For the last year, SIMS students have engaged in online discussion
on their own IRC channel. Interactions occur both inside and outside of class, when
users are co-present or in separate physical environments. Conversations sometimes augment class
discussion, allowing for different types of in-class participation, as well as
the sharing of relevant material, URLs, etc. We are seeking to enable
professors and teachers to visualize chatroom discussions in order to evaluate
how successfully they are teaching and how well students are comprehending the
material.
In addition to examining SIMS-specific IRC data, we have explored
the benefits of such a visualization for educators wishing to understand the
relation of chat activity to the classes they teach. Some courses now
incorporate chat into the classroom, by creating sanctioned IRC channels for
students to engage with one another. A visual analysis of log data from such discussions may prove interesting or
useful for professors. While we have not created a fully-functioning tool, we have created a prototype visualization
using SIMS data in order to examine the possibilities of such a tool.
We expected to see fairly consistent daily usage patterns,
with highs and lows at specific times of day. We also expected that academic events within SIMS (exams, papers
assigned or due) would produce sharp peaks that fell outside of these
patterns. We also anticipated that certain classes would generate more participation and higher numbers of users. We
hoped to analyze these possibilities by implementing a proof of concept and
performing preliminary testing to suggest that these types of results could be
useful for educators. Below are some hypotheses that we used to help direct our
design and user tests:
• Implementations of the backchannel will and should vary across different contexts and domains
• Teaching styles should take advantage of the social and educational affordances the backchannel offers
• Chatrooms should enable teacher self-assessment
• A backchannel should encourage social interactions and community building
• A backchannel etiquette will need to be developed
Research has been done on the role of online backchannel
discussions that take place among co-present participants[i] ,
such as attendees of an academic conference[i]
This research is highly relevant and uses
chat log data to draw conclusions about the role of this type of communication.
However, we have found little research on the role of these discussions in the classroom setting. Many of the studies on the use of
technology-mediated discussion in the classroom focus on the overall effects of
multi-tasking on student performance
[iii] or how technology may be incorporated to encourage discussion
[iv], rather than an examination of the patterns of online discussion within the classroom
setting.
How do we enable a quick and intuitive response that requires little effort for a professor or teacher?
Coterie (Figure 1) and PeopleGarden (Figure 2) are examples of semantic visualization,
where the meaning and relevance of the data is conveyed in the design of the visualization.
Semantic visualizations will help a professor to visualize the social and academic environment in the chatroom in order to
better understand how well he or she is teaching and how well the students are learning the material.
Although this information could conceivably be displayed through a simple table with student names and number of contributions,
we feel that a semantic visual design could be a more pleasing and desirable method for a professor to view this information.
It would facilitate a quick mental depiction of the social dynamics and the material being discussed.
PeopleGarden offers an interesting metaphor of life, growth, and health, through its use of flowers, petals, and gardens as visual representations.
We will look to explore similar metaphorical visual icons that can be used to convey academic properties to a professor.
Coterie
Figure 1. A Coterie[xvi]
display showing three simultaneous
conversational threads: one related to aircraft, one about screens, and one
with a comment about a previous statement’s usage. Six users are currently
active, but many more are listening.
In Coterie, one can see at a glance how many people are logged into a channel. The participants are
represented as colored ovals that bounce and become brighter when that person
speaks. This highlights the active participants and conveys the vitality of the
discussion. By visualizing presence and conversational activity, Coterie
creates an environment that looks like the conversation—coherent discussions
form a solid, central core and scattered chats are, well, scattered all over the
place. One easily senses how populated the channel is and who are the primary
participants.
PeopleGarden
Figure 2. Two PeopleGarden[v] visualizations
Left: a group with many members at different levels of participation.
Right: a group with a single dominant member.
PeopleGarden uses a
flower and garden metaphor to visualize participation on a message board.
Participants are each represented by a flower. The longer they have been
involved, the higher the stem; the more they have posted, the more petals.
Initial postings are shown in red, replies in blue. One can easily get a sense
of an individual’s role as an active participant or a long-time lurker.
