-1
$\begingroup$

Al-Mubaid et al. proposed a semantic similarity measure in their research paper [1]. They see ontologies as connected graphs but refer to clusters within ontology graphs without ever defining what they are.

I am not sure what they meant with clusters.

This is an example of two clusters that Al-Mubaid et al. presented in their paper. The nodes starting with 'a' are part of cluster A, whereas those starting with 'b' are part of cluster B. enter image description here

There are two possible ways to interpret the definition of clusters:

  1. The clusters are separate ontologies.
  2. every root node's branch forms an independent cluster?

[1] Al-Mubaid H, Nguyen HA. A cluster-based approach for semantic similarity in the biomedical domain. Conf Proc IEEE Eng Med Biol Soc. 2006;2006:2713-2717. doi:10.1109/IEMBS.2006.259235

Improve this question
$\endgroup$
1
  • $\begingroup$ Clusters are indeed the branches of the root node. Nevermind the community answers. $\endgroup$
    – AzLimbiate
    Commented Aug 14, 2020 at 20:37

1 Answer 1

Reset to default
0
$\begingroup$

There is a whole study of hierarchical clustering. You start with a discrete set of nodes and iteratively connect the ones that are closest according to some similarity measure.

The branches of the root node will correspond to the largest clusters but they may have subclusters corresponding to other subtrees that may be of more interest and are more clearly defined as meaningful separate clusters.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Okie thanks for sharing that. However, the research paper doesn't show what dissimilarity measure was used to form such clusters. It takes the Snomed-CT/MeSH ontology and assumes that it's formed of clusters. Then, it uses its semantic similarity measure on pairs of concepts and compares that with points given by physicians. Moreover, knowing how the clusters were formed would important too (they don't mention anything about it). Then again, it's counterintuitive to form clusters using some similarity measure on which you will test your own similarity measure. $\endgroup$
    – AzLimbiate
    Commented Aug 17, 2020 at 2:12
  • $\begingroup$ In other words, it's unlikely that they formed clusters. $\endgroup$
    – AzLimbiate
    Commented Aug 17, 2020 at 2:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.