GloVe methods for semantic clustering

Question

After implementing GloVe (global vector) method to convert words to word vectors, how should one proceed ahead for semantic clustering?

Thanks in advance!!

Answer 1

I am currently working on clustering of Glove vectors with K-means and Fuzzy C-means algorithms. K-means shows some good results such that semantically similar words get grouped into one cluster. However, Fuzzy C-means seems not to work properly with these vectors such that when we cluster it using specified number of clusters, it does not produce expected number of clusters, probably it forces some clusters to overlap.

Regarding the number of clusters and performance evaluation, existing methodologies such as Elbow criteria for Kmeans and Fuzzy Partition Coefficient for Fuzzy C-means can be used to see what are the optimal number of clusters