QueryLines
QueryLines: approximate query for visual browsing [xiii]
QueryLines approximates query techniques for searching and
analyzing two-dimensional data sets such as line or scatter plots. The
techniques allow users to explore a dataset by defining QueryLines: soft
constraints and preferences for selecting and sorting a subset of the data. By
using both preferences and soft constraints for query composition, it allows
greater flexibility and expressiveness than previous visual query systems. When
the user over-constrains a query, for example, a system using approximate
techniques can display "near misses" to enable users to quickly and
continuously refine queries.
Although the concept behind QueryLines is useful, we rejected the idea of soft constraints for our visualization
because we feel that it is too complex and detailed. While a professor could certainly be capable of using a similar feature,
they likely will not want the time overhead of deducing when a soft constraint query is necessary.
With our goal of creating a simple and intuitive design
we think it is better to have fewer visualization options as a tradeoff for a more clear visualization design.
Figure 3. Example queries for a sample dataset: (a) all the
plots, (b) min and max hard constraints prune the set of exact
matches, (c) an over-constrained query (i.e., no exact matches)
with two soft matches, and (d) a preference, goal query.
Milestones in Time
Figure 4. The Milestones in Time: The Value of Landmarks in Retrieving
Information from Personal Stores[vi] showed that users experienced a faster search time when the
timeline contained landmarks than when it didn’t. Although the context of this
study is slightly different, we suspect that similar trends might be seen in a
classroom environment because a professor’s lecture content is still personal
to him or her. This indicated to us that professors might want to overlay
topical lecture information, as a type of “landmark”, onto the graph to more
quickly searches for the information they are looking for.
We displayed topics in our test file using TimeSearcher, see Figure 16,
as one method in which professors could quickly visualize topics.
This topic is further discussed in our Conclusions and Future Work section.
How should the professor select a time frame to visualize?
We want to enable dynamic querying, in which the professor can select a range of time frames with
real-time feedback in the visualization. We want to enable the following components, as described in Professor Marti Heart's Dynamic Querying lecture:
• Visual presentation of query's components
• Visual presentation of results
• Rapid, incremental, and reversible control
• Selection by pointing, not typing
• Immediate and continuous feedback
• Support browsing
• Details on demand
Based on the related work below, we suspect that the two best options for displaying time frames are drop-down boxes or sliding scales or a combination of the two.
We will explore the different features offered by each of these methods.
The following project describes the use of timeboxes, which are like drag-and-drop boxes, to enable selection of a time frame.
Timesearcher
Timesearcher: Visual Exploration of Time-Series Data [xiv]
Timesearcher is prototype environment for interactive querying and exploration of time-series data.
Queries are built using timeboxes, a graphical, direct-manipulation metaphor for the specification of queries over
time-series datasets. These timeboxes support interactive formulation and modification of queries.
The prototype includes windows for timebox queries, individual
time-series, and details-on-demand. Other features include drag-and-drop support for
query-by-example and graphical envelopes for displaying the extent of the entire data set and result set from a given query.
Query manipulation tools including pattern inversion and 'leaders
& laggards' graphical bookmarks provide additional support for interactive
exploration of data sets.
The properties of the timebox are further described in Dynamic query tools for time series data sets: timebox
widgets for interactive exploration[xvii].
They are rectangular widgets that can be used in
direct-manipulation graphical user interfaces (GUIs) to specify query
constraints on time series data sets. Timeboxes are used to specify
simultaneously two sets of constraints: given a set of N time series
profiles, a timebox covering time periods x1...x2(x1<x2) and
values y1...y2(y1≤y2) will retrieve only those n ∈N that have values y1≤y≤y2
during all times x1≤x≤x2.
Figure 5. The screenshot of Timesearcher below shows a visualization of students in a classroom.
We created this visualization with a text file formatted for use in TimeSearcher that documented student names and number of entries.
The top left window shows student participation, over a single class lecture, with an orange timebox drawn to select a specific time frame.
The window on the top right displays time periods during the lecture with lecture topics during a specific time period.
The bottom right window shows student names. Clicking on these names highlights their participation on the graph.
A Temporal Communication Flow Visualizer for Social Network Analysis[vii]
This project describes a sliding time frame algorithm to visually analyze the
evolution of communication patterns over time. The layout of the graph is automatically
recalculated every day, resulting in an interactive movie.
The essential concept of the sliding time frame algorithm is to display active ties between actors in a
sliding time frame covering a flexibly selected interval of n days starting
from the specific day the visualization is showing. The window frame moves
forward day by day, and new ties are subsequently added to the graph each day
until the desired width n of the sliding time frame is reached. This time frame
window allows users to foresee the activities happening inside the time frame
after the current day. By default, all the old communication activities before
the current time frame window are included in the layout of the graph.
