CIRCA:Viral Analytics Paper Proposal

From CIRCA

(Difference between revisions)
Jump to: navigation, search
 
(11 intermediate revisions not shown)
Line 1: Line 1:
-
==Viral Analytics: Embedded Voyeur==
+
==Viral Analytics: Embedding eVoyeur in Content Systems==
-
Viral Analytics built an embeddable Voyeur, eVoyeur, and is testing it in different environments.  This tool is able to perform various text analysis while being embedded in a site.  This means that users can access these text analysis tools while looking at the article.  Since eVoyeur is in its primary stage, we are conducting interviews with journal and blog editors in the Digital Humanities to determine what kinds of tools would be beneficial for these sites.  The interviews will also help us determine the frameworks that these sites are using.
+
Authors: Geoffrey Rockwell, Ashley Moroz, Stéfan Sinclair, and Corey Slavnik
-
Therefore, in this paper we will
+
===Introduction===
-
* Describe the current text analysis tools available and their benefits and limitations.
+
How can we better mash tools together with texts for interpretation? One model is to create embeddable badges that, like a YouTube video can be dropped into content. To that end the Viral Analytics project has developed and tested an embeddable text analysis micro-environment based on Voyeur Tools [1] called eVoyeur. eVoyeur is able to perform text analysis on web pages on which it is embedded without authors needing to install software on their server. It is an example of what Siemens calls a professional reading tool. [2] For more functionality eVoyeur can also be tucked in as a plug-in for common Content Management Systems (CMS) like WordPress for blogs, Drupal and the Open Journal System (OJS) – the CMS plugins interact intelligently with the backend database and are able to operate on customized sets of documents as defined, say, by categories or tags. Authors thus can give readers access to analytical badges like word clouds that run in the client’s browser when they look at a blog post or article. This paper will discuss eVoyeur and the results of usability interviews with journal and blog editors in the digital humanities that helped us understand how we can better support content editors. This paper will:
-
* Demonstrate eVoyeur within Open Journal Systems, Wordpress, and Drupal
+
-
* Describe the methodology of the interviews
+
# Introduce embeddable tools and plug-ins in general,
-
* Discuss the results from the interviews
+
# Demonstrate eVoyeur in the context of common content management systems like WordPress and OJS,
-
* Conclude with the future of the project.
+
# Discuss the feedback from usability interviews that we conducted, and
 +
# Conclude with a discussion of the future of the project.  
-
# Embed eVoyeur into journals and blogs
+
===Demonstrate eVoyeur in Drupal, Wordpress and OJS===
-
# Conduct an analysis on how users used the tool
+
eVoyeur is an embeddable tool capable of performing text analysis on any page on which it sits. It builds on the TAToo project that created a Flash embeddable tool for blogs. [3] The eVoyeur micro-environment is a sizable badge that can show different views on the content of the page including a word cloud, a list of high-frequency words, a graph of collocations and statistical summary of the content of the page. Cirrus is a Wordle-like (see wordle.net) word cloud where the most frequent words in the document are sized and posititioned according to their frequency. The Frequency Grid shows the overall word frequencies for the entire corpus as well as information about how word frequencies are distributed over documents within the corpus. Links finds collocates for words and displays a network graph of the linked words. The Summary tool provides the longest and shortest document, highest vocabulary density, most frequent words, notable peaks in frequency and distinctive words. In this paper we will demonstrate how the tools work as analytical badges for the reader and then how they can be installed as plug-ins in common CMS systems.
-
# Create three levels of the tool (beginner, intermediate, advanced)
+
-
====Demonstrate eVoyeur in Drupal, Wordpress and OJS====
+
===Methodology and Results of Usability Interviews===
-
