Re: DSE TC, Friday May 25

One of the features of Exhibit is faceted browsing--the ability to 
narrow down a collection of things by various dimensions in no 
predefined order. I'm wondering how faceted browsing can be usefully 
integrated with the hierarchical view that you want.

One way is that the hierarchy is a facet itself. For example, if what 
the exhibit is showing are patient records, then a hierarchical facet 
might be about geographical locations of those records:

    Massachusetts (37)
       Clinic A (12)
       Clinic B (25)
    Ohio (28)
       Clinic C (16)
       Clinic D (12)
    ...

Selecting Massachusetts would show all 37 records in Massachusetts, 12 
from clinic A and 25 from clinic B.

Another way of integrating faceted browsing with the hierarchy is that 
the hierarchy visualizes the data being browsed. (This is probably what 
you meant originally, but without faceted browsing.) So, since I don't 
know much more about clinical data stuff, I'll give an example about 
artists and the hierarchy is about their influences on one another's style:

    Artist A masters techniques X and Y; influences
       Artist B masters techniques X and Z; influences
          Artist D masters technique Z
       Artist C masters techniques Y; influences
          Artist E masters technique Y; influences no other artist
    Artist F masters techniques X and Z; influences
       Artist G masters technique Z; influences no other artist

Now if you use the "technique" facet and pick only "technique Z", then 
the hierarchy is reduced to

    Artist B masters techniques X and Z; influences
       Artist D masters technique Z
    Artist F masters techniques X and Z; influences
       Artist G masters technique Z; influences no other artist

showing only artists who know technique Z. The ability to view data in a 
hierarchy and yet filter that hierarchy on-the-fly is pretty new and I 
wonder if that's of any use to you.