The sliding time frame concept would be a useful way of allowing a professor to select a particular period of time.
It would not need to be limited to day-long periods, but could also be adjusted according to individual lectures, weeks, or semesters.
Figure 6. The time frame moves to day d as the animation progresses. Day d is the current
day that the visualization is showing and the current time frame is [d, d+n].
All communications through day (d+n) are calculated and displayed, and if a
communication takes place before or on day d, it is active.
Figure 7.
In the “No history” mode, only the edges in the current window are included. The time
frame moves to day d as the animation goes. Thus, day d is the current day that
the visualization is showing and the current time frame is [d, d+n]. Only communications
inside the current time frame are calculated and displayed, and only
communications on day d are considered active.
See Appendix B for more related visualization examples.
We used existing visualization tools such as
Spotfire and Tableau to perform our data explorations. For the prototype design of our visualization, we used Photoshop combined with HTML and Javascript.
For next semester, we are investigating other tools such as Java Applets, Flash, or AJAX, but do not feel we are at a point to
commit to any particular tool or level of interactivity.
The data we will use is made up of chat logs for the IRC
channel from October 2004 to the present. Because some of the discussions that
take place on the channel are private or sensitive, the privacy of all
participants has been protected by changing their usernames. The chat logs contain a date and time stamp,
user names, and log entries. There are also robot and channel operator entries.
Nov 21 10:51:32 <tamara_> "What is the matrix?"
Nov 21 10:51:52 <benji>this is pretty cool
Nov 21 10:52:35 <jen>I want to know if they will display rich media types
Nov 21 10:52:35 < jen >ie. pictures and such
Nov 21 10:52:42 < benji >the xml and such
Nov 21 10:52:43 <whitney>good question jen
Nov 21 10:55:19 <gopher>benji jen: are we supposed by buy Johnny's book for the reading?
We have used this data to extract information about times (which can be overlaid
with class times, important dates, etc), users, and lines of contribution
within a discussion.
Figure 8. Total Chat Entries By User Between Oct 2004 and Oct 2005
Any member of the SIMS community has access to the backchannel.
Most students have entered the backchannel at least once, although not all of them have and some only enter it sporadically.
A visualization of users indicates that usage follows a Power Law curve.
Usage follows an exponential drop off with a gradual decline at the tail end of the curve.
Figure 9. Trends in Chat Log Usage over a 24 Hour Time Frame
Participants in the SIMS chatroom are driven by a desire for a sense of community.
People are "searching for a feeling of community that's been lost as many "third places"
(not work, not home, but a third place where people congregate and interact) have closed down."
There are many indicators of the SIMS backchannel as a third place.
Participant usage increased during class time but also in the evenings, as this graph shows.
This suggests that an in-class chat tool could be enhanced by out of class uses, such as study sessions.
Figure 10. Peaks in Usage During Particular Class Hours in the Fall of 2004
Chat use peaks during class hours, indicating that students voluntarily choose to participate in the chatroom while in class.
Student motivations will have to be further analyzed to determine how to facilitate similar motivations when the backchannel is
an academically drive environment, rather than a combination of academic and social.
Figure 11. Number of Chat Entries Over First Six Weeks of Spring 2005 Academic Semester
User count is plotted versus the first six weeks of the Spring 2005 academic semester.
Visualizations of user entries versus time periods highlights adoption curves over time.
The visualization suggests that community participation increases as the academic semester progresses.
We conducted three informal interviews of SIMS professors to determine their perspectives on a backchannel discussion.
These were not formal interviews, nor did we ask about visualizations in particular. Given the novelty, but relevance, of this topic
we wanted to acquire a general idea of professor's view of the backchannel before begining our design process.
The professors we interviewed were Bob Glushko, Larry Downes, and Ray Larson. The key points from our discussions are highlighted here.
Because the nature of the backchannel could potentially cause concerns about privacy and personal involvement, we chose to compile the salient points from our interviews in an anonymous format.
• Disconcerting sometimes because we don't know what's happening in that backchannel.
• When a whole bunch of people start smiling broadly or snickering, you sometimes wonder if you said something weird or what.
