[[ Following up on an older post of mine – Why Search is still poor ]]
Somebody searched for this, “music from dx that people can listen“, and got to my blog from the first offered link. Damn!
Popular search engines don’t have much in terms of language skills. If it does not know something, it should admit it. How about, “I can’t understand what dx is. Can you explain what dx is?”. (i.e., the human computation approach can be used with human users feeding back into the system)
Alternatively, it can use the wikipedia which is an offline and continous feedback. See this page at the wiki with all possible disambiguations of ‘dx’. It can query the Wikipedia! This struck me sometime back (hehe) and when I was searching for it, I came across this paper WikiRelate! Computing Semantic Relatedness Using Wikipedia.
As for me, I do not know any ‘dx’ other than the ‘dx’ of my blog. 😉
The challenge of continously changing language (you know, parlance) will also be met by the constantly changing wikipedia. The real problem will be the reliability of the search engine. Maybe, search companies can have additional screening of the wiki. But practical implementation of this would mean search engine servers accessing the wikipedia (incase of realtime result generation) which would entail additional internet traffic. Ofcourse, companies would have a local index of the wiki on their servers and get frequent updates. Time shall tell.
The same query on Yahoo, yielded reasonably better results. Looks like ‘DX’ in that query referred to a music group. That looks like tagging at work. You see, the human feedback! 😉 I don’t know if Yahoo is using Yahoo Answers feedback for its search results. Might be worth a shot.
After seeing the above, I am forced to say this might be a bias in Google’s algorithm. Hmm…