Irene Ros

Designing and building software, tools and experiences on the web that aim to improve people's lives.

The last 8–9 years of my career have been focused on Data Visualization, which has given me plenty of time to develop a philosophy or two about my approach to this field. I say two, but I really mean about half a dozen of semi-distraught career-crisis generating moments (that lasted weeks or months) questioning what it is I want out of this field, whether what I am doing is the “right thing” and whether I should be doing something else.

Sure, it can messy. But this Open Web workflow will test your assumptions and give you better results.

US Federal Legislation is a common subject of discussion and advocacy on the web.The contents of bills present a significant challenge to both experts and average citizens due to their length and complex legal language. To make bills more accessible to the general public, we present Many Bills: a web-based visualization prototype that reveals the underlying semantics of a bill. We classify the sections of a bill into topics and visualize them using different colors. Further, using information retrieval techniques, we locate sections that don't seem to fit with the overall topic of the bill. To highlight outliers in our 'misfit mode', we visualize them in red, which builds a contrast against the remaining gray sections. Both topic and misfit visualizations provide an overview and detail view of bills, enabling users to read individual sections of a bill and compare topic patterns across multiple bills. We obtained initial user feedback and continue collecting label corrections from users through the interface.

Reading congressional legislation, also known as bills, is often tedious because bills tend to be long and written in complex language. In IBM Many Bills, an interactive web-based visualization of legislation, users of different backgrounds can browse bills and quickly explore parts that are of interest to them. One task users have is to be able to locate sections that don't seem to fit with the overall topic of the bill. In this paper, we present novel techniques to determine which sections within a bill are likely to be outliers by employing approaches from information retrieval. The most promising techniques first detect the most topically relevant parts of a bill by ranking its sections, followed by a comparison between these topically relevant parts and the remaining sections in the bill. To compare sections we use various dissimilarity metrics based on Kullback-Leibler Divergence. The results indicate that these techniques are more successful than a classification based approach. Finally, we analyze how the dissimilarity metrics succeed in discriminating between sections that are strong outliers versus those that are 'milder' outliers.

US Federal legislation is a hot topic for discussion and advocacy on the web. Yet legislative bills present a significant challenge for both experts and average citizens to navigate and understand. To explore solutions to this problem, we have created DocBlocks: a prototype visualization and website that enables users to explore the content of congressional bills and communicate their findings to others. Our technique enables us to take any document from a categorized corpus, classify its sections, and visualize its topic structure. With the launch of this service, we hope to provide a valuable tool for open governance and learn from our users at this critical intersection of visualization, advocacy, social software, and civil society.

Safety has been a monumentally important issue in the medical field. Preventable medical errors in hospitals are estimated to cause the death of over 98,000 patients a year. Typical errors include miscommunication between medical professionals, incorrect drug administration, miscalculated drug doses, and many other minor yet life threatening mistakes. One of the reasons that such mistakes occur is that medical processes are complex and have many stages that require the collaboration and coordination of several professionals and departments. This kind of complexity often leaves processes insufficiently defined, so participants are unsure how or what needs to be done in unusual situations or make mistaken assumptions about the behavior of other participants.

The overall goal of the project was to enhance the safety and efficiency of complex medical processes by applying new methods developed in software engineering. These techniques support formalizing the process definitions and using verification techniques to check them for possible errors.

My thesis work in the Laboratory for Advanced Software Engineering Research (LASER) concentrated around the use of Little-JIL, an agent coordination language, to continue the modeling and analysis of a real-world medical process: the Adult Outpatient Chemotherapy Process that is being performed at the Baystate Hospital's D'Amour Cancer Center. I concentrated my efforts around defining the process itself, applicable medical terminology, participating agents, the resources that are required, artifacts (such as medical charts) that are created and used, the non-normative behaviors that must be accommodated and the safety properties that must all be maintained.

In this thesis I describe the process and the methodology I used to elicit it, along with findings indicating that defining and evaluating the process helps in identifying weaknesses in it, thus leading to an improved medical process and greater patient safety.

In this paper we describe a cross-platform, multimedia courseware presentation system developed as a natural evolution of our existing technologies that deliver course content to on-campus and distance learning students. Following more than a decade of developing and deploying web-based education (WBE) systems to support on-campus and distance education students, a number of factors – the need to support a variety of platforms, the increasing availability of high-bandwidth network access, new pedagogies for teaching and learning, etc. – led us to consider a completely new approach to content capture, authoring and delivery. In designing these new delivery and capture systems, we were committed to maintaining our tradition of creating systems that place few demands on instructors and their choice of teaching styles and pedagogy, that provide mechanisms for supporting cooperative and inquiry-based learning [Edelson99], and that recognize the increasing role that content capture and delivery plays in teaching and learning for on- campus as well as distance education. Our goal is to create an open-source, platform- independent and effective content delivery system that retains the functionality from earlier technologies that had proved valuable (though extensive assessment and evaluation) to students and instructors in a variety of teaching and learning scenarios. We also want to capture content in classrooms that are moving from traditional lecture formats to active and collaborative learning where a variety of technologies and pedagogical approaches are employed. As a result, we created a pure Java content delivery system that uses a “plug-in” architecture to facilitate incorporation of tools for annotation, notation, collaboration and navigation; and a capture system that acquires “significant” events, in the form of video keyframes, from unconstrained computer-based presentations, video streams of instructors, students and whiteboard use, and other sources.

A key CASA innovation, Distributed Adaptive Sensing (DCAS), relies on the networking of adaptive sensors to achieve observations not possible with traditional sensors. Innovative wireless networks, in particular, are needed to support off-the-grid radars such as those being developed in Puerto Rico as part of the CASA IP5 test-bed.

In fall 2005, CASA faculty and students explored recent research in wireless networking through a novel, hands-on undergraduate course. Prof. James Kurose and CASA Research Scientist Mike Zink led a team of 9 undergraduate and graduate students and researchers to develop, deploy and measure novel outdoor networks based on the 802.11 standard.

The course, UMASS Computer Science 496A, was archived in a novel multimedia format called jMANIC that was developed by CASA student Byron Wallace in collaboration with the UMASS RIPPLES group as part of the CASA Education and Outreach thrust. jMANIC provides a browser-based viewer of synchronized video, audio and graphics with search and logging capabilities.