• "You're speaking Dutch today, we [students] don't understand you."
• Depends on style of course - 200 people, lecture-style versus small discussion seminar.
• Elicit certain kinds of problem solving skills that might be valuable.
• Should it be anonymous?
• Don't want to stifle legitimate discourse, but want to stifle illegitimate discourse. What's the right metrix?
• How can it encourage more class discussion?
• Students could double-check references on google.
• Measure of coherence based on density and bandwidth of class. "Would you explain that in English?"
• Might be useful, interesting possibility that people can actually combine lecture and debate at the same time.
• Why would you need a backchannel? Is there a communication void?
• Disconnect of not knowing what's going on.
• What's missing from main conversation? Main conversation is limited by size, distance, features.
We conducted three interviews of SIMS professors and TAs using our visualization prototype. Again, to retain the privacy of the individuals, we compiled their feedback into a set of key points in an anonymous format.
• In a small class, could have names overlayed directly on the lines or mapped to a list of names below instead of having to hover over each line to see who the user is.
• Like seeing total participation in black
• Hard to distinguish between overlapping pastel lines.
• Useful if hovering over a line would make it brighter and larger, or more easy to visualize that individual line.
• Don't like drop-downs!
• Drop-downs would take too many clicks
• Could have clickable calendar with time and duration selection options.
• Could have sliding scale that adjusted real-time, ie between month/day/hour, etc.
• Sliding scale could have beginning time on left end of scale and ending time on right end of scale.
• Would be useful to have average participation rates to provide some context for this lecture or time frame as compared to average usage.
• Use NLP to visualize patterns, such as when a students posts a question and others post links to URLs as answers.
• Need to better indicate what high and low participation rates actually mean.
• Will the lines always be as smooth as they are here? What if a student only contributes one comment? Perhaps they can be compiled over a particular scale, such as 8 15-minute sample points in a two hour period or 4 week-long sample periods over a month.
• How to visualize discourse?
• Difficult to track or make a judgment about what's going on with a particular user.
• Drill-down through visualization to see logs - drill down into time period to get to actual log entries
• Zoom in on context, slice by time, such as a two minute period with a lot of activity and remember "oh, that was that joke I did!"
• How could it be used to run real-time in the classroom and displaying what's happening?
• Need more context to know what students are talking about.
• If it's more like an official IRC then people will stay more on topic.
• Helpful to indicate if people understand the material.
• Teachers could use to help determine grades based on class participation.
Our main goal was to design a visualization that would allow instructors to explore data related to in-class IRC discussions. Past experience in designing software for professors has shown that there is a wide range in terms of attitudes and experience with technology. It has also shown that professors consistently desire tools that are easy to use and efficient. Though the purpose of visualization is often exploratory rather than task-driven, we nevertheless recognized the need to create a tool that would be intuitive enough to use at a glance.
Aside from the usability of the tool, it was also imperative that we examine what views of the data would be most interesting to an instructor. We determined that the following aspects would be relevant for our purposes:
Discussion during class. Instructors may wish to visualize discussion over the course of a particular class or lecture.
Discussion over the semester. Instructors may also wish to view trends in discussion over an entire semester.
User distribution. This would be especially beneficial to instructors wishing to view discussion on an individual basis, e.g. to determine participation rates of students.
Relation of chat to course material. Instructors may want to draw links between material and chat participation.
From this analysis, we realized that the most important area of focus for our visualization would be viewing trends over time. This would include viewing discussion over short periods (a single class) as well as long term trends (a semester or year view). We chose to use trends over time as the major focus of our tool, coupled with the ability for the user to zoom in order to adjust the time period. Because we aimed to keep the visualization simple and clean, we kept the design spare, with additional details available on demand, when the user mouses over or clicks a section of the visualization.
In the design of our visualization, we encountered several interesting design challenges. First, choosing a method for zooming that was simple, apparent, and efficient was difficult. We decided to begin with drop down menus, as we felt these would be simple for users to understand and use, though perhaps not the best use of visualization techniques. We hoped to refine the zooming technique with user feedback, and plan to create a more visually-based zoom feature as part of our future work.
We also thought that text analysis might be an interesting feature to incorporate, possibly supplying users with a list of frequent words, or using it to draw correlations between discussion content and course topics. After a brief analysis of common words in our data set, we realized that this would be a tremendous challenge, but one we are certainly still interested in experimenting with.
