NLTK and Mallet

I am quite impressed by the available open source NLP softwares. Recently, I have used two of them after playing with most the major softwares. I like NLTK because it is in python and hence it is easy to use and it has a lot of utilities. Mallet is the other one I like. It's in java and it seems to have good sequence modeling tools such as FSM, HMM and MRF etc. It's fantastic that there are also robust parser available in both python and in java. I remember that I wrote a HTML parser a long back after getting completely frustrated by mal formed htmls while working on a project. Today I found Jericho (java) and BeautifulSoup both work wonderfully for bad HTML data. However, the core theoretical NLP problem is still very hard because human languages are quite ambiguous and humen communication system is quite buggy (actually I should say it is high performance distributed and fault tolerant system and languages are inherently, intentionally vague) than a computer program. It is amazing that human can understand the meaning of a sentence in a language like English by seeing a few context words only.

Being efficient at Work place

1. Find places to work: home, the library, the coffee shop, conference room whatever.
   
2. Say “no” to things. Many of us are abysmally poor at this, but eventually it becomes a necessity. Figuring out when to say “yes” vs. “no” is hard.
3. Don’t be afraid to do a bad job on things.
   
4. Find ways to keep work fun. Pick up new languages, read what top academicians are talking about, attend conferences, look at the problem sets given in the lecture notes by the best professors in an area.
   
5. Stay in shape.
   
6. Keep doing projects that has at least a little non-triviality. Work out the math. If you do not know the math, read it. If the math requires a background, ignore that and get the intuition you need to understand.
   
7. Stop caring about things that you are not working on.
   
8. Do not waste your energy to argue with a fool.
9. Do not attend a meeting if it is not necessary. Be strict on this.
10. Publish if possible.
    

A beautiful book by Jeff Ullman

I found that Jeff Ullman recently wrote a book on foundation of computer science. It is a great book pointing the significance of theoretical aspects of computer science with respect to the real problems.

Bing shopping

Bing shopping search seems to be sadly poor. They have a lot more ads than google. Their UI looks dull. Search results have a bias towards lower price. Product images can be ugly. They dont seem to apply a successful query understanding module for its shopping search.