Welcome, Guest. Please Login
Tinderbox
  News:
IMPORTANT MESSAGE! This forum has now been replaced by a new forum at http://forum.eastgate.com and no further posting or member registration is allowed. The forum is still accessible via read-only access for reference purposes. If you wish to discuss content here, please use the new forum. N.B. - posting in the new forum requires a fresh registration in the new forum (sorry - member data can't be ported).
  HomeHelpSearchLogin  
 
Pages: 1 2 3 4
Send Topic Print
Tinderbox for Textual Analysis (Read 110437 times)
Loryn
Full Member
*
Offline



Posts: 97

Tinderbox for Textual Analysis
Jun 18th, 2009, 5:51pm
 
I hope Mark doesn't mind, but this thread speculates on the possibility of developing new Tinderbox capabilities as analogues to existing Tinderbox functionality, for some specific specialised domains.

As way of introduction, my academic training is in Linguistics and Computational Linguistics, following the European tradition of linguistics as a social phenomenon (following Firth, Halliday, Hassan), rather than the American tradition of linguistics as a branch of psychology. So, for me, syntagmatic linguistics is closely related to textual analysis, textual criticism, literary analysis.

The functionality I describe below would be useful in a range of fields:
- linguistics
- literary criticism
- law
- biblical studies
- any academic study in which close study of extant texts is prevalent.

Tinderbox - The metaphors with which I use Tinderbox
What attracts me to Tinderbox, as a knowledge worker in general, is the ability to seamlessly combine visual-analytic, data-analytic and textual-analytic modes of representation together with programmatic facilities to enable searching and large-scale data manipulation.

The basic metaphors for Tinderbox are:
1. Tinderbox as a database.
2. Tinderbox as a semantic-network.
3. Tinderbox as an outliner.
4. Tinderbox as simple UI-builder (workflow mechanism).

My primary metaphor is Tinderbox-as-a-database: Tinderbox Notes are Records; the Prototype as a Type (Table); the Agent as a Query Language / Data Manipulation Language.

Once data has been assembled in a database-like fashion. I seek to structure the data through analysis. This type of analysis seeks to classify like-with-like. Sometimes I use attributes (usually boolean or set attributes) applied with stamps. When I'm being informal, I use colors. When I'm tentative, and not yet ready for over classification, I switch into map view and use spatial and visual cues to arrange the data and record 'observations'---before I know which observations are significant. Relations between various pieces of data are signified through links.

Once the data has proceeded a certain way, I use agents to pull out groupings for further analysis. Observations, hypotheses, etc. become new notes.

When the research is nearing write-up, I assemble existing notes, create new ones, and generally use Tinderbox as a standard outliner.

Slightly outside of this flow of discussion, I also use Tinderbox as a simple workflow-driven UI-builder. I have an "application" that presents me with a text to study each day. (It's implemented by me scheduling in the texts for the next month, by date. Each day, I open the Tinderbox document, it shows me the text for the day---highlighted in yellow. When I've recorded my observations in the note, the note turns black; excepting one day per week in which it turns gray.)

The desired metaphor - Tinderbox for Textual Study
The new metaphor I'd like to add is that of Tinderbox as a tool for textual study.

In this metaphor, we have the following components:
1. A Text. The text that informs our study. This is like the contents of a Tinderbox note, excepting that the words are not "ours": they are the topics of our study. Also, there is often a need for referencing particular features in this text: page and line numbers in legal texts; line numbers in poetry; verses in biblical study. Multiple texts comprise a corpus.

2. A Textspan. A Textspan consists of a span of a target Text, i.e. a point at which the span begins, and a point at which it ends. Textspans are like current Tinderbox notes, with the additional property of their referencing a span in a Text. Applying a Textspan to a Text enables some unique visual properties: e.g., one could use a Textspan to cause the spanned text to appear as a different color, or to have an underline, or a double or triple underline, to have a straight vs wavy underline, etc. Textspans inherit from Textspan prototypes just like Notes inherit from Note prototypes. In a map-type view, a Textspan can be represented as the span-across-the-text and as a callout to the "note body" (like Microsoft Word 2003 comments).

When you open the note for the Textspan, at the top of the note, you see the underlying text which comprises the Textspan. Should other Textspans exist within the current span, they are visible. Thus you can recursively navigate Textspans, allowing the linguistic notion of Rank to be implied.

3. An overlay. The Text, and any Textspan, has the notion of an overlay. A Textspan belongs to an overlay. I can switch-on or switch-off overlays. (Some overlays may be useful when semi-transparent.) I can use agents to gather textspans into overlays. Using overlays, I can execute multiple studies on the one text. For example, one overlay could contain my generic structure analysis; another contain my grammatical analysis; and a third my cohesion analysis. Then I can arbitrarily switch them on-and-off, or adjust transparency, to see how my multiple analyses inform each other.