The idea of overlaying class times and important events (such as exams and paper due dates) with log data also seemed promising. We examined the possibilities of collecting such data, including user tagging or manual entry from another source. This is another feature we would like to explore further in our future work.
Lastly, we found the idea of users tagging their own content to be interesting. For example, users entering a relevant question in the channel during lecture could tag the line with a [question] or similar tag, which the system could use to parse the content for instructor analysis. This is an idea that would certainly need to be tested thoroughly with users.
Prototype
The first iteration of our prototype was a simple mockup, created with Photoshop. It's intent was to elicit initial feedback, and convey the idea of interaction techniques. In Figure 12, the user is mousing over one of the graph lines, which displays a tooltip with the name of the individual user (in this case, Sarita Yardi).
Figure 12. First mockup, created with photoshop.
Next, we moved to creating a more interactive prototype. This prototype used simple HTML and Javascript, coupled with wizard of oz techniques to give the appearance of interaction. The user could mouse over the endpoints on each line to get information about the user. During testing, the participants were asked to zoom to a specific date, between 9:00 am and 1:00 pm, at which point they were taken to a zoomed in visualization.
Figure 13. Interactive prototype screen.
Figure 14. Zoomed in view of interactive prototype, after the user selects the more narrow date and time range.
Our primary question in designing the visual data presentation was: Which visual components make it easiest for a professor or teacher to visualize the data in the most appropriate ways? We hoped to create a tool that was simple enough in appearance not to overwhelm the average user, while still providing the necessary features for full exploration.
We selected a simple line graph
visualization because we thought it to be both intuitive and familiar to professors and
teachers. We were looking to balance between novelty and visual appeal with functionality
and efficiency. We would lean towards more emphasis on the latter constraints
in order to increase the likelihood that professors would feel comfortable
using the tool. We liked Martin Wattenberg’s BabyNames visualization as a model
because it is easy, simple, and intuitive. Minimal instructions are needed and
the learning curve is quick. Furthermore, the colors are visually pleasing and
easy on the eye for continuous interaction.
We felt that zooming would be an important feature of our tool, and hope to expand on this component in the future. Our prototype used a simple drop down menu for zooming in or out, though our research on existing tools has given us alternate ways methods of zooming through time visually.
We also implemented brushing so that
the user could quickly and intuitively scan the graph for salient information.
For example, he can search by specific student, by time scale, or by
discussion. All of these can be done by simply brushing the mouse over the
visualization and pausing on the relevant curves. In particular, brushing is
useful to display point connections between closely connected threads, to
better understand which conversations were relevant and active, based on the
professor's lecture. Highlighting provided a very quick, high-level ability to
visualize students who participated in a particular chat. We also used a details on demand feature to keep the interface free of extraneous information, and give the interface a simple, uncluttered appearance. By mousing over relevant points on the visualization, the user can obtain specific information that is relevant to him or her, and thus explore the data effectively.
Through our design and testing of this visualization, we found that creating a meaningful visualization of discourse is extremely challenging. Not only is the data itself complex, but discovering the needs and goals of users and developing a tool to address those needs also proved difficult. This may be in part due to the exploratory nature of visualizations, as opposed to designing software with more well-defined and specific tasks. Nevertheless, exploring those needs has been highly enlightening and we have made significant progress in our conceptualization of the tool and its design.
With our design, we hoped to create a prototype for a tool that would be simple and efficient. Through user testing, we discovered that our method of zooming was not preferred, which was an expected result. Users did not like the drop downs, and instead suggested sliding scales or calendar-like interfaces. We also found that our colors were not distinguishable enough, possibly due in part to browser displays, and that users would prefer highlighting methods to aid in tracking a single user. Users understood the complexity of correlating discussion content with a visualization, but also felt that it would make the visualization more useful.
We plan on continuing the development of our visualization, incorporating user feedback into the next design phase. Because our design method used simple HTML and JavaScript, it was fast and flexible enough to accommodate significant design revisions. We plan to incorporate features requested most by our participants, such as visually-based zoom, highlighting, and text analysis.