David

Dobson, Stephen wrote:
> I imagine something similar to the OBO bioportal when visualizing the
> geneontology 
> (http://www.bioontology.org/ncbo/faces/pages/ontology_details.xhtml?onto
> logy_display_name=Gene%20Ontology then hit the visualize button). 
>
> The left side of the screen would show the elements of a taxonomy in a
> hierarchical view.  The right side could show something about the
> element chosen, whether descriptions or observations 
>
> Examples include various taxonomies we'd be interested in creating and
> quickly sharing around the organisation. 
>
> --Showing how we distinguish between a hierarchy of terms such as
> portfolio, therapeutic area, programs, studies, projects. 
>
> --How a clinical protocol could be broken down into events, findings,
> interventions, the datasets underneath them and then the variables below
> that.  
>
> --From the paper Kirsten pointed out to me Informatics: The Fuel For
> Pharmacometric Analysis http://www.aapsj.org/view.asp?art=aapsj0901008.
> You could use this hierarchy to view a taxonomy of pharmacometric
> entities and their relationships
>
> --How a governance committee is organized and related to other
> governance committees.  In this case if a person is on multiple
> governance committees then I suspect something could be done with the
> properties functionality to allow jumping into different areas of the
> hierarchy.  
>
> --A simple visualization of the meddra hierarchy if not too large for
> the JS to handle. 
>
> Cheers
>
> Stephen.
>
>
> -----Original Message-----
> From: public-hcls-dse-request@w3.org
> [mailto:public-hcls-dse-request@w3.org] On Behalf Of David Huynh
> Sent: Saturday, May 26, 2007 6:50 PM
> To: Dobson, Stephen
> Cc: Eric Neumann; Forsberg, Kerstin L; public-hcls-dse@w3.org;
> karger@mit.edu
> Subject: Re: DSE TC, Friday May 25
>
>
> Hi Stephen,
>
> Glad you like Exhibit! Could you show me how you want those hierarchical
>
> views to look/behave like? I'm not versed in the particulars of your 
> data domain.
>
> Thanks,
>
> David
>
> Dobson, Stephen wrote:
>   
>> I love it.
>>
>> Yes it won't take the full set of data I'm just creating a smaller 
>> subpopulation to make it work. Thanks for the dates thing, hadn't 
>> worked that out.
>>
>>
>> Sure. I can generate a SAS file with some metabolizing enzyme 
>> polymorphisms very easily and see if that can link in.
>>
>> I'd love some sort of quick hierarchical viewing mechanism if that's 
>> possible.
>>
>> Similar to how the NCBO's bioportal works. It would be great for 
>> viewing things like the meddra hierarchy, organizational charts, 
>> multidimensional information. I'm forever creating mindmaps to demo 
>> little bits of this type of hierarchical data.
>>
>> Cheers
>>
>>
>> Stephen.
>>
>>
>>     
> ------------------------------------------------------------------------
>   
>> *From:* Eric Neumann [mailto:eneumann@teranode.com]
>> *Sent:* Friday, May 25, 2007 4:44 PM
>> *To:* Dobson, Stephen; Forsberg, Kerstin L; public-hcls-dse@w3.org
>> *Cc:* dfhuynh@csail.mit.edu; karger@mit.edu
>> *Subject:* RE: DSE TC, Friday May 25
>>
>> Stephen,
>>
>> I was amazed how large the treament.xls file was, and so when I passed
>>     
>
>   
>> it to Babel with the other 3, it choked... happy to hear you managed 
>> to make it work. Please append it to the DSE site.
>>
>> Some Babel tips:
>>
>> - make sure the main row item ID is in the first column, then babel 
>> will construct a URI for it (extra property).
>>
>> - additional entity ID's (linker columns), need a URI to be generated 
>> manually (I believe), but will still use these local IDs as cross 
>> linkers if they have the same prop name in other files: e.g., AE_ID
>>
>> - append a ':url' after a attribute (column) name if you wish it to be
>>     
>
>   
>> used as a URI
>>
>> - other data types can be appended to column names to help Babel (like
>>     
>
>   
>> :item)
>>
>> - object types can also be assigned (People, Events, etc) by adding a 
>> 'type' column
>>
>> - dates should by in the ISO 8601 format (yyy-mm-dd)
>>
>> If we can get all 4 files together correctly, building viewers should 
>> be easy.... should we try this also for some simple SAS tables?
>>
>> I am working with the MIT researchers (David Karger and David Hyunh 
>> have been copied on this email;) to identify possible future features 
>> in Exhibit that we would find useful-- please pass along any general 
>> recommendations!
>>
>> Eric
>>
>>
>>
>> -----Original Message-----
>> From: Dobson, Stephen [mailto:Stephen.Dobson@pfizer.com]
>> Sent: Fri 5/25/2007 4:18 PM
>> To: Forsberg, Kerstin L; Eric Neumann; public-hcls-dse@w3.org
>> Subject: FW: DSE TC, Friday May 25
>>
>> Hi all,
>>
>>
>>
>> The Exhibit system provides a great way of visualizing the tools.
>>
>>
>>
>> Based on our conversation today am I right in thinking that this
>>     
> system
>   
>> should "know" how to join data from multiple spreadsheets based on the
>> same URI's. It looks like it has managed it for the AE's and
>>     
> demography
>   
>> in the demonstration Eric provided. Eric I assume you joined both
>>     
> these
>   
>> datasets together simply by loading into bable?
>>
>>
>>
>> I've managed to join treatment, demog and vitals together in the same
>> way. It took me a while to work out you needed a unique record
>> identifier (e.g. vt_1, similar to Eric's AE_1) but after I discovered
>> that it works great. I need to greatly reduce the spreadsheet to have
>> any hope of visualizing through the JSON file though.
>>
>>
>>
>> I imagine we could eventually have applications where data could be
>> dragged and dropped and based on the URI's the system would know how
>>     
> to
>   
>> join the data and what visualizations could be performed. This would
>> be really nice.
>>
>>
>>
>> Cheers
>>
>> Stephen.
>>
>>
>>
>>
>>
>> ________________________________
>>
>> From: public-hcls-dse-request@w3.org
>> [mailto:public-hcls-dse-request@w3.org] On Behalf Of Forsberg, Kerstin
>>     
> L
>   
>> Sent: Friday, May 25, 2007 8:15 AM
>> To: Eric Neumann; public-hcls-dse@w3.org
>> Subject: RE: DSE TC, Friday May 25
>>
>>
>>
>> Hi Eric,
>>
>> looks like we have a lot of interesting stuff to cover!
>>
>>
>>
>> I have been playing around with the Exhibit viewer and JSON data.
>>     
> Great
>   
>> stuff!!!
>>
>>
>>
>> I would like us to have a discussions around what "web of data"
>> considerations that needs to made/applied to really make this the
>>     
> killer
>   
>> app. That is, what we should consider in the process of taking data
>>     
> from
>   
>> existing sources such as SAS/Excel datasets, rendering them as
>>     
> instances
>   
>> of common entities (observations), enriching them by assigning URI:s
>>     
> and
>   
>> annotating them using different ontologies, before making up the JSON
>> files
>>
>>
>>
>> Talk to you soon
>>
>> Cheers Kerstin
>>
>> -----Original Message-----
>> From: public-hcls-dse-request@w3.org
>> [mailto:public-hcls-dse-request@w3.org]On Behalf Of Eric Neumann
>> Sent: 25 maj 2007 03:08
>> To: public-hcls-dse@w3.org
>> Subject: DSE TC, Friday May 25
>>
>>
>>
>> SW-HCLS(DSE) Friday, May 25, 8:30am-9:30am/13:30-14:30 UTC
>> for a duration of 1 hour
>> Zakim Bridge +1.617.761.6200 <tel:+1.617.761.6200> , conference
>> code
>> 3731 ("DSE1")
>> Coordinator: Eric Neumann <eneumann@teranode.com>
>>
>> Agenda:
>> a) Roll call, minutes review
>>
>> b) Propose next DSE call June 8, 2007
>>
>> c) 'Brief' review of CTO workshop meetings - take away messages?
>> - On a related point, paper from Kerstin on 'informatics and
>> pharmacometrics': http://www.aapsj.org/view.asp?art=aapsj0901008
>>
>> d) Planning Preparation for DIA panel June 20
>>
>> e) SIMILE Exhibit viewing of SDTM data being constructed (see
>> attachments); will this be appropriate for a SDTM+biomarker RDF demo?
>>
>> f) BioONT DSE DCM discussion; see
>> http://esw.w3.org/topic/HCLS/OntologyTaskForce/BIONTDSEDCM
>>
>> g) (if we have time) DSE paper: goals, (journal), content, and
>> timing
>>
>>
>> cheers,
>> Eric
>>
>> ----------------------------------------------------------------------
>> LEGAL NOTICE
>> Unless expressly stated otherwise, this message is confidential and 
>> may be privileged. It is intended for the addressee(s) only. Access to
>>     
>
>   
>> this E-mail by anyone else is unauthorized. If you are not an 
>> addressee, any disclosure or copying of the contents of this E-mail or
>>     
>
>   
>> any action taken (or not taken) in reliance on it is unauthorized and 
>> may be unlawful. If you are not an addressee, please inform the sender
>>     
>
>   
>> immediately.
>>
>>     
>
>
>
>

Received on Thursday, 31 May 2007 20:51:14 UTC