4. A concordance. A concordance is like an agent, excepting that it searches Texts, and generates not aliases, but Textspans. The output of a concordance is list of Textspans (not a list of aliases of Textspans). The Textspans produced by a concordance can, optionally, be manually adjusted as to their beginning and ending.

I believe these concepts would significantly extend Tinderbox's utility for all fields involving textual analysis.
Back to top
 
 
  IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #1 - Jun 18th, 2009, 7:09pm
 
I've posted a close variant of this post onto my (very NEW) personal blog:

http://loryn.me

Back to top
 
 
  IP Logged
Mark Bernstein
YaBB Administrator
*
Offline

designer of
Tinderbox

Posts: 2871
Eastgate Systems, Inc.
Re: Tinderbox for Textual Analysis
Reply #2 - Jun 18th, 2009, 11:18pm
 
This makes good sense; indeed, it could make a good research paper at Hypertext or IVICA or Computing in the Humanities.  

A lot of the infrastructure is already in Tinderbox -- the concordance, for example.  And something like your "textspans" (which needs a new term, since that one is already taken) should be part of Tinderbox 5.

I'm not quite clear how overlays work; are they interchangeable files? Intermedia used to do something like that.
Back to top
 
 
WWW   IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #3 - Jun 18th, 2009, 11:56pm
 
Yes, it's exciting how much of the infrastructure for this is already in place within Tinderbox. I've been using Tinderbox in multiple projects where I've been trying to emulate some of these ideas. The current conceptual infrastructure is all in place; so this extension would fit quite naturally at some stage in Tinderbox's development.

Overlays. I'm thinking of overlays as a visibility layer, similar to what you'd see in CorelDraw or PhotoShop. Remember back to the days when people used to use overhead transparencies? An "overlay" was an additional transparency laid down to provide additional annotations.

In terms of implementation, one could imagine an implementation for overlays where you could save and export them independently of each other. Indeed, you could establish the base text in a separate file from the Textspans / overlays. That would allow for workgroups (think of a college class) to provide their analyses on a common base text, and enable sharing with each other; or allow the lecturer to mount each student's analysis over the same base text for review / marking.

That's all possible. But the important feature is that you can visually impact a text analogous to marking it up by hand, and then modify your view (i.e. change visibility on the overlays) for further work. Otherwise the page gets too busy.

(As an aside, remember back in the dim, dark days of the late nineties. There was a company that provided a browser plugin that enabled  annotations on websites. One of the reasons the concept failed was because they provided a single global channel for all annotations on the website. They failed to provide layers / overlays / channels; they failed to establish a community where I was willing to view Mark Bernstein's website markups and exclude Joe the Plumber's, so instead ended up with 5,000 Joe-the-Plumber-level (i.e. spam-like) remarks defacing websites.)
Back to top
 
« Last Edit: Jun 19th, 2009, 3:45am by Loryn »  
  IP Logged
Mark Bernstein
YaBB Administrator
*
Offline

designer of
Tinderbox

Posts: 2871
Eastgate Systems, Inc.
Re: Tinderbox for Textual Analysis
Reply #4 - Jun 19th, 2009, 7:10pm
 
There was a good paper at IVICA today about an experimental system for analyzing archives of texts and visual artifacts, such as letters and drawings.   It, too, has a notion similar to your overlay layer.

There's some real latitude for good work in this area.
Back to top
 
 
WWW   IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #5 - Jun 21st, 2009, 1:31am
 
Thanks for the encouragement, Mark.

During the weekend I've been gathering some thoughts regarding this research. I'll be writing up some aspects of this thinking on my blog, http://loryn.me


BTW: Is there any way we can get a copy of that paper you're referring to?
Back to top
 
« Last Edit: Jun 21st, 2009, 1:32am by Loryn »  
  IP Logged
Mark Bernstein
YaBB Administrator
*
Offline

designer of
Tinderbox

Posts: 2871
Eastgate Systems, Inc.
Re: Tinderbox for Textual Analysis
Reply #6 - Jun 21st, 2009, 1:57pm
 
<strike>I don't yet have the URL for this paper</strike>, but the system -- CritSpace, by Richard Furuta and his students at Texas A&M --  is also discussed here:

http://www.ieee-tcdl.org/TCDL/Bulletin/current/audenaert/audenaert.html

and here

http://llc.oxfordjournals.org/cgi/content/abstract/24/2/143


(Added later) The IVICA Proceedings are here:

     http://www.csdl.tamu.edu/IVICA/2009proceedings.pdf

The specific paper I had in mind is the paper on CritSpace, but there are plenty of interesting papers in the bundle.
Back to top
 
« Last Edit: Jun 21st, 2009, 4:46pm by Mark Bernstein »  
WWW   IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #7 - Jun 23rd, 2009, 8:15am
 
I'm writing http://loryn.me entries first in Tinderbox. When an article has been published, I hit the $IsPublished attribute, and it shifts automatically to the Published folder (the original location is stored in a string attribute).



