0

I am new to carrot2 here so please bear with me. I am using carrot2workbench 32-bit in order to use Aduna cluster map. It's helpful to see the relationships among the clusters, but how do you identify the documents in an overlap (merged bubble) between the sets. For example, if I have 1 cluster of 20 docs and another 30 docs, and the 5 in the overlap, how to find out what these 5 docs are? I have not found an option to do that. Without being able to do that, it can't be very useful. Please help. Thanks a million!

Casey
  • 41,449
  • 7
  • 95
  • 125
user2599757
  • 5
  • 1

1 Answers1

0

Unfortunately, this is a current known limitation of Carrot2 Workbench. For various reasons, the interactions between Aduna visualization and the rest of Workbench are limited, it is not possible to inspect the documents shared by two or more clusters.

At some point we're hoping to replace Aduna with another visualization of similar capabilities. For this reason we're not planning to hack around Aduna to enable the full set of interactions.

Some old binaries of Aduna visualization are available from their Maven repository though, so if you'd like to play with them and Carrot2 clustering on your own, all the required resources are there.

Stanislaw Osinski
  • 1,231
  • 1
  • 7
  • 9
  • Thanks for the info. Now is exporting clusters also not doable at this point? I want to feed selected clusters them back into C2 for further clustering (which is an option in the Search tab, but I am not seeing a way to export the clusters in order to do so. thanks. – user2599757 Jul 26 '13 at 16:38
  • See the File -> Save as... menu option. – Stanislaw Osinski Jul 26 '13 at 20:18
  • but that's only doc IDs, not full set of citation with abstract as in the XML I uploaded for clustering. I was hoping to get the clusters out in the same format so that I can feed the selected onese in again for further clustering. – user2599757 Jul 29 '13 at 14:15
  • The saved XML should also contain the documents you submitted for clustering (unless you unchecked the relevant check box in the Save dialog). The clusters reference documents from the same file by ID. – Stanislaw Osinski Jul 30 '13 at 07:58