3 ms·

I think a major breakthrough will occur when we ditch the whole "meaning" thing. I think humans probably use language the same way a computer uses numbers, i.e

I think a major breakthrough will occur when we ditch the whole "meaning" thing. I think humans probably use language the same way a computer uses numbers, i.e. manipulation of abstract symbols. Words, like numbers, are ultimately just empty buckets with a bunch of tags attached that tell us something about how it interacts with the world. Our tagging system is very complex, and I suspect that much of human creativity stems from experimentally assigning inappropriate tags to concepts and seeing if the result makes a more coherent pattern.

timcederman 18y ago
Our search company (Trovix) assigns meaning to words. Google does not. They use statistical analysis to derive contextual links.
- MissBrackets 18y ago
  I agree with you on that one.
slackenerny 18y ago
I don't want to get overly philosophical here on breakthrouh happening or not but I too think language structure may not be tied to or convey any particular meaning. Quine explored this much on philosophical grounds and then there is poetry, a craft of aligning text to our inner brain cache buffers so that it gives a pleasant read e.g. by the means of rythm. What I mean is not that semantic analysis is futile, but that it may simply not be intristically fundamental to the language. For example in this comment on basics of statistical machine translation http://michaelnielsen.org/blog/?p=577 http://michaelnielsen.org/blog/?p=577 Gil Kalai expresses his astonishment at the results despite himself being a noted statistician. On the other hand I recently bought a book of poems, side work of a translator, who showcased therein his skills at morphing his language with the goal set at scrambling it badly - striping out of any flow and rythm and obeying grammar only in most pathological ways, still the poems are beautiful and meaningful.

timcederman 18y ago

Our search company (Trovix) assigns meaning to words. Google does not. They use statistical analysis to derive contextual links.

MissBrackets 18y ago
I agree with you on that one.

slackenerny 18y ago

I don't want to get overly philosophical here on breakthrouh happening or not but I too think language structure may not be tied to or convey any particular meaning. Quine explored this much on philosophical grounds and then there is poetry, a craft of aligning text to our inner brain cache buffers so that it gives a pleasant read e.g. by the means of rythm. What I mean is not that semantic analysis is futile, but that it may simply not be intristically fundamental to the language. For example in this comment on basics of statistical machine translation http://michaelnielsen.org/blog/?p=577 http://michaelnielsen.org/blog/?p=577 Gil Kalai expresses his astonishment at the results despite himself being a noted statistician. On the other hand I recently bought a book of poems, side work of a translator, who showcased therein his skills at morphing his language with the goal set at scrambling it badly - striping out of any flow and rythm and obeying grammar only in most pathological ways, still the poems are beautiful and meaningful.