Tuesday, October 31, 2006

Collaborative Filtering

Directory of Related papers

TASTE
Wikipedia
WEKA

rashmi sinha's papers/ blog
jun wangs collaborative filtering resources

COLLABORATIVE FILTERING FOR USENET NEWS


search google for : "collaborative filtering" taste

Things I need to know more about CF.

1. Bayesian text classifier
2. Collaborative filtering. Neighbourhood based algorithms (KNN?)
3. Pearson Correlation

Sunday, October 15, 2006

The Chain Game and its implications in Managing Companies

I just had an outing with the children of Vishwas (an NGO) at lalbagh, bangalore. We played a couple of games. The children were bubbling with enthusiasm throughout..

One of the games that we played was the Chain Game. Let me describe the game for you..
1. One player is chosen as the initiator of the chain [The nucleus]
2. This nucleus has to catch the other players who are the part of the game.
3. Once he gets hold of someone, the victim and the nucleus have to form a chain and go after the remaining people.
4. This continues till only one player is left [he is the winner]

I observed that:
1. Smaller the size of the chain, the better the
a] Agility
b] Co-ordination
c] Swift reaction to changes in the external environment.

As the chain grows, the internal friction increases, coherence decreases, conflicts of interest arise, the system becomes clumsy [externally]..
Although there is lots of energy in the system in terms of [force]energy of individual components, the forces of the components tend to nullify one another...

Thats what happens to big conglomerates like GE, Intel.. and this may also happen to Google in may be 50 years or so..

Ideal strategy would be breaking the link.. This will create multiple nuclei which will act independantly of each other but work towards the common goal.. Capturing the entire market [or killing out the competition]..

Something similar to the Agent in Matrix [the virus]..
What do I name this technique.... Viral capture?