Grammar Analysis...

[hometown]

Do you know where I can analyze the grammar of an input English sentence ? Could you tell me how such a machine works ?
Where can I find similar information or any tutorial for that ?

Thanks a lot for any help...
Regards,
-Nina

[hometown]

...

[IndikaNiva]

for the first part of you post
Do you know where I can analyze the grammar of an input English sentence ?

MS Word can do that
Tools ->language -> Thasurses

for the rest of your post I have no clue.

[hometown]

Originally posted by IndikaNiva
for the first part of you post
Do you know where I can analyze the grammar of an input English sentence ?
MS Word can do that
Tools ->language -> Thasurses
for the rest of your post I have no clue.

It is Okay, Thanks a lot for your answer though...:)

But do often use a translator ? I mean a translation machine to help you translate English into any other languages ? Do you know how such a machine works ?
I really appreciate any of your help...:)

Regards,
-Nina

[Joseph_R_Thomas]

you should try searching google

[hometown]

No, the OP and the previous one are two different posts....I didnot mean I was looking for a translation page but I would like to ask if there is someone here who can tell me how such a translator works...Thats the previous one. The Op is about sentence's grammar analysis, just like how to learn to analyze English grammar for ESL students....:) Yes, that is what I am looking for...

[chmanish]

If you want to analyze a sentence in English, what you need to develop is an Artificial Intelligence application.

See, analyzing any language requires 3 steps:
1) Syntactic analysis
2) Semantic analsis
3) Lexical analysis

You also need to form a set of rules(as they exist in english) and store them in the computer. (PROLOG is usually used for such applications).

The subject needs a great amount of study and learning. What I can suggest is that you should get yourself a good book on AI or Language Processing.

:D And don't depend upon Amazon for that

[hometown]

Originally posted by chmanish
If you want to analyze a sentence in English, what you need to develop is an Artificial Intelligence application.

See, analyzing any language requires 3 steps:
1) Syntactic analysis
2) Semantic analsis
3) Lexical analysis
You also need to form a set of rules(as they exist in english) and store them in the computer. (PROLOG is usually used for such applications).
The subject needs a great amount of study and learning. What I can suggest is that you should get yourself a good book on AI or Language Processing.
:D And don't depend upon Amazon for that

Thanks alot for your answer...
But do you have any link to take me there ? I dont have much money like software developers here, I have to save all of what I am given each day by my brother and I put them all in my account in the bank...It s been years but there is only $240, so could you help me ?

Regards,
-Nina

[chmanish]

You can visit the following:


http://research.microsoft.com/research/nlp/

http://nlp.stanford.edu/fsnlp/

http://www.cs.cmu.edu/afs/cs.cmu.edu/project/ai-repository/ai/html/faqs/ai/nlp/top.html



Search google for more info.

[Andreas Masur]

[Moved thread]

[hometown]

Originally posted by Andreas Masur
[Moved thread]
Thanks Andreas alot since you have taken care of Nina so more carefully than anyone here on board...*:)*
Oh,no, I think I should use this smiley :D
You have been working so hard that Nina sure you got an A+ for the test...(by Rogue)
[Just a joke]




chmanish:) Thank you very much for your links....

Regards,
-Nina

[Sam Hobbs]

One of the earliest samples of Natural Language, and the most popular, is "Eliza". So I searched for:+eliza +language +artificialAnd I got many results, many of which look useful. In the web site for the American Association for Artificial Intelligence (AAAI) (http://www.aaai.org) is their Natural Language (http://www.aaai.org/AITopics/html/natlang.html) page that looks very useful. I think generation5 - Artificial Intelligence Tour (Page 5) (http://www.generation5.org/content/1999/page04.asp) answers many of your questions. There is a lot in Computers Artificial Intelligence Natural Language Chatterbots Polish Yellow Pages (http://yellowpages.pl/ca/58115/Computers/Artificial_Intelligence/Natural_Language/Chatterbots) but I don't know how much is useful. In Homepage of Elizaneth, sister of Joseph Weizenbaum's Eliza (http://www.uib.no/People/hhiso/eliza.html) are links for:

a web page that has many versions of source code for Eliza
a web page where you can play with Eliza

You might enjoy reading How my program passed the Turing Test (http://www.compapp.dcu.ie/~humphrys/eliza.html).

