Regarding the Dirichlet distribution

In 5th chapter of Mining Text Data [http://rd.springer.com/chapter/10.1007%2F978-1-4614-3223-4_5#] it is mentioned that the Dirichlet distribution has following property:
The Dirichlet distribution favors imbalanced multinomial distributions, where most of the probability mass is concentrated on a small number of values. As a result, it is well suited for models that reflect commonly observed power law distributions in human language.

I am not able to understand how does the Dirichlet distribution reflects power law distributions in human language.
Request you to elaborate the connection between the Dirichlet distribution and human language.

I also request you to provide me intuition behind the use of Dirichlet distribution in LDA.
Please also give me pointers to appropriate resources.

lda dirichlet

asked May 27 '13 at 05:25

swapnilhingmire
31●15●15●16

One Answer:

oldest newest most voted

1	I believe this sentence is false. DIrichlet distributions do not obey power laws. There is, however, an adaptation of a dirichlet process which can be made to obey a power law, the Pitman-Yor process. answered May 27 '13 at 08:14 Alexandre Passos ♦ 25541●54●278●421

Subscription:

Once you sign in you will be able to subscribe for any updates here

Tags:

lda ×114
dirichlet ×26

Asked: May 27 '13 at 05:25

Seen: 647 times

Last updated: May 27 '13 at 08:15

Regarding the Dirichlet distribution

Subscription:

Related questions