FRIHOSTFORUMSSEARCHFAQTOSBLOGSCOMPETITIONS
You are invited to Log in or Register a free Frihost Account!


A morphosyntactical analyzer





ratanegra
When questioned about what my final papers would be about, I automatically thought of mixing both of my knowledge cores: computers and languages. This was the result.

As you know, there are many kinds of softwares to analyze text, but they are all very simple, since the possibilities of combinations of words can be infinite, especially since words are invented everyday. Nonetheless, I thought of a solution to that: through analyzing the context of each word, the frequency in which it is used, dictionary definitions and translations and it's morphological elements (the letters it contains can indicate the kind of word that it is, like for example, if it has an -ing ending, it is a verb or derivates from a verb). There are many other functionalities I have thought for it, but right now I can't remember them all. But, overall, I think I will be able to make it happen when I learn how to program.
fuzzkaizer
so maybe you will have to feed your machine with sort of a 'dictionary of a words contexts'? i think there even exist some studies of linguistics, but only for some words, and maybe not exaustivley... might be allready quite a project to set up such a database? compared to that frequency is maybe a more viable factor to trace the meaning of a word by computer. still some linguistic research necessary.
how do the usual softwares work? do they combine a several of such probabilities?
i am curious how you would manage things like creative use of language, irony, ambiguity and double sense etc. this might be the most tricky part of the project. keep the thread informed!
ratanegra
fuzzkaizer wrote:
so maybe you will have to feed your machine with sort of a 'dictionary of a words contexts'? i think there even exist some studies of linguistics, but only for some words, and maybe not exaustivley... might be allready quite a project to set up such a database? compared to that frequency is maybe a more viable factor to trace the meaning of a word by computer. still some linguistic research necessary.
how do the usual softwares work? do they combine a several of such probabilities?
i am curious how you would manage things like creative use of language, irony, ambiguity and double sense etc. this might be the most tricky part of the project. keep the thread informed!


I had not thought of that last part, to be sincere. But I have studied linguistics and it is possible to put a category on each word. Some categories are controversial between schools of thought, but I could just choose one from all the possibilities. This would not be a one-man project, since that would be extremely difficult. I would set up the bases and let the language community contribute when they are doing their own research in the field.

I don't know much about softwares, so I cannot tell you how they usually work. However, algorithms, as far as I understand them, work with probabilities. If I could get the help of someone who knew about this, it could help with this side of the program. And if the software worked with probabilities, I assume it would work with several of them at the same time. There is no doubt about that.

And now thinking about that last part, I think that probing the context of each phrase, the program would be able to detect, at least on a minimum level, the lack of correspondence between each particle and its context. As I understand this multifunctional tool, it would be able to understand the meaning of words, sentences, paragraphs and texts based on mathematics and the use of dictionaries.

This last point comes from a net of word meanings that was done by the team of WordNet. I would probably make use of it for the development of the algorithms that, with the help of connectors (such as prepositions, conjunctions, adverbs, etc.), would help the program understand the relation between each word.
coolclay
Wow, this sounds like it could be awesome if it ever comes to fruition. It is a very daunting project however.

My analysis on the difficulty of this project hinges on the fact that Google can't even get it's voice recognition software even remotely accurate. I use Google voice as my primary means of telephone conversation. They attempt to convert my voicemails to text, but I have yet to see one that is even close to correct. It gives a good source of entertainment however.

So good luck with the project if successful it has the ability to make a lot of money and change the way voice recognition works, but it will be a ton of work!
Related topics
A "small" list of free apps
How to Protect ur windows System
Need comments for my site
Does anyone know this???
site submission for free
Master Tutorial on SEO
Javascript rapid picture change
Of Diodes, Capacitors, and No Child Left Behind
Robot.txt and Meta tags
my best flash site!!!
how to do SEO?
Biowindows® Xp Vision™ DVD Svcpack 4 XXL [porjoderunrato]
I'll change the world, and so will you.
Great free photo editor
Reply to topic    Frihost Forum Index -> Lifestyle and News -> Amazing Projects

FRIHOST HOME | FAQ | TOS | ABOUT US | CONTACT US | SITE MAP
© 2005-2011 Frihost, forums powered by phpBB.