Dear Sergio,

You can use Davies-bouldin (DB) index for hard clustering, and Xie-beni (XB) and Partition (SC) indices for soft clustering quality assessment.

Brief description and several examples of the mentioned indices can be found in the following article:

http://www.sciencedirect.com/science/article/pii/S0924271614002056

Furthermore, you could easily find their source codes for implementation in MATLAB application.

Bests,

Saman.

Dear Friend

                I think that purity measure would be a good choice. But its limitation is the trade-off measure; which could be mitigated using normalized mutual information and the trivial accuracy measure.

You can use FBcubed, compared with other metrics FBcubed satisfies a serie of constraints (Cluster Homogeneity, Cluster Completeness, Rag Bag, and Clusters size vs. quantity), the implementation is not difficult.

Here is the link of the paper:  http://nlp.uned.es/docs/amigo2007a.pdf