Microsoft Research Connections Blog
Next at Microsoft
Social Media Collective
Windows on Theory
Posted by Allison Linn
A summary of Microsoft researchers' papers from CHI 2015 are listed below.
About 70 Microsoft researchers are traveling this week to ACM CHI 2015, the premiere conference on human-computer interaction, to present research results and tools that can do everything from create an interactive, 3D map of your neighborhood to simply deliver you better web search results.
Microsoft’s researchers will be presenting papers, participating on panels, organizing the doctoral consortium and leading plenary sessions at this year’s conference.
The annual conference, which is being held this year in Seoul, comes as researchers at Microsoft and elsewhere are finding ways to make computing even more personal, catering to the very specific needs of surgeons in the operating room, CEOs in far-flung boardrooms and students attending online lectures.
That means figuring out what people need from technology, and also what they don’t.
See also: Abstracts and downloads of Microsoft researchers' CHI 2015 papers
Take blood pressure monitoring, for example. In one of the papers being presented at CHI, Microsoft researchers asked 34 people to monitor their blood pressure five times a day, in order to better understand whether it would be useful to have continuous blood pressure monitoring in a gadget such as Microsoft Band.
They found that while people were interested in their blood pressure, they didn’t necessarily understand what the data meant.
Dan Morris, a senior researcher who worked on the project, said some participants didn’t trust the readings if results went against their own perceptions of their health, or were really different from a previous reading.
Participants also were likely to draw conclusions about their blood pressures that weren’t necessarily accurate. For example, a person who got a high blood pressure reading might blame it on a recent salty meal, Morris said. While high salt intake in general could have an effect on blood pressure, it’s unlikely that one meal will cause an immediate spike.
Some participants also reported that just the act of monitoring their blood pressure made them feel stressed and anxious.
The reports’ findings reinforce the idea that even though technologies are evolving to the point that they can give us continuous health data about things like weight and blood pressure, that’s not necessarily the best thing for the user.
“It’s not the Holy Grail in and of itself,” Morris said.
In the field of search, researchers also are working hard to figure out how to understand a person’s intent when they make a search query, and provide the most relevant results.
“Although we have made tremendous strides in improving the quality and capabilities of search over the last decade, we are still just scratching the surface of what we can do,” said Susan Dumais, a distinguished scientist and deputy managing director of the Microsoft Research lab in Redmond.
In her CHI keynote, Dumais will describe how large-scale behavioral data, such as the queries that people give and the results they click on, are transforming the way in which researchers understand search behavior. That’s also allowing them to design and evaluate new search algorithms and interfaces.
Although behavioral data provides a great deal of information about what people are doing, it provides little insight about why they are doing it, or whether they are satisfied. Dumais said that’s why it’s important to use complementary methods such as field observations, laboratory studies and panels, to get a more complete understanding of, and support for, search.
See also: Search, Information Retrieval, and Knowledge Management research
The field of search also must evolve to be more responsive to new technologies, especially mobile devices. For example, a person typing a search query into a computer screen might use different language than someone who is speaking their request on a mobile phone. That’s because spoken queries tend to be longer, and people use more natural language than when they are typing. The search technology needs to understand they still might want the same thing.
At the same time, Dumais said, people have less patience for technology that doesn’t deliver the results they want right way.
“Today, people just expect search to work, and their expectations about what search can do are growing,” she said.
In some cases, making human-computer interaction more personalized means using technology to replicate the kinds of experiences we have when we don’t use any gadgets at all.
That’s part of the logic behind ImmerseBoard. It’s a Microsoft technology that uses huge touch screens called SurfaceHubs combined with Microsoft Kinect cameras to let two people standing hundreds of miles apart look at each other, virtually “shake hands” and even write on the same white board.
See also: Communication and Collaboration research at Microsoft
The interactive, 3D technology replicates the feeling two people have when they are standing next to each other in a conference room chatting about a project. That allows for much more natural interaction than the often-awkward conference call or video chat.
Of course, the technology also adds plenty of bells and whistles that your regular conference room white board can’t.
For example, if two people are in one room, it uses face and clothing recognition to figure out who is at the white board at any given moment, so it can later separate out who wrote what on the board. In addition, it can distinguish between right and left hands, so you can use one as a pencil and the other as an eraser, and it lets you do things like use your finger as a laser pointer from far away.
“We can make meetings more personalized,” said Yinpeng Chen, a senior research software development engineer at Microsoft who is helping to develop the technology.
A framework for automatically generating interactive tutorials: Researchers at Microsoft, Cornell University and the University of Washington are presenting a system that would automatically create step-by-step tutorials, which could be used for things like educational games.
Automatic game progression design through analysis of solution features: Researchers from Microsoft, Cornell and the University of Washington will present a system that automates the progression from one level of a game to another.
Mixed-initiative approaches to global editing in slideware: Researchers at Microsoft, the University of Waterloo and the University of Tokyo collaborated to create a system to improve the visual consistency of slide decks.