We also hope to develop a method for viewing the data in alternate ways, perhaps by selecting specific students and time periods. Instructors should be able to visualize topics quickly, and associate them with the data set. We would like to enable more complex dynamic querying through visual components, as in Figures 15 and 16. This allows instructors to choose a query through selection, and be presented with a visualization of query results. We will also explore alternate means of zooming, including the sliding time frames and drop boxes discussed previously, as well as the calendar-like interfaces suggested by test participants.
We expect that our design will change significantly as we develop it further and incorporate more features, but our initial visualization design has given us tremendous insight into the possibilities of visualizing chat. We now have a solid conceptual framework for refining a prototype tool, and hope to explore the possibilities of utilizing such a tool in a general classroom setting.
Figure 15.
Left: Student names
Right: Lecture times, topics, and number of entries for a particular student. While we did not incorporate these tools into our prototype,
we feel that there is potential for professors to use similarly structured menus to select student names and visualize their interactions.
These menus are simple and easy to use and would work well with the visualization we designed.
Figure 16. This shows each student's participation over the course of the lecture.
This visualization might be a useful option for professors to select when they wanted to zoom in on a particular user's behavior.
The menus in Figure 15 could be use for this purpose.
Figure 17. This picture shows 2nd grade students using laptops in the classroom to participate in a common chatroom.
It was taken in Nancy Durhake's classroom at the Concord Elementary School in Darien, Illinois.
[vxiii]
At the top left of the picture there is a television monitor which is displaying the chatroom discussion in real-time.
There is clearly an opportunity to expand on the role of a chat visualization tool outside of the SIMS community and into other environments,
such as the K-12 classroom. We hope that our visualization has provided solid guidelines for further development of such a tool.
Figure 21. Loom visualization.[ix]
The threading patterns in two groups. Related posts are placed in a circle.
Left: a group with almost no threading—posts are sent but seldom
replied to. The single dots scattered across the screen are a sign of a
nonsocial group.
Right: a group with dense threads. Numerous and dense circles
suggest a vibrant conversational arena. The circular form is suggestive of a
conversational gathering and the scattered dots lend themselves to an
interpretation of disconnection.
Figure 22. The original Chat
Circles [x]
interface. The local user’s circle is bordered in white (in this image, it is user
called magenta”, with the blue circle). Everyone is speaking, except for “it”
(in green). Only “ann” is inside of magenta’s hearing range.
Figure 23. Chat Circles history
interface. Each vertical line represents a user. The horizontal bars represent
posting, with hollow bars standing for messages posted outside the local user’s
hearing range. The text of postings, shown as solid bars, can be retrieved by
mouse-over.
Figure 24. In this paper we introduce Conversation Thumbnails
[xi],
a prototype discussion interface that provides skimming friendly navigation,
exploits available metadata to help spotlight important sections of the
discussion, and uses a special composition-time search mechanism to help maintain
conversational coherence. Figure 24 shows a portion of the interface for an illustrative
discussion. The interface has three parts: a navigable overview of the entire
conversation, a detail window displaying individual messages, and a composition
area for new messages.
Figure 25. After a query: The target is represented by the thick
black line; matching companies are shown with thumbnail graphs
Each time the user releases the mouse while drawing on the graph
entry panel, the applet performs the query and displays the results as a series of thumbnail graphs in the lower
portion of the screen, as seen in Figure 25. Several considerations turned out
to be important in designing the result display. First, it is helpful to
present miniature thumbnail graphs of the stocks retrieved by the query. These
snapshots provide instant, comprehensible feedback on the quality of the results. In addition,
moving the mouse over a thumbnail displays the full version of the thumbnail
overlaid on the main graph entry panel. This allows the user to compare the result
directly with their query. Each thumbnail is also accompanied by a numerical
indicator of how well it matches the target pattern.
Second, performing the query on mouse release seems to be the
most desirable degree of interactivity. Forcing the user to click a “submit”
button adds a gratuitous step to each query. Dynamic updates as the mouse moves
cause an uninformative and distracting shuffling of the results. Third,
it is important to allow users instant access to “drill down” on the thumbnails for more information.
Since the point
of the query is to find stocks to invest in, we designed the applet so that
clicking on a thumbnail brought the user to SmartMoney.com’s stock research web
pages.
Figure 26. The Netscan dashboard combines the thread tree, piano roll, reply-based sociogram views, message display,
and header information.
