"Finding Top relevant document in kmeans cluster"

amir_askary_sha
amir_askary_sha New Altair Community Member
edited November 5 in Community Q&A

Hi,

 

After running kmeans clustering, how can I find out which document is the most relevant (top document) in one cluster?

 

Right now the documents in a cluster are sorted ascendingly by their id. I want to have them sorted by a weight score showing how relevant this document is in this cluster, or at least to see the most relevant doc in the cluster.

Answers

  • MartinLiebig
    MartinLiebig
    Altair Employee

    Hi,

     

    how do you define relevancy?

     

    Best,

    Martin

  • amir_askary_sha
    amir_askary_sha New Altair Community Member

    I don't know exactly; any kind of relevancy. For example let's say every cluster has some top words in it (the centroids that kmeans finds), and then the document which has the shortest cosine/euclidian distance to those top words of the cluster, is the most relevant doc in the cluster.