Some other sites that are probably useful:

Open Directory - Computers: Artificial Intelligence: Natural Language (http://www.dmoz.org/Computers/Artificial_Intelligence/Natural_Language/)
Joseph Weizenbaum's ELIZA: Communications of the ACM, January 1966 (http://i5.nyu.edu/~mm64/x52.9265/january1966.html)
ELIZA (http://www-ai.ijs.si/eliza/eliza.html)

There are many more sites, it is just a matter of searching for them. You are very lucky that the internet has so many very useful resources. It was not long ago when, without the internet, there would have been so very little available to you.

[hometown]

Thank you Sam,...I do not know what to say really, you helped me so much...

Thanks...

Regards,
Nina

[Yves M]

Originally posted by hometown But do often use a translator ? I mean a translation machine to help you translate English into any other languages ? Do you know how such a machine works ?

chmanish only highlightes one view of NLP, but it's not the only one. There are two fundamentally different approaches to NLP (Natural Language Processing):

The analytical approach:
The programmers code their understanding of language into the program. This used to be the preferred way of approaching the problem. You basically have:
text > morphology (is this word a noun / verb etc) > grammatical analysis of sentences (constructing trees) > semantic relationships.
This is the method that is used most of the time in MT (machine translation). The results are very good for sample sentences, medium to good for structered text in a given domain (like law e.g.), but very erratic on "free" text where sentences may be incomplete, not very well structured in the first place or ambigous. IMHO the analytical method has hit a dead end and as a pure solution cannot get much better.
The statistical approach:
The programmer does not code their understanding of language into the program. The program gathers this information from a (usually quite large) sample of text. Here you have:
sample text > statistical analysis > internal representation
new text > statistical analysis > comparison with sample > result.
This is what EBMT (Example-based machine translation) is using. Take millions of sentences in english for which you have (good, since done by hand) translations sentence to sentence into french. Using this information, an EBMT system can then translate some new text from english into french.

The fun thing about this is that the entire problem cannot be solved by a computer in theory. I'm on a mailing list of professional translators and many times a translation question comes up over which people argue endlessly. The point is that many times there is no single correct answer and many other times, what seems to be correct actually isn't.

To finish this off, here is one example of a sentence that is impossible to analyze with only one meaning.

I see the man on the hill with the telescope.
What is it ? Is the man on the hill holding a telescope and I'm watching him ? Is it a hill with a telescope and I see a man standing on that hill ? Is it just a man on a hill and I'm watching him through my telescope ?

[Yves M]

Actually, just to highlight the ambiguity in english, here is another example of a problem. When you analyze a sentence to find out what its structure is, you need to decide whether a given word is a noun, verb, adjective, adverb, determiner...

Sounds simple, doesn't it ? Now let's take a well-known word like "barbecue". What is it ?

a noun: "I have invited our neighbours for the barbecue tomorrow"
a verb: "Tomorrow, I will barbecue some chicken"
an adjectiv: "We will eat the chicken with barbecue sauce".

[Sam Hobbs]

Originally posted by Yves M
Sounds simple, doesn't it ? Now let's take a well-known word like "barbecue". What is it ?Looking at a dictionary, "barbecue" can be a noun or a verb. I don't see a definition for it as an adjective. So according to that dictionary, "barbecue" is either a noun or a verb; it is not possible to say which one without a context.

Are you sure that "barbecue" in "barbecue sauce" is an adjective? I did not pay attention to grammar when I was in school so I don't know as much as you do but I thought that "barbecue sauce" is a name.

You say "ambiguity in english"; is english the only ambiguous language?