(A) The thread tree
visualization presents information about the structural and temporal history of
the selected thread. Selecting a message displays its contents in the message
pane (D) below.
(B) The piano roll component displays
a list of all posters who are present in the selected thread ordered by the
number of posts they contributed. Columns for each day the thread was active
contain bars if the poster posted on that day. When the user mouses over each
poster, the related posts are highlighted in the thread tree and the interpersonal
connection component.
(C) The interpersonal
connection component displays a sociogram that relates users with whose they
reply to and who reply to them. Posters are located based on the number of
responses they send (the x axis) and the number they receive (the y axis).
(D) The message display pane presents
the contents of a selected message and controls for replying via Usenet or
email.
Figure 27. The Netscan thread
tree for the thread with the subject “Visual Basic versus Visual
C++” from the microsoft.public.vb.general.discussion
newsgroup starting on July 13, 2000. In this view, boxes represent messages. A
line between two messages indicates that the more recent message is a reply to the
earlier message.
Figure 28. The Netscan piano
roll display presents a ranked list of authors by their total activity and an
indication of how much they posted on each day over the lifetime of a specific
thread.
Figure 29. The Netscan
interpersonal connections pane displays the authors in a thread and an indication
of how much they replied to and were replied from all other participants over
the lifetime of a specific thread.
Figure 30
[xii] illustrates the progress over time of the communication activities around preparing and
conducting a Webinar, i.e. a Web based conference. The picture in the top left
of figure 30 shows the structure of the team preparing the Webinar. This team
has high density and relatively low group betweeness centrality, as shown by
the first arrow in the center of the graph. The picture in the top right of
figure 30 shows a screen shot of the communication pattern during the first time
the Webinar was conducted. The practice leader (blue dot) is sending and
receiving information in a star structure, the graph in the center as pointed
out by the second arrow displays now relatively high centrality. The third
picture at the lower left displays the team preparing a rerun of the webinar, again communicating with relatively
low centrality (third arrow). The final screen shot in the lower right of
figure 30 displays the practice coordinator (red dot) rerunning the Webinar,
communicating in a star structure with relatively high centrality (fourth
arrow) with his audience.
[i]
McCarthy, Joseph F and d. boyd, 2004.
Digital Backchannels in Shared Physical Spaces: Attention, Intention, and Contention.
In Proc. Of the 2004 ACM Conf.
On Computer Supported Cooperative Work (CSCW 2004), 550-553.
[ii]McCarthy, Joseph F and d. boyd, 2005.
Digital Backchannels in Shared Physical
Spaces: experiences at an academic conference.
In Extended Abstracts of the 2005 Conference of Human Factors in
Computing Systems (CHI 2005), 1641-1644.
[iii]
Hembrooke, H. and G. Gay. 2003.
The Lecture and the Laptop: The Effects of Multitasking in the Classroom.
Journal of Computing in Higher Education, 15(1), 46-65.
[iv]
Ratto, M., Shapiro, R.B., Truong, T.M., and Griswold, W.G. (2003).
The ActiveClass Project: Experiments in encouraging classroom participation.
In Computer Support for Collaborative Learning, 2003.
[v]
Xiong, R. and Donath, J. 1999. PeopleGarden: creating data portraits for users.
In Proceedings of the 12th Annual ACM Symposium on User interface Software
and Technology (Asheville, North Carolina, United States, November 07 - 10,
1999). UIST '99. ACM Press, New York, NY, 37-44. DOI=
http://doi.acm.org/10.1145/320719.322581
[vi]
M. Ringel, E. Cutrell, S. Dumais, E. Horvitz. Milestones in Time: The Value of
Landmarks in Retrieving Information from Personal Stores, INTERACT 2003,
September 2003.
[vii]
Gloor, P., Zhao, Y.,
TeCFlow – A Temporal Communication Flow Visualizer for Social Network Analysis.
[viii]
Wattenberg, M. 2001. Sketching a graph to query a time-series database.
In CHI '01 Extended Abstracts on Human Factors in Computing Systems
(Seattle, Washington, March 31 - April 05, 2001). CHI '01. ACM Press, New York, NY,
381-382. DOI= http://doi.acm.org/10.1145/634067.634292
[ix]
boyd, d., Lee, H-Y, Ramage, D. and Donath, J. Developing legible visualizations for
online social spaces. In Proceedings of the Hawaii International Conference on System
Sciences. (Jan.7–10, 2002, Big Island, Hawaii).
