to improve the future of web search.”
Since the launch of its slash-tag powered, human-curated social search engine last November, Blekko has introduced the Spam Clock — which led to the entertaining and pyro-rific Burning Spam celebration — a mobile app and publisher tools. The company also teamed up with programming Q&A resource Stack Overflow, which recently changed its name to Stack Exchange and garnered $12 million in funding.