eVoyeur is an embeddable tool capable of performing text analysis on the site.  The current tool can perform cirrus, frequency grid, links, reader, and summary.  Cirrus is a word cloud where the most frequent words in the document are displayed. Frequency Grid shows the overall word frequencies for the entire corpus as well as information about how word frequencies are spread out over documents within the corpus. Links finds collocates for words and displays links between them using a force directed graph. This is like a web where all words are a different color depending on which post there are the most hits (places where the word appears). With the reader all posts are color coded on the left hand side of eVoyeur. By navigating through the left side bar you can read all the posts one by one. Summary gives you the longest and shortest document, highest vocabulary density, most frequent words, notable peaks in frequency and distinctive words.
+
In order to determine how analytical badges could be useful to editors of online content we conducted usability interviews with journal and blog editors in the digital humanities. Before the interview an illustrated introduction to the eVoyeur tool was sent to the interviewees to give them ideas. The interviews were conducted over Skype following a loose interview guide that allowed editors to tell us what kinds of text analysis tools would be useful to their audiences. Interview notes were shared back with the interviewees for correction. A preliminary analysis of the interviews shows that the most useful tools are the Cirrus word cloud and the Frequency Grid. Interviewees felt that many of tools suffered because of the small size of the badge imposed by the design of the content management systems. We are now trying to more efficiently use the space allocated to the badge and adding a feature that lets the user expand the view for those tools that don’t work in the small.
-
====Methodology of interviews====
+
We also asked interviewees for suggestions for new features. A couple wanted the tool to operate not just on the page it appears on but also on collections of articles. Another wanted the tool adapted to Omeka and one wanted link analysis. Finally, this being the digital humanities, someone wanted a TEI wrapper. In this presentation we will summarize the results of our interviews and how we have adapted the tool in response.  
-
In order to determine what tools would be useful in the embedded eVoyeur tool we conducted interviews with journal and blog editors. Before the interview a pdf explaining the eVoyeur tool was sent to the interviewee.  This pdf gave a quick summary of how eVoyeur is displayed in OJS and a brief description of the various tools.  The interviews were conducted over skype and questions were asked about what they liked about the Voyeur tool and what kinds of text analysis tools would be useful to their audiences.
+
 +
===Conclusions===
 +
Our goal is to make eVoyeur as attractive as possible to editors of digital humanities. We believe that easy to use analytical badges are one way to bring interpretative tools to content, but not the only way. Large projects will want tools that are more closely integrated and under their control, but for the small online publication eVoyeur can provide accessible functionality that resembles the badge tools that have proliferated on the web. The next step once we have responded to our editor interviewees with an improved micro-environment is to test the tool in real journals and blogs. To that end we are looking for projects willing to include eVoyeur in their interface.
-
====Results of interviews====
+
Notes
-
5 out of 8 people liked Cirrus.  5 out of 8 people liked Frequency Grid. 4 out of 8 people thought links was useful, while one person did not like links. 2 people liked reader and 2 people thought it was not effective since it was too small.  3 people thought Summary was good and 1 person thought it was not useful.  2 people thought the window was too small. One person liked that you could scrape particular issues in an article.  One person found it hard to see the functionality of the tool.
+
 +
[1] Voyeur Tools can be used at <http://voyeurtools.org>.
-
2 people would find a concordance useful. One person would like the Links to navigate between articles.  5 people would like searchability or link ability for articles. One person would like a tool to conduct text analysis on all articles in journal. One person would like a trend list to find trends in a particular field. One person would like summary to link to an article.  One person would like collocates.  One person would like eVoyeur on Omeka.  One person would like more visualizations.  One person would like link analysis.  One person would like a TEI wrapper.  One person would like a personography/placeography markup.
+
[2] Siemens, Ray et al. "A Study of Professional Reading Tools for Computing Humanists." A report at <http://etc.uvic.ca/public/pkp_report/>.
-
 