Mudslide: A spatially anchored census of student confusion for online lecture videos: Microsoft researchers have created a prototype system that allows students to give online lecturers feedback about what parts of the lecture were least clear to them, linked to particular slides in the lecture.
RIMES: Embedding interactive multimedia exercises in lecture videos: Microsoft researchers have created a way to make online lectures more interactive, allowing students to record audio, video and text that teachers can evaluate.
Accurate, Robust and Flexible Real-Time Hand Tracking: Microsoft researchers have developed a system that can track, in real time, all the sophisticated and nuanced hand motions that people make in their everyday lives.
Measuring crowdsourcing effort with error-time curves: Researchers at Microsoft and Stanford University have created a data-driven model for figuring out the fair price to pay for crowdsourced jobs such as transcription and search.
“For Telling” the present: Using the Delphi Method to understand personal information management practices: Researchers from Microsoft and five universities collaborated to come up with better systems for managing personal information such as documents and e-mails.
Break it down: A comparison of macro- and microtasks: Researchers from Microsoft and Stanford worked together on a paper looking at the benefits of breaking up major jobs, such as a huge transcription project, into small tasks that can be completed by crowdsourcing. Although the “microtask” system generally took longer, the researchers found that they got higher-quality results.
ModelTracker: Redesigning performance analysis tools for machine learning: Microsoft researchers have developed an interactive and efficient way of displaying machine learning performance while simultaneously supporting direct access to data for inspection and debugging.
(s|qu)eries: Visual regular expressions for querying and exploring event sequences: Researchers at Microsoft and Brown University have developed a system for non-programmers to use visual tools to better find and analyze sequences in data sets such as programs or web logs.
FluxPaper: Reinventing paper with dynamic actuation powered by magnetic flux: Researchers at Microsoft and Keio University in Japan have developed paper that contains a very thin magnetic layer, allowing for much more sophisticated interactions than regular paper.
The known stranger: Supporting conversations between strangers with personalized topic suggestions: Researchers at Microsoft, the University of Minnesota and Carnegie Mellon sought to solve an awkward problem: Talking to new people. The study used a wearable device to offer topic suggestions to strangers conversing for the first time.
Sangeet Swara: A community-moderated voice forum in rural India: Researchers at Microsoft and the University of Washington have created a system for curating audio messages for communities in developing regions to share with one another.
Playing the legal card: Using ideation cards to raise data protection issues within the design process: Researchers at Microsoft and University of Nottingham developed a deck of cards to help system designers comply with emerging European Union data protection regulation. The cards take a game-based approach to support more creative compliance.
Exploring time-dependent concerns about pregnancy and childbirth from search logs: Researchers from Microsoft and the University of Waterloo used anonymous search engine logs to create a methodology for better understanding what pregnant women and new moms are looking for at various stages of pregnancy and early parenting.
Health vlogs as social support for chronic illness management: Researchers at Microsoft, University of Washington and Michigan State University analyzed video blogs and user comments from people diagnosed with HIV, diabetes and cancer. They found implications for how vlogs could be enhanced to further support people managing chronic illness.
Gauging receptiveness to social microvolunteering: Researchers at Microsoft, Carnegie Mellon and University of Rochester created Visual Answers, a Facebook application that posts visual questions from people who are blind, to test the concept of asking friends to do small volunteer tasks.
Understanding data videos: Looking at narrative visualization through the cinematography lens: Researchers at Microsoft, University of Manitoba and University Toulouse analyzed data videos to better understand how to create a tool that would let more people do their own data storytelling.
Modeling ideology and predicting policy change with social media: Case of same-sex marriage: Researchers at Microsoft and MIT analyzed four years of Twitter posts related to same-sex marriage to build models for using social media to successfully predict the outcome of proposed policy changes.
Designing social and emotional skills training: The challenges and opportunity for technology support: Microsoft researchers worked with Vienna University of Technology to come up with ways that human-computer interaction could help support other tools for developing better social and emotional learning.
Data-in-place: Thinking through the relations between data and community: Researchers at Microsoft, University of Nottingham and Newcastle University will present findings of a yearlong study on data from one street and its community.
The Semantic Paintbrush: Interactive 3D mapping and recognition in large outdoor spaces: Researchers from Microsoft and elsewhere collaborated on a project that can create big, 3D maps of outdoor spaces and when draw on that map with a laser pointer.
Digital collections and digital collecting practices: Researchers at Microsoft and Cardiff University have developed a system for classifying what constitutes a digital collection of data, and present a framework for creating more meaningful and valuable digital collections.
Effect of machine translation in interlingual conversation: Lessons from a formative study: Researchers at Microsoft and University of Maryland evaluated how people use a translator application and found that they naturally adapt their style of speech to accommodate the technology.
Voice or gesture in the operating room: Researchers at Microsoft and several universities found that, when performing cardiothoracic surgery, surgeons can benefit from having both voice and gesture control in the operating room.
Industry is changing, and so must we: In this case study, Microsoft researchers argue that course curriculum and training in human-computer interaction must evolve to reflect the skills required in today’s software industry.