[x]
Donath, J. and Viégas, F. B. 2002. The chat circles series: explorations in
designing abstract graphical communication interfaces. In Proceedings of the
Conference on Designing interactive Systems: Processes, Practices, Methods, and
Techniques (London, England, June 25 - 28, 2002).
DIS '02. ACM Press, New York, NY, 359-369.
DOI=http://doi.acm.org/10.1145/778712.778764
[xi]
Wattenberg, M. and Millen, D. 2003. Conversation thumbnails for large-scale
discussions. In CHI '03 Extended Abstracts on Human Factors in Computing
Systems (Ft. Lauderdale, Florida, USA, April 05 - 10, 2003).
CHI '03. ACM Press, New York, NY, 742-743.
DOI=http://doi.acm.org/10.1145/765891.765963
[xii]
Gloor, P., Laubacher, R., Zhao, Y., Dynes, S.
Temporal Visualization and Analysis of Social Networks.
[xiii]
Ryall, K., Lesh, N., Lanning, T., Leigh, D., Miyashita, H.,
and Makino, S. 2005. QueryLines: approximate query for visual browsing. In CHI
'05 Extended Abstracts on Human Factors in Computing Systems (Portland, OR,
USA, April 02 - 07, 2005). CHI '05. ACM Press, New York, NY, 1765-1768.
DOI= http://doi.acm.org/10.1145/1056808.1057017
[xvi]
Donath, J. 2002. A semantic approach to visualizing online conversations.
Commun.
[xvii]
Hochheiser, H. and Shneiderman, B. 2004. Dynamic query tools
for time series data sets: timebox widgets for interactive exploration. Information
Visualization 3, 1 (Mar. 2004), 1-18. DOI=
http://doi.acm.org/10.1145/993176.993177
Figure 1. A Coterie
Figure 2. Two PeopleGarden
Figure 4. The Milestones in Time: The Value of Landmarks in Retrieving
Information from Personal Stores
Figure 6. The time frame moves to day d as the animation progresses. Day d is the current
day that the visualization is showing and the current time frame is [d, d+n].
All communications through day (d+n) are calculated and displayed, and if a
communication takes place before or on day d, it is active. 
Figure 7.
In the “No history” mode, only the edges in the current window are included. The time
frame moves to day d as the animation goes. Thus, day d is the current day that
the visualization is showing and the current time frame is [d, d+n]. Only communications
inside the current time frame are calculated and displayed, and only
communications on day d are considered active.
Figure 8. Total Chat Entries By User Between Oct 2004 and Oct 2005
Figure 9. Trends in Chat Log Usage over a 24 Hour Time Frame
Figure 10. Peaks in Usage During Particular Class Hours in the Fall of 2004
Figure 11. Number of Chat Entries Over First Six Weeks of Spring 2005 Academic Semester
Figure 17. This picture shows 2nd grade students using laptops in the classroom to participate in a common chatroom.
It was taken in Nancy Durhake's classroom at the Concord Elementary School in Darien, Illinois.
Figure 18.
Figure 19.
Figure 20.
Figure 21. Loom visualization.
Figure 22. The original Chat
Circles 
Figure 23. Chat Circles history
interface. Each vertical line represents a user. The horizontal bars represent
posting, with hollow bars standing for messages posted outside the local user’s
hearing range. The text of postings, shown as solid bars, can be retrieved by
mouse-over.
Figure 24. In this paper we introduce Conversation Thumbnails
Viewing Results of a Query
Figure 26. The Netscan dashboard combines the thread tree, piano roll, reply-based sociogram views, message display,
and header information.
Figure 27. The Netscan thread
tree for the thread with the subject “Visual Basic versus Visual
C++” from the microsoft.public.vb.general.discussion
newsgroup starting on July 13, 2000. In this view, boxes represent messages. A
line between two messages indicates that the more recent message is a reply to the
earlier message.
Figure 28. The Netscan piano
roll display presents a ranked list of authors by their total activity and an
indication of how much they posted on each day over the lifetime of a specific
thread.
Figure 29. The Netscan
interpersonal connections pane displays the authors in a thread and an indication
of how much they replied to and were replied from all other participants over
the lifetime of a specific thread.
Figure 30 