My plan for the remainder of the Textual Analysis Example 1 series ( http://loryn.me/taeg1/ ) is as follows.



And then, there are the exciting plans for outlining my concepts regarding potential Tinderbox UI affordances.



(What I've referred to as Textspan (in this thread) I'm referring to as "Syntagm Lense" in this picture. I don't much like the name; can anyone think of alternative names?)

If anyone sees anything "missing," or have additional suggestions as to the way forward, feel free to speak up. More minds are better than one, etc.
Back to top
 
« Last Edit: Jun 23rd, 2009, 8:18am by Loryn »  
  IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #8 - Jun 24th, 2009, 5:14pm
 
In my latest installment ( http://loryn.me/journal/2009/6/24/assembling-the-text.html ), I manage to massage text into Tinderbox, as shown below.

To do so, I use Microsoft Word regular expression search and replace functionality. Now, I keep Microsoft Word for Windows hanging around on a Windows XP virtual machine almost exclusively for this capability. Does anyone know how to to non-programmatic regex search and replace on a Mac. (Short of firing up Python or purchasing MS Office for Mac OS X.)

Back to top
 
 
  IP Logged
Mark Bernstein
YaBB Administrator
*
Offline

designer of
Tinderbox

Posts: 2871
Eastgate Systems, Inc.
Re: Tinderbox for Textual Analysis
Reply #9 - Jun 24th, 2009, 5:36pm
 
BBEdit/TextWrangler does very nice regex search and replace.  

So does TextMate.
Back to top
 
 
WWW   IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #10 - Jun 25th, 2009, 5:46am
 
Thanks, Mark. I'll make sure I give one or both of them a try in the near future.

Now, just wondering if anyone can extend my thinking in terms of the Contexts of Situation (that's the linguistic name for it) or Scenario Analysis (the interaction design name for it) for Textual Analysis. In the article http://loryn.me/journal/2009/6/21/textual-analysis-in-three-situational-contexts... I outline just three basic scenarios involving Textual Analysis.

Surely I've missed some. What other scenarios can you think of that would involve different user goals, materials, tools or approaches than the three scenarios I've sketched out there?
Back to top
 
« Last Edit: Jun 25th, 2009, 5:48am by Loryn »  
  IP Logged
Mark Anderson
YaBB Administrator
*
Offline

User - not staff!

Posts: 5689
Southsea, UK
Re: Tinderbox for Textual Analysis
Reply #11 - Jun 25th, 2009, 6:17am
 
Loryn, TextWrangler is just the free baby brother of BBEdit. TextWrangler has find/replace (including regular expression based search) for single files and even multiple files. Likely it has everything you need in the context of your request for tools. Spending $$$ on BBEdit will; likely bring you little extra in the way of search/replace functionality (I use BBEdit - but for other purposes such as HTML/JS/CSS editing, etc.)
Back to top
 
« Last Edit: Jun 25th, 2009, 6:23am by Mark Anderson »  

--
Mark Anderson
TB user and Wiki Gardener
aTbRef v6
(TB consulting - email me)
WWW shoantel   IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #12 - Jun 25th, 2009, 2:48pm
 
Hey, thanks MarkA. I'll certainly look into TextWrangler a whole lot earlier now that I know that. Smiley
Back to top
 
 
  IP Logged
Victor Cova
Full Member
*
Offline



Posts: 5

Re: Tinderbox for Textual Analysis
Reply #13 - Jun 25th, 2009, 6:04pm
 
Hey Loryn, I'm not sure whether this really fits in any of your three scenarios, but in sociology, using qualitative methods requires
1- being able to type a lenghty text from taped recording (e.g.: interview E1, E2, E3 with women victim of domestic violence)
2- "coding" various parts, that is identifying sentences by categories (C1 for childhood, C2 for stigma, C3 for why they don't leave their partner)
3- sorting different text interviews by code and comparing them (ex: E1: "When I was a child, my father used to drink a lot" could be coded C1)
4- visualising the succession of codes and comparing different successions in different interviews (E1 started talking about her childhood, then about her partner, then about stigma, whereas E2 did the contrary) and how much each interview devotes to each code
5- linking different excerpts,  among them, with excerpts from articles and to parts of the analysis.

And for various disciplines dealing with images, one would need a way to link parts of an image to bits of text or to other bits of other images, and why not as well with sound and video excerpts. I'm thinking film studies, anthropology, musicology, history, but also probably biology, healthcare.But that's outside textual analysis.
Back to top
 
 
  IP Logged
Loryn
Full Member
*
Offline



Posts: 97

Re: Tinderbox for Textual Analysis
Reply #14 - Jun 26th, 2009, 12:37am
 
Thanks, Victor. That's exactly the sort of input I'm looking for. I'll certainly consider the sociological and image analysis uses you've mentioned.

To everyone: What else is there?
Back to top
 
« Last Edit: Jun 26th, 2009, 12:38am by Loryn »  
  IP Logged
Pages: 1 2 3 4
Send Topic Print