About 15 years ago someone with a degree in language designed a computer program that read in COBOL programs and then logically (not just physically) restructed them so that they were easier to understand. The relevance is that his training about language and parsing of language helped him to design computer language parsing also. In that situation, the parsing had to do more than a compiler did, since in this situation the parsing had to be able to determine what result of the instructions would be; it had to understand a sequence of instructions and then to be able to rewrite those instructions. It might seem easy enough but if so then I am just not explaining it very well. It might be the equivalent of reading a C program and creating a C++ program with classes when appropriate. The challenge would be to be able to identify what is an object and to be able to create a class when one does not exist.

The COBOL restructuring tool had only one other competing with it, one written by IBM. I don't know how much the IBM one cost but the non-IBM one cost (was being sold for) $100,000 per copy. I am sure there is ample opportunity remaining for language parsing and natural language stuff and such.

[Yves M]

Originally posted by Sam Hobbs
Looking at a dictionary, "barbecue" can be a noun or a verb. I don't see a definition for it as an adjective. So according to that dictionary, "barbecue" is either a noun or a verb; it is not possible to say which one without a context.

Are you sure that "barbecue" in "barbecue sauce" is an adjective? I did not pay attention to grammar when I was in school so I don't know as much as you do but I thought that "barbecue sauce" is a name.

Indeed, the dictionary gives only noun and verb definitions and my example is not well-chosen. There is the added difficulty that the function of a word may be different from its category. This means that something which is unequivocally a noun for example, can be used in a sentence as some other thing. I'll have to see whether I can come up with a good example of this.


You say "ambiguity in english"; is english the only ambiguous language?

Of course not. All human languages contain some ambiguity. Ambiguity depends on technical reasons like lack of inflections (as in English), lack of adequate words, ambiguous sentence construction (the telescope example), pronouns (like "he", "she", "it"). and probably some other things.

Also, ambiguity is different in each language. I don't remeber quite how many different words Inuits have for snow, but it's more than ten. If an english person says "snow" it sounds ambiguous to an Inuit.


About 15 years ago someone with a degree in language designed a computer program that read in COBOL programs and then logically (not just physically) restructed them so that they were easier to understand. The relevance is that his training about language and parsing of language helped him to design computer language parsing also. In that situation, the parsing had to do more than a compiler did, since in this situation the parsing had to be able to determine what result of the instructions would be; it had to understand a sequence of instructions and then to be able to rewrite those instructions. It might seem easy enough but if so then I am just not explaining it very well. It might be the equivalent of reading a C program and creating a C++ program with classes when appropriate. The challenge would be to be able to identify what is an object and to be able to create a class when one does not exist.

The COBOL restructuring tool had only one other competing with it, one written by IBM. I don't know how much the IBM one cost but the non-IBM one cost (was being sold for) $100,000 per copy. I am sure there is ample opportunity remaining for language parsing and natural language stuff and such.
It's a problem that's indeed very similar to what analytical NLP people have been doing. Trying to analyze "language" by constructing an internal representation and map the structures to some meaning. Then remap that meaning to some other "language". For computer languages this is of course much easier than for human languages, but even so, it can still be very difficult.

[hometown]

Thanks Yves M so very much for your information....
I amnot a native, so I actually dont understand what is barbecue sauce ? May I ask for another sound explanation from you YvesM ? :).
Because I rarely go out having barbecue with my friends...I am like a stupid about that kind of sauce...and I have to honestly admit that from my childhood I have not learnt anything about sauce...If I amnot mistaken, YvesM know alot right ? Could you teach me now ?

I would much appreciate if You, YvesM could give me some more help on that....

Thank you... :)

Regards,

-Nina

[Sam Hobbs]

Barbecue sauce is made of tomatoes, vinegar and spices.

[hometown]

Originally posted by Sam Hobbs
Barbecue sauce is made of tomatoes, vinegar and spices.
Okay....:)

Thanks a lot...:)

Regards,
-Nina