Paper Reading #25: Twitinfo: aggregating and visualizing microblogs for event exploration

References
Twitinfo: aggregating and visualizing microblogs for event exploration by 

Adam Marcus, Michael S. Bernstein, Osama Badar, David R. Karger, Samuel Madden, Robert C. Miller.  Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.

Author Bios

• Adam Marcus
• Michael S. Bernstein is a graduate student focusing on human-computer interaction at MIT in the CSAIL.  His research is on crowd-powered interfaces: interactive systems that embed human knowledge and activity.
• Osama Badar is currently a member of the CSAIL at MIT.
• David R. Karger is a member of the CSAIL in the EECS department at MIT.  He is interested in information retrieval and analysis of algorithms. 
• Samuel Madden is currently an associate professor in the EECS department at MIT.  His primary research is in database systems.
• Robert C. Miller is an associate professor in the EECS department at MIT and leads the User Interface Design Group.  His research interests include web automation and customization, automated text editing, end-user programming, usable security, and other issues in HCI.

Summary

• Hypothesis
• TwitInfo can provide a useful tool for summarizing and searching twitter for information about events and trends.
• Methods 
• The researchers asked 12 participants to use TwitInfo to research different aspects of a recent event.  During this part of the process they gathered usability feedback and observed which interface objects were useful or ignored.  The second part of the testing involved adding a time limit.  Participants were given 5 minutes to research the event using Twitinfo and then 5 minutes to compose a report about their findings.  At the end of the session, the participants were interviewed about their reactions to the Twitinfo system.
• Results
• They found that participants were able to reconstruct reasonably detailed information about events even without prior knowledge of it.  They found that when users were performing the freeform exploration they tended to explore the largest peak thoroughly and read tweets completely.  They also drilled in on the map and followed links to related articles.  Most of the tweets were only used to confirm event details rather than generate the information.  When the time constraint was introduced, the focus shifted to skimming peak labels for a broad sense of the event a chronology, and a few people honed in on only one or two links to outside news sources to minimize time spent searching through repeated information.
• Contents
• This article mainly spends time explaining how Twitinfo works and the details that went into its creation.  It describes the user testing and results, and identifies several key trends in how people use the tool.  

Discussion
This article was not particularly convincing to me; while the authors successfully created their product, it didn't seem as useful as they'd thought it might be.  Frankly, I can't imagine it being used other than as a way to garner general public opinion on events because it is simply much easier to search for related news articles.  It might potentially be useful in a smaller group setting, perhaps if a high school student wanted to read some first-hand posts about a fight that went down after lunch or something.

Paper Reading #24: Gesture avatar: a technique for operating mobile user interfaces using gestures

References

Gesture avatar: a technique for operating mobile user interfaces using gestures by Hao Lu and Yang Li. Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.

Author Bios

• Yang Li is currently a Senior Research Scientist working for Google.  He spent time at the University of Washington as a research associate in computer science and engineering.  He holds a PhD in Computer Science from the Chinese Academy of Sciences.
• Hao Lu is currently a graduate student at University of Washington in Computer Science and Engineering.  He is also a member of the DUB Group.  

Summary

• Hypothesis
• Gesture Avatar can provide a useful, viable solution to the problem of imprecise finger input on touch-screen surfaces.  The paper also presents three separate hypothesis regarding GA's relation to Shift: 
• • (H1) Gesture Avatar will be  slower than Shift on larger targets, but faster on small targets. 
  • (H2) Gesture Avatar will have fewer errors than Shift. 
  • (H3) Mobile situations such as walking will decrease the time performance and increase the error rate of Shift, but have little influence on Gesture Avatar
• Methods
• Participants were asked to test both Shift and Gesture Avatar, with half of the group starting on one technology and the other half starting with the other.  Participants were also asked to complete tasks both sitting and walking.  To begin with, they were asked to select different targets of varying size, complexity, and ambiguity and and performance time was measured between the tap of the start target and the selection of the final target.  The researchers aimed to address both letter ambiguity and and commonness by using 24 different letters and controlling the distance between the objects and the number of letters used.
• Results
• Gesture Avatar was much slower than Shift when target size was 20px, but faster than shift when target size decreased to 10px.  The two techniques were approximately equal at 15px, and both techniques increased in speed with larger target sizes.  Shift with MobileState was faster when sitting than when walking, but Gesture Avatar was roughly the same between the two activities.
• Contents
• This paper presents and explores Gesture Avatar, an application designed to combat the issue of imprecision from finger-based touch screen technology.  The authors developed their product on an Android and tested it against the current Shift technology to better understand where it was limited and where it excelled.  Overall, their results match their hypotheses and the product itself received a positive reception from test subjects.

