Search Website

Testimonial

We love the software and it is truly enabling to use to manage and analyze a mountain of data.

Michael A. Raiber, Ph.D.
Assistant Professor of Music Education
University of Oklahoma


Members Login
Members Online
None
Current Poll
What New HyperRESEARCH Feature Do You Want?
 
Home > Community > Forums > Visit Our Forums > Online Chats > Chat Transcript 10/9/09: "Text Mining of...
 ResearchWare Forum :: Online Chats
Board Index Latest Posts Welcome Guest   
Chat Transcript 10/9/09: "Text Mining of Qualitative Data using Synonyms"   by ann on 2009-10-10 03:09:44
 Subject :Chat Transcript 10/9/09: "Text Mining of Qualitative Data using S.. 2009-10-10 03:09:44 
ann
Staff Moderator
Joined: 2009-07-07 13:14:08
Posts: 6
Location: Boston, MA
Subject :Chat Transcript 10/9/09: "Text Mining of Qualitative Data using Synonyms"

Feel free to continue this conversation by posting to this thread!

(transcript edited for spelling and clarity)

* paul joins Main Room

14:04:30 * richard joins Main Room
14:05:04 ‹paul› While we're waiting, what interests you about using synonyms in coding or mining qualitative data?
14:08:10 ‹guest1› I am doing research on what is available in software and it seems that filtering for synonyms has some positive respects and some complications and I wanted to see what others are talking about related to using synonyms.
14:08:50 ‹paul› That's exactly what this chat is intended to explore!
14:10:43 ‹paul› Well, there are still some people who haven't come over to chat yet, but lets get started. I have a few observation/questions/comments to get the discussion started.
14:11:18 * guest2 joins Main Room
14:11:21 ‹paul› First let me welcome everyone to an on-line chat discussion of using word synonyms to construct automatic code relationships in the text mining of qualitative data.
14:1‹paul› I'd like to explore the pros and cons to using synonyms to key word(s) in automatically coding qualitative data - at least in a first pass. Is the use of synonyms "valid" when coding qualitative data? What level of human review is warranted?
‹paul› To what "level" should sysnonyms be used? B is a synonym for A and C is a synonym for B but not directly for A. Should someone be able to code for A, B, and C? or even deeper "levels"?
14:13:01 ‹guest2› Why not!
‹paul› Some qualitative folks might argue that the further removed a synonym is the less it is associated with the original "meaning" of the underlying source materials?
14:14:31 ‹guest2› Do you mean searching for synonyms of "valid", like "well grounded", "exact"
14:14:43 ‹paul› Yes, exactly
14:15:10 ‹paul› Or "hate" has "dislike", etc. as synonyms
14:15:36 ‹guest2› Yes, but if one can review hits before coding them, there is no issue of searching for those and eliminating those that are irrelevant
‹paul› So, when selecting a key word to find and code in source material, a QDA package could present a list of synonyms for that word. The user could then select which synonyms to search for or add them all.
14:18:14 ‹guest2› Yes!
14:18:57 ‹paul› Once it starts searching, should you be able to review each instance of matching source material before coding or review after they have been coded to confirm the codes?
14:20:56 ‹guest2› Before and after.  One may work quickly and then review things afterward or take time to review all matching source
‹paul› it seems you are in favor of synonym support in auto coding of sources. Would you mind taking a momement to indicate how it would help?
14:22:44 ‹guest2› I can't stay for long, but basically, if I'm searching for something specific, it would be difficult for me to imagine all the different way one could express an idea.
14:23:37 ‹paul› So broad synonym coverage lets you err on the side of catching MORE instances of a concept and then filtering down to what you're really looking for?
14:23:53 ‹guest2› Absolutely!
‹paul› Anoy other thoughts from others in this chat?
14:26:24 ‹paul› Okay, I have another question for folks...
14:26:47 ‹paul› Does the use of synonyms need to be tracked or recorded so that one can reconstruct that code A was attached to source material X becuase it contains B which is a synonym of A?
14:27:09 ‹paul› Does that question make sense?
14:27:22 ‹paul› I can rephrase if not.
14:27:39 ‹guest1› I think I understand it.
14:28:05 ‹guest2› Yes it has something to do with keeping an audit trail
14:28:20 ‹guest1› I also think that it does need to be recorded or tracked.
‹paul› Exactly - how essential is that audit trail to publishing, reporting or defending your research?
14:29:46 ‹guest2› In principal, it is essential. In practice, many qualitative researchers don't mind
‹paul› Here's wehere there may be the "Devil in the details". For example an audit trail could just keep that C is a synonym of B and B is a synonym of A OR...
‹paul› It could also track the defintions of A, B, and C so that their context to how C relates to B and to A. Thought on what is important to kepe track of?
‹paul› Said another way, what minimally and ideally would you like to see regarding the use of synonyms in coding?
‹paul› I also have one last question for folks to consider.
14:35:58 ‹guest2› I like to know which one I used, if it is just not be able to replicate the search  later
14:36:00 ‹paul› Code Networks or Maps are graphical representations of code relationships found in HyperRESEARCH and some other QDA tools. What is the value of using synonyms in the construction of code maps?
14:36:43 ‹paul› So if the search could be saved, that could serve as a minimal "audit trail"?
14:37:13 ‹guest2› minimal yes!
14:38:01 ‹paul› I assume ideal would be to track the A to B to C  to source material relationships behind the scenes?
14:38:23 ‹guest2› Better yes!
14:38:48 ‹paul› Any further thoughts on audit trails?
‹guest2› My week is off, I have to go now.  Have a nice week-end.
14:39:00 ‹paul› Thanks for coming!
14:39:36 ‹paul› For folks left, any futher thoughts on audit trails?
14:41:43 ‹guest1› I am sorry, I am just getting my feet wet with this type of research, so pardon my possible skewed understanding, but wouldn't you need to know how you got there in case there are multiple meanings of A, so you could back out and explore other meanings.
14:43:36 ‹paul› Yes, that's what I was suggesting with the posisble need to track that a chunk of source material was coded with Code X becuase it contained key word B and B is a synonym of A which was the key word originall search for to be coded with X. Is that what you'd like to see?
14:45:‹guest1› Yes, otherwise you might go down a path when there were multiple options and end up with only one of what might have been multiple interpretations.
14:46:42 ‹paul› Thanks, good feedback. As you're "getting your feet wet" have you used code networks or code maps?
14:47:07 ‹guest1› No, just read about them.
14:47:47 ‹paul› Any other thoughts, comments, or questions about this topic?
14:48:37 ‹guest1› No, thanks for letting me lurk.
14:49:24 ‹paul› Okay then. Thanks to everyone for coming. Please keep an eye out on the event list on this web site for other chats that may interest you.
‹paul› This was very helpful and informative for me to understand researchers' needs in this area. Thanks again!
14:50:0*guest1 quit
IP Logged
--
Ann Dupuis
ResearchWare, Inc.
ann@researchware.com
Page # 


Powered by ccBoard


Copyright © 1997-2010 ResearchWare, Inc. All Rights Reserved.

HyperRESEARCH™ and HyperTRANSCRIBE™ are trademarks of ResearchWare, Inc. All other trademarks are the property of their respective owners.

Your use of this web site is governed by our Terms of Use. The privacy of any information we collect or you share with us is governed by our Privacy Policy.

Questions? See our Contact Information for ways to reach us.


JoomlaWatch Stats 1.2.9 by Matej Koval