4 ms·

To do research in natural language processing. Roadblocks: * Time. Or if time = money then I need money to be able to spend time on this. Also, in itself this

To do research in natural language processing. Roadblocks: * Time. Or if time = money then I need money to be able to spend time on this. Also, in itself this wouldn't bring in much income I don't think. * No CS background. Don't know where to start.

ananthrk 16y ago
Three good books that can get you started: 1) Natural Language Understanding - James Allen 2) Speech and Natural Language Processing - Jurafsky & Martin 3) Statistical Foundations of Natural Language Processing - Manning You can brush up your probability and Linear Algebra from the prep guides available from Andrew Ng's machine learning course. My best wishes.
- pchristensen 16y ago
  I have copies of 2) Speech and Natural Language Processing - Jurafsky & Martin 3) Statistical Foundations of Natural Language Processing - Manning that I'm willing to sell. I tried to bite off more than I could chew and moved on to a different project. peter at pchristensen dot com
thestoicattack 16y ago
What part of NLP? It's actually a pretty wide field.
- felideon 16y ago
  Not exactly sure, but after skimming through Wikipedia, natural language generation sounds pretty interesting. But text-to-speech and speech recognition is probably what I've had in mind.
  - thestoicattack 16y ago
    Conisder "Statistical Methods for Speech Recognition" by Fred Jelinek. Lots of math, not so much CS, if that'll make things easier for you.
robryan 16y ago
You will probably want to start with CS concepts, given that from what I have seen NLP is a very algorithmic area of research, having some CS/math background will help you a lot to understand what has already been done and to build on it. Now I'm not sure the best way to go about this, either learn the abstract concepts which might not be the best way to start or pickup one of the more pure languages which in turn will teach you concepts when you learn how they are implemented in the language.
- felideon 16y ago
  Thanks. I'm assuming Common Lisp is a suitable language? Not sure which ones you would consider pure, other than Scheme or Haskell.
  - robryan 16y ago
    Not pure as such, just something you can get the concepts from without to much language specific stuff or at to high a level. Don't think I'm the best person to suggest though having started with PHP in my teens, then only really had my mind opened up to proper CS when I got to uni.
weirdwonka 16y ago
Some good advice 1. use python + NLTK 2. read papers and posters from current conferecnce such as ACL or even WWW 3. start an incremental project and try at least getting a poster published in a tier-1 conference 4. If you have good grades + work ex go to CMU Advanced Lang. Tech Masters program

ananthrk 16y ago

Three good books that can get you started: 1) Natural Language Understanding - James Allen 2) Speech and Natural Language Processing - Jurafsky & Martin 3) Statistical Foundations of Natural Language Processing - Manning You can brush up your probability and Linear Algebra from the prep guides available from Andrew Ng's machine learning course. My best wishes.

pchristensen 16y ago
I have copies of 2) Speech and Natural Language Processing - Jurafsky & Martin 3) Statistical Foundations of Natural Language Processing - Manning that I'm willing to sell. I tried to bite off more than I could chew and moved on to a different project. peter at pchristensen dot com

thestoicattack 16y ago

What part of NLP? It's actually a pretty wide field.

felideon 16y ago
Not exactly sure, but after skimming through Wikipedia, natural language generation sounds pretty interesting. But text-to-speech and speech recognition is probably what I've had in mind.
- thestoicattack 16y ago
  Conisder "Statistical Methods for Speech Recognition" by Fred Jelinek. Lots of math, not so much CS, if that'll make things easier for you.

robryan 16y ago

You will probably want to start with CS concepts, given that from what I have seen NLP is a very algorithmic area of research, having some CS/math background will help you a lot to understand what has already been done and to build on it. Now I'm not sure the best way to go about this, either learn the abstract concepts which might not be the best way to start or pickup one of the more pure languages which in turn will teach you concepts when you learn how they are implemented in the language.

felideon 16y ago
Thanks. I'm assuming Common Lisp is a suitable language? Not sure which ones you would consider pure, other than Scheme or Haskell.
- robryan 16y ago
  Not pure as such, just something you can get the concepts from without to much language specific stuff or at to high a level. Don't think I'm the best person to suggest though having started with PHP in my teens, then only really had my mind opened up to proper CS when I got to uni.

weirdwonka 16y ago

Some good advice 1. use python + NLTK 2. read papers and posters from current conferecnce such as ACL or even WWW 3. start an incremental project and try at least getting a poster published in a tier-1 conference 4. If you have good grades + work ex go to CMU Advanced Lang. Tech Masters program