[bioontology-support] Possible cycling problems in MeSH

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[bioontology-support] Possible cycling problems in MeSH

Bekhuis, Tanja

Hello,

 

For the last few weeks I have been exploring the structure of several terminologies available via the NCBO BioPortal. As a consequence, I have become familiar with displays of trees in the left pane. So I was rather perplexed by the tree for MeSH as it appears in the portal.

 

I attached a brief document with 4 images: 2 from the NCBO BioPortal and 2 from the NLM online browser for comparison.

 

Notably, only 4 of the 16 categories display in the initial screen for MeSH in the portal. If I try to add a category that should be there, e.g. the Humanities category, topmost concepts repeat as one traverses the tree and the system ‘hangs’ when I try to explore further.

 

If you read through the attached document and compare the two sources, you will see odd discrepancies that suggest internal cycles in the NCBO version.

 

I would very much appreciate guidance regarding how to interpret your version of MeSH, as I had intended to use it in a research project underway.

 

Sincerely,

Tanja

Tanja Bekhuis, PhD, MS, MLIS

 

Postdoctoral Associate

Department of Biomedical Informatics / School of Medicine

University of Pittsburgh

Pittsburgh, Pennsylvania 15232, USA

E-mail: [hidden email]

Phone at UPMC Shadyside:  +1 412-647-6705

 

 

 


_______________________________________________
bioontology-support mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/bioontology-support

TBekhuis_Problems_with_MeSH_via_NCBOBioPortal_04072011.docx (873K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: [bioontology-support] Possible cycling problems in MeSH

Ray Fergerson

Tanja,

 

This is pretty confusing. First, we should not even be displaying the MeSH hierarchy. After investigating this thoroughly some time ago we decided that we could not support an accurate display of the hierarchy so, rather than support the wrong display, we would not show any hierarchy. The MeSH terms themselves should be available in BioPortal but the hierarchy should not. That you seen something is actually a bug.

 

The problem that we discovered with our support of MeSH is that the “hierarchical” relationship between nodes in MeSH is not transitive. The effect of this is that the children of a term depend upon the path that you used to reach that term. This is not consistent with usual ontology  is-a relationships which BioPortal was designed around. Addressing this correctly for MeSH would require a redo of the UI, backends, and of all of our REST services.  This is just not feasible. BioPortal is really designed to explore transitive is-a and part-of relations and we have to limit ourselves to this domain. Fortunately this domain covers the overwhelming number of terminologies. Unfortunately it does not cover MeSH.

 

Ray

 

 

From: [hidden email] [mailto:[hidden email]] On Behalf Of Bekhuis, Tanja
Sent: Thursday, April 07, 2011 1:59 PM
To: [hidden email]
Cc: Bekhuis, Tanja; Rebecca Crowley
Subject: [bioontology-support] Possible cycling problems in MeSH

 

Hello,

 

For the last few weeks I have been exploring the structure of several terminologies available via the NCBO BioPortal. As a consequence, I have become familiar with displays of trees in the left pane. So I was rather perplexed by the tree for MeSH as it appears in the portal.

 

I attached a brief document with 4 images: 2 from the NCBO BioPortal and 2 from the NLM online browser for comparison.

 

Notably, only 4 of the 16 categories display in the initial screen for MeSH in the portal. If I try to add a category that should be there, e.g. the Humanities category, topmost concepts repeat as one traverses the tree and the system ‘hangs’ when I try to explore further.

 

If you read through the attached document and compare the two sources, you will see odd discrepancies that suggest internal cycles in the NCBO version.

 

I would very much appreciate guidance regarding how to interpret your version of MeSH, as I had intended to use it in a research project underway.

 

Sincerely,

Tanja

Tanja Bekhuis, PhD, MS, MLIS

 

Postdoctoral Associate

Department of Biomedical Informatics / School of Medicine

University of Pittsburgh

Pittsburgh, Pennsylvania 15232, USA

E-mail: [hidden email]

Phone at UPMC Shadyside:  +1 412-647-6705

 

 

 


_______________________________________________
bioontology-support mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/bioontology-support