Discussion
I think the researchers did a very good job of testing and presenting this project.  I am glad to see that there is significant strides being taken in this area of precision as it is a prevalent issue among touch-screens today.  I don't have any problems with the paper itself; it is well written, thoroughly researched and tested, and logically presented.

Paper Reading #23: User-defined Motion Gestures for Mobile Interaction

References
 User-defined Motion Gestures for Mobile Interaction by Jaime Ruiz, Yang Li, Edward Lank.  Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.


Author Bios

• Jaime Ruiz is currently a fifth-year doctoral student in the HCI Lab in the Cheriton School of Computer Science at the University of Waterloo.
• Yang Li is currently a Senior Research Scientist working for Google.  He spent time at the University of Washington as a research associate in computer science and engineering.  He holds a PhD in Computer Science from the Chinese Academy of Sciences.
• Edward Lank holds a Ph.D. in Computer Science from Queen's University.  He is currently  an Assistant Professor in the David R. Cheriton School of Computer Science at the University of Waterloo

Summary 

• Hypothesis
• Although modern smartphones contain sensors to detect three-dimensional motion, there is a need for better understanding of best practices in motion-gesture design.  
• Methods
• 20 participants were asked to design and perform a motion gesture with a smartphone device that could be used to execute a task on the smartphone.  These gestures were then analyzed and several were selected to be included in the rest of the study.  In the following experiment, the participants were given a set of tasks and and a set of motion gestures.  The participants performed each gesture and rated them based on how well the gesture matched the task and how easy it was to perform.  
• Results
• Participants tended to design gestures that mimicked normal motions, such as putting the phone to their ear.  They also were able to relate interacting with the phone to interacting with a physical object, such as turning a phone upside down to hang up like the old phones.  Tasks that were considered opposites always resulted in similar gestures, but performed in opposite directions.  A diagram of the resulting gesture set is shown at the top of this post.
• Contents
• The paper begins by running an experiment to determine how participants feel gestures ought to be mapped to make them easy and intuitive.  The results were fairly consistent, though some unexpected.  After the research, the paper describes a second question: the set of parameters manipulated by the participants.  They determine that there are two different classes of taxonomy dimensions: gesture mapping and physical characteristics.  Gesture mapping can be further interpreted based on metaphor, physical, symbolic, and abstract.

Discussion

I felt that this was a fascinating read and is a good example of progress in the mobile interaction arena.  The hypothesis and testing was a lot more open ended than many of the previous papers, but I feel that this approach lent itself to a better understanding overall.  I think the authors achieved their research goals, but I would also be interested to see follow up studies to verify their results.  

Paper Reading #22: Mid-air pan-and-zoom on wall-sized displays

References
Mid-air pan-and-zoom on wall-sized displays by Mathieu Nancel, Julie Wagner, Emmanuel Pietriga, Olivier Chapuis, Wendy Mackay.  Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.


Author Bios

• Mathieu Nancel is currently a PhD student in HCI in the Université Paris-Sud XI under the supervision of Michel Beaudouin-Lafon and Emmanuel Pietriga.
• Julie Wagner is a PhD student in the insitu lab in Paris, working on new tangible interfaces and new interaction paradigms at large public displays.
• Emmanuel Pietriga is currently a full-time research scientist working for INRIA Saclay - Île-de-France.  He is also the interim leader of INRIA team In Situ.
• Olivier Chapuis is a Research Scientist at LRI.  He is also a member of and team co-head of the InSitu research team.  
• Wendy Mackay is a Research Director with INRIA Saclay in France,  though currently on sabbatical at Stanford University. She is in charge of the research group, InSitu. 

