I respectfully disagree. You write: "if you know what you want to search for, you search for it". This description is obviously adequate for only *one* sort of web searching, which is rather unexploratory.
But, in many cases search engines are useful in that they allow us to explore knowledge domains we were not aware of. Suggesting potentially related terms enables exactly this sort of searching.
Don't think of it as search for information you already know you want. Think of it as being exposed to information you are not aware of yet (I, e.g., don't underestimate my own ignorance).
Here's an example, an undergrad student is searching for "belief revision". He finished clicking belief, and "belief net" pops up from google. Sounds related. Great, he makes a mental note to check it out, and is introduced to belief revision models based on Bayesian Reasoning.
TO sum, I find the suggestion box helpful, and would think it would be even more helpful if suggestions were based on terms exctracted from a semantically-related thesauri (e.g., Wordnet) rather than being just lexically related.
I agree that abandonware should be archived and publiclly shared. Here's a prime example: Alta-Vista Personal search engine, which was developed in 1997, and worked on win89 and win2000 flawlessly (until a win2K bug forced me to index manually) -- it was pretty quick, and did a better job than google's current desktop search. It also didn't run constantly in the background, but updated it's index at preset times.
Finding altavista personal search is almost impossible now (and it won't run on XP), but it's still useful for other WIN OSs.
FOr those interested: you can get it here
Right, this is called the McGurk effect, and has been known since.. hmm. the mid 70's.
Catch a demo here
http://www.media.uio.no/personer/arntm/McGurk_engl ish.html
As someone who actually read the entire article, I can attest it can really pass a 1.5 hour flight. It *might* also be interesting reading for those interested in some cutting edge child research methods such as ERP electrophysiology for kids.
What's not clear to me is the value in Slashdot putting up a pointer to an article that can only be read with subscription service that costs an arm and a leg, and is usually only freely available only to lucky folks in the.EDU domains.
Finally, let me drop my 2 cents on the original posting that cited the paper as saying about infants: "They also learn phonotactic rules".
This statement is phrased rather loosely. Just because infants' behavior indicates that they can determine whether stimuli correspond or do not correspond to a rule certainly does not mean that the mental representation system that afforded this discrimination actually works by representing anything akin to rules.
You don't need a rule-based system to be able to determine whether a certain input corresponds or doesn't correspond to a set of constraints (see the classical debates between Pinker and McLelland on the acquisition of the past-tense in English).
Saying that infants learn "rules" is therefore a bit misleading.
Slashdot Mirror
I respectfully disagree. You write: "if you know what you want to search for, you search for it". This description is obviously adequate for only *one* sort of web searching, which is rather unexploratory.
But, in many cases search engines are useful in that they allow us to explore knowledge domains we were not aware of. Suggesting potentially related terms enables exactly this sort of searching.
Don't think of it as search for information you already know you want. Think of it as being exposed to information you are not aware of yet (I, e.g., don't underestimate my own ignorance).
Here's an example, an undergrad student is searching for "belief revision". He finished clicking belief, and "belief net" pops up from google. Sounds related. Great, he makes a mental note to check it out, and is introduced to belief revision models based on Bayesian Reasoning.
TO sum, I find the suggestion box helpful, and would think it would be even more helpful if suggestions were based on terms exctracted from a semantically-related thesauri (e.g., Wordnet) rather than being just lexically related.
I agree that abandonware should be archived and publiclly shared. Here's a prime example: Alta-Vista Personal search engine, which was developed in 1997, and worked on win89 and win2000 flawlessly (until a win2K bug forced me to index manually) -- it was pretty quick, and did a better job than google's current desktop search. It also didn't run constantly in the background, but updated it's index at preset times.
Finding altavista personal search is almost impossible now (and it won't run on XP), but it's still useful for other WIN OSs. FOr those interested: you can get it here
Right, this is called the McGurk effect, and has been known since.. hmm. the mid 70's. Catch a demo here http://www.media.uio.no/personer/arntm/McGurk_engl ish.html
As someone who actually read the entire article, I can attest it can really pass a 1.5 hour flight. It *might* also be interesting reading for those interested in some cutting edge child research methods such as ERP electrophysiology for kids. .EDU domains.
What's not clear to me is the value in Slashdot putting up a pointer to an article that can only be read with subscription service that costs an arm and a leg, and is usually only freely available only to lucky folks in the
Finally, let me drop my 2 cents on the original posting that cited the paper as saying about infants: "They also learn phonotactic rules".
This statement is phrased rather loosely. Just because infants' behavior indicates that they can determine whether stimuli correspond or do not correspond to a rule certainly does not mean that the mental representation system that afforded this discrimination actually works by representing anything akin to rules.
You don't need a rule-based system to be able to determine whether a certain input corresponds or doesn't correspond to a set of constraints (see the classical debates between Pinker and McLelland on the acquisition of the past-tense in English).
Saying that infants learn "rules" is therefore a bit misleading.