+
[3] For the TAToo (Text Analysis for me Too) project see <http://ra.tapor.ualberta.ca/~tatoo/>. For a discussion of this tool and our experiments see “Ubiquitous Text Analysis” in the Poetess Archive Journal, <http://paj.muohio.edu/paj/index.php/paj/article/view/13>.
-
One person said readers are suspicious of text analysis since it is seen as a bibliographical searching. One person said there are too many options that will overwhelm users. One person said it was good for quantitative research. One person said it would be better if based in the cloud. One person said it needs more robust digitization tools. 2 people would like visulizations to bring together all 6 tools.
+

Current revision as of 13:36, 18 March 2011

Contents

Viral Analytics: Embedding eVoyeur in Content Systems

Authors: Geoffrey Rockwell, Ashley Moroz, Stéfan Sinclair, and Corey Slavnik

Introduction

How can we better mash tools together with texts for interpretation? One model is to create embeddable badges that, like a YouTube video can be dropped into content. To that end the Viral Analytics project has developed and tested an embeddable text analysis micro-environment based on Voyeur Tools [1] called eVoyeur. eVoyeur is able to perform text analysis on web pages on which it is embedded without authors needing to install software on their server. It is an example of what Siemens calls a professional reading tool. [2] For more functionality eVoyeur can also be tucked in as a plug-in for common Content Management Systems (CMS) like WordPress for blogs, Drupal and the Open Journal System (OJS) – the CMS plugins interact intelligently with the backend database and are able to operate on customized sets of documents as defined, say, by categories or tags. Authors thus can give readers access to analytical badges like word clouds that run in the client’s browser when they look at a blog post or article. This paper will discuss eVoyeur and the results of usability interviews with journal and blog editors in the digital humanities that helped us understand how we can better support content editors. This paper will:

  1. Introduce embeddable tools and plug-ins in general,
  2. Demonstrate eVoyeur in the context of common content management systems like WordPress and OJS,
  3. Discuss the feedback from usability interviews that we conducted, and
  4. Conclude with a discussion of the future of the project.

Demonstrate eVoyeur in Drupal, Wordpress and OJS

eVoyeur is an embeddable tool capable of performing text analysis on any page on which it sits. It builds on the TAToo project that created a Flash embeddable tool for blogs. [3] The eVoyeur micro-environment is a sizable badge that can show different views on the content of the page including a word cloud, a list of high-frequency words, a graph of collocations and statistical summary of the content of the page. Cirrus is a Wordle-like (see wordle.net) word cloud where the most frequent words in the document are sized and posititioned according to their frequency. The Frequency Grid shows the overall word frequencies for the entire corpus as well as information about how word frequencies are distributed over documents within the corpus. Links finds collocates for words and displays a network graph of the linked words. The Summary tool provides the longest and shortest document, highest vocabulary density, most frequent words, notable peaks in frequency and distinctive words. In this paper we will demonstrate how the tools work as analytical badges for the reader and then how they can be installed as plug-ins in common CMS systems.

Methodology and Results of Usability Interviews

In order to determine how analytical badges could be useful to editors of online content we conducted usability interviews with journal and blog editors in the digital humanities. Before the interview an illustrated introduction to the eVoyeur tool was sent to the interviewees to give them ideas. The interviews were conducted over Skype following a loose interview guide that allowed editors to tell us what kinds of text analysis tools would be useful to their audiences. Interview notes were shared back with the interviewees for correction. A preliminary analysis of the interviews shows that the most useful tools are the Cirrus word cloud and the Frequency Grid. Interviewees felt that many of tools suffered because of the small size of the badge imposed by the design of the content management systems. We are now trying to more efficiently use the space allocated to the badge and adding a feature that lets the user expand the view for those tools that don’t work in the small.


We also asked interviewees for suggestions for new features. A couple wanted the tool to operate not just on the page it appears on but also on collections of articles. Another wanted the tool adapted to Omeka and one wanted link analysis. Finally, this being the digital humanities, someone wanted a TEI wrapper. In this presentation we will summarize the results of our interviews and how we have adapted the tool in response.

Conclusions

Our goal is to make eVoyeur as attractive as possible to editors of digital humanities. We believe that easy to use analytical badges are one way to bring interpretative tools to content, but not the only way. Large projects will want tools that are more closely integrated and under their control, but for the small online publication eVoyeur can provide accessible functionality that resembles the badge tools that have proliferated on the web. The next step once we have responded to our editor interviewees with an improved micro-environment is to test the tool in real journals and blogs. To that end we are looking for projects willing to include eVoyeur in their interface.

Notes

[1] Voyeur Tools can be used at <http://voyeurtools.org>.

[2] Siemens, Ray et al. "A Study of Professional Reading Tools for Computing Humanists." A report at <http://etc.uvic.ca/public/pkp_report/>.

[3] For the TAToo (Text Analysis for me Too) project see <http://ra.tapor.ualberta.ca/~tatoo/>. For a discussion of this tool and our experiments see “Ubiquitous Text Analysis” in the Poetess Archive Journal, <http://paj.muohio.edu/paj/index.php/paj/article/view/13>.

Personal tools