Summary 

• Hypothesis
• The main hypothesis of the paper is that there is a need for more research on complex tasks when dealing with high resolution wall-sized displays.  The authors made seven smaller hypotheses about how people best interact with tools.
1. Two hands are faster than one
2. Two-handed gestures should be more accurate and easier to use.
3.  Linear gestures should map better to the zooming component, but should eventually be slower because of clutching.
4. Users will prefer clutch-free circular gestures.
5. Techniques using fingers should be faster than those requiring larger muscle groups.
6. 1D path gestures should be faster, with fewer overshoots than techniques with lesser haptic feedback.
7. 3D gestures will be more tiring.
• Methods
• They conducted an experiment with 12 participants based on three primary factors: handedness, gesture, and guidance.  They controlled for potential distance effects by introducing the Distance between two consecutive targets as a secondary factor. The pan-zoom task involved navigating through two groups of concentric circles, starting at a high zoom level and zooming out until the neighboring group is visible.  Then they pan and zoom until they reach the target group.
• Results
• The data from the pan-zoom task strongly supported the first hypothesis, as well as numbers 5 and 6.  They were surprised to find that their third hypothesis was not supported; linear gestures are faster than circular ones.  Finally, as they expected, participants found that hypothesis 7 held true, and that IDPath guidance was least tiring but 3DFree was most tiring.
• Contents
• The paper discusses how users might best interact with a large screen by studying several different motions and commands.  They wanted to observe ease of use, causes of fatigue, and how simple or complex the interactions might be.  They proposed several ideas at the beginning to guide their research, and then performed an experiment to highlight the key points.  They found that while most of their hypotheses were supported, they had judged one or two points inaccurately.

Discussion
    I think this was a very well put together paper.  I feel that the authors did a good job defining their parameters and goals, and then delivering the results thoroughly.  I think it would have been interesting if they had done a little bit more experimentation, maybe trying a few different approaches to get the information they were looking for.  However, overall I was impressed by the research and findings.

Paper Reading #21: Human model evaluation in interactive supervised learning

References
Human model evaluation in interactive supervised learning by Rebecca Fiebrink, Perry R. Cook, and Daniel Trueman. Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems


Author Bios
Rebecca Fiebrink is currently an assistant professor in Computer Science at Princeton University. She holds a PhD from Princeton and was a postdoc for most of 2011 at the University of Washington.
Perry R. Cook is a professor emeritus at Princeton University in Computer Science and the Department of Music.   He is no longer teaching, but still researches, lectures, and makes music.
Daniel Trueman is a musician, primarily with the fiddle and the laptop.   He currently teaches composition at Princeton University.


Summary

• Hypothesis
• Because model evaluation plays a special role in interactive machine learning systems, it is important to develop a better understanding of what model criteria are most important to users.
• Methods
• The authors performed three studies of people applying supervised learning in their work.  In the first study they led a design process with seven composers to focus on refining the Wekinator.   Participants met regularly to discuss the software in relation to their work and suggest improvements.  In the second study, students were told to use the Wekinator in an assignment focused on supervised learning in interactive music performance systems.   Specifically, they were asked to use an input device to create two gesturally controlled music performance systems.  The third study was a case study completed with a professional musician to build a gesture recognition system for a sensor-equipped cello bow.  The goal of this study was to build a set of gesture classifiers to capture data from the bow and produce musically appropriate labels.
• Results
• In the first study participants found that the algorithms used to control the sound were difficult to control in a musically satisfying way using either a GUI or an explicitly controlled sequence.  Unlike the first study, the second and third both made some use of cross validation.  Users in the second study indicated that they considered high levels of cross validation accuracy to be indicative of good performance, and made use of it as such.  In the third study, however, it was used more as a quick way to check.  The people in all three studies used direct validation much more frequently than cross validation.  The direct validation was broken down into six categories: correctness, cost, decision boundary shape, label confidence and posterior shape,  and complexity and unexpectedness.
• Contents
• The researchers present work studying how users evaluate and interact with supervised learning systems.  They examine what sort of criteria is used in the evaluation and present observations of different techniques, such as cross-validation and direct validation.  The purpose of the research is both to make judgments of algorithm performance and improve training models, in addition to providing more effective training data.

Discussion
     I think this paper did a very good job of presenting findings and being thorough with the research and methodology.  By my estimation the researchers did accomplish their goal of gathering useful data regarding evaluation of supervised learning systems, and I think that this work will be very beneficial in the future.  I did not find any gaping faults with the paper itself; it presents its purpose, carries out the research and gives the findings, and even discusses potential benefits and uses.

Paper Reading #20 : The Aligned Rank Transform

References
The Aligned Rank Transform for Nonparametric Factorial Analyses Using Only Anova Procedures by Jacob O. Wobbrock, Leah Findlater, Darren Gergle, and James J. Higgins.  Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.


