# Dirichlet distribution
**URL:** https://discourse.pymc.io/t/dirichlet-distribution/7411
**Category:** Questions
**Created:** [May 8, 2021, 6:36am UTC](https://discourse.pymc.io/t/dirichlet-distribution/7411 "2021-05-08T06:36:52Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@800635753](https://discourse.pymc.io/u/800635753)
#### Post date: [May 8, 2021, 6:36am UTC](https://discourse.pymc.io/t/dirichlet-distribution/7411/1 "2021-05-08T06:36:52Z")
Hi,
I have a directed graph with the label. can you explain to me how to compute Prior Dirichlet for each node? I need to use the Prior Dirichlet in MAP. I keen to use it in my work.
---
### Author:  [@twiecki](https://discourse.pymc.io/u/twiecki)
#### Post date: [May 11, 2021, 9:13am UTC](https://discourse.pymc.io/t/dirichlet-distribution/7411/2 "2021-05-11T09:13:36Z")
Unfortunately I don’t understand what it is you’re trying to do and what your issue is.
---
### Author:  [@800635753](https://discourse.pymc.io/u/800635753)
#### Post date: [May 12, 2021, 6:59am UTC](https://discourse.pymc.io/t/dirichlet-distribution/7411/3 "2021-05-12T06:59:20Z")
I will use a GMM model in my work and I am going to use MAP for linkprediction in GMM model. the MAP is prior\*likelihood and i compute Dirichlet for prior and likelihood. i am doing this beacause last papers only count the number of edges and it leads to eliminate some of nodes.
is it possible to use Dirichlet just for Prior?? can i use a kernel similarity for likelihood?? I know that if i use this method, the type of prior and likelihood are not same, but I am not sure it work truly for MAP.