Hi everyone, need help from developers with a news article clustering problem I am Working on.

Hi everyone, need help with a news article clustering problem I am Working on. The performance measure to be used is silhouette score... So far I have tried  tf-idf and gensim doc2vec to convert the documents to learnable  features and then used kmean and hierarchical clustering technique to cluster the docs... The max score I am able to achieve is around 0.13 . Given that the silhouette score range is between -1 to 1, this is low score. Is their anything I can try to improve this?  Any help would be greatly appreciated.