Author Bios

• Jacob O. Wobbrock is currently an Associate Professor in the Information School and an Adjunct Associate Professor in the Department of Computer Science & Engineering at the University of Washington. 
• Leah Findlater is a postdoctoral researcher in The Information School, working with Dr. Jacob Wobbrock. She holds a PhD from the University of British Colombia.
• Darren Gergle  is an Associate Professor at Northwestern University and has a PhD from Carnegie Mellon University.
• James J. Higgins is currently a Professor at Kansas State University and holds a PhD from the University of Missouri-Columbia.

Summary
Hypothesis
    The Aligned Rank Transform is a useful and easily accessible tool for pre-processing nonparametric data so that it can be seen and manipulated on a level beyond current nonparametrics tests.

Methods
     The ART procedure consists of 5 steps: 

1. Computing residuals: for each raw response Y, compute residual = Y - cell mean
2. Computing estimated effects for all main and interaction effects: these are calculated such that Ai is the mean response Yi for rows where factor A is at level i. AiBj is the mean response Yij for rows where factor A is at level i and factor B is at level j. And so on.
3. computing the aligned response Y', assigning average ranks Y'' where Y' = residual + estimated effect.
4. performing a full-factorial ANOVA on Y''.

Results
     The paper re-examined three different studies using their ART procedures to demonstrate its usefulness.  The first case showed how ART can uncover interaction effects that may not be seen with Friedman tests.  The second case showed how the ART can free analysts from the distributional assumptions of ANOVA.  The last case demonstrated the nonparametric testing of repeated measures data.  
Contents
     The authors presented their Aligned Rank Transform tool, which useful for the nonparametric analysis of factorial experiments and makes use of the familiar F-test.  They discuss the exact process in detail, then go on to show three examples of where it could prove useful and effective with real data.


Discussion
I am not particularly qualified to comment much on the usefulness or ingenuity of this project, but it seems to me like the authors did a fine job of creating a tool or technique to handle data that was previously more cumbersome and less obvious.  I cannot find any faults with the paper, and I think that they did a good job of selecting several different test cases to highlight particular areas of usefulness with the ART.  

Paper Reading #19 : Reflexivity in Digital Anthropology

References
     Reflexivity in Digital Anthropology by Jennifer A. Rode.  Published in the CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems.


Author Bio
     Jennifer Rode is currently an Assistant Professor at Drexel's School of Information in Pennsylvania.  She is also a fellow in Digital Anthropology at University College London.  She holds her PhD from the University of California, Irvine.

Summary

• Hypothesis   Rode believes that digital anthropologists can contribute to the field of HCI by writing reflexive ethnographies, which is different from other more positivist approaches.

• Methods  -  This paper was more of a discussion than an actual research and development project, so the author did not have any methods to present.   

• Results  -   The author posits that digital anthropologists are not studying technology, but rather studying in the context of technology.  She also clarifies some of the traditional approaches to ethnographic study, namely Positivist versus Reflexivity.  In addition to the approaches, she described some of the writing styles, specifically Realistic, Confessional, and Impressionistic. 
• Positivist: Data is collected, studied, and tested with the aim of producing an unambiguous result.  
• Reflexivity:  According to Burawoy, reflexivity embraces intervention as an opportunity to gather data, it aims to understand how the data gathering impacts the data itself, and reflexive practitioners look for patterns and attempt to draw out theories. 
• Realistic: 
  the need for  experimental 
  author(ity), its typical forms, the native’s point of view, and interpretive omnipotence.
• Confessional: 
  broadly provides  a written 
  form for  the  ethnographer  to  engage  with  the  nagging doubts surrounding the study and discuss them textually with the aim of demystifying the fieldwork process
• Impressionistic: based on dramatic recall and a well told story.

• Content  -  The author showed how ethnography has various forms and orientations, and how reflexivity can contribute to design and theory in HCI.  She also describes three forms of anthropological writing and the key elements of the technique.  Finally, she describes how ethnography is actually used in the design process of computer-human interaction.

Discussion
This paper, while surely valuable in its field, is extremely hard to read.  There is a lot of information and it seems to me that it could have all been summarized much more succinctly.  I don't know who she expects to actually read the entire paper all the way through and understand everything that she is going on about, because it is simply too much.  The paper serves as a good overall synopsis of a lot of different approaches in the ethnographic realm, but I feel it has virtually no relevance outside of that.