Improving Software Development through Human-Centered Approaches

Tuesday, September 30, 2014

Abstract:
My Natural Programming Project is working on making programming languages and environments easier to learn, more effective, and less error prone. We are taking a human-centered approach, by first studying how people perform their tasks, and then designing languages and environments that take into account people's natural tendencies. We are designing new programming languages for people who are not professional programmers (sometimes called "end-user programmers") based on how people think about expressing algorithms and tasks. For example, the InterState system uses a visual notation that combines states and constraints to make web behaviors easier to express. We also are working on improving programming environments and libraries for professional programmers. For example, programmers often need to backtrack out of changes while exploring how to achieve their goals, which is poorly supported by today's IDEs, so we developed a selective undo mechanism that makes them twice as effective. We studied the usability of APIs, such as the Java SDK and the SAP eSOA APIs, and discovered some common patterns that make programmers up to 10 times slower in finding and using the appropriate methods, so we have developed new tools to compensate. This talk will provide an overview of our studies and resulting designs and tools, which benefit from applying both Software Engineering and Human-Computer Interaction approaches.

Bio:
Brad Myers is the principal investigator for the Natural Programming Project and the Pebbles Handheld Computer Project. He has been a consultant on user interface design and implementation to over 75 companies, and regularly teaches courses on user interface design and software. He is IEEE Fellow, ACM Fellow, a member of the CHI Academy and winner of three Most Influential Paper Awards. Myers' research interests focus on user interface development systems, user interfaces, handheld computers, programming environments, programming language design, programming by example, visual programming, interaction techniques, and window management.

The Expanding Influence of Computer Science across Higher Education

Tuesday, October 28, 2014

Abstract:
Computer Science as a discipline has matured and expanded. Many computer scientists are leading major higher education institutions and computer science courses are more popular than ever among non-majors. In this talk, I will present the tremendous pressure higher education institutions are facing regarding delivery, resources, etc. and a pivotal role that computer science and technology could play in addressing these issues.

Bio:
Satish Tripathi was appointed the 15th president of the University at Buffalo in 2011. He served as UB’s provost from 2004-2011, was dean of the Bourns College of Engineering at the University of California-Riverside from 1997-2004. Previously, he spent 19 years as professor of computer science at the University of Maryland, including seven years as department chair. Fellow of the IEEE and AAAS, Tripathi also he holds an honorary doctorate from the Indian Institute of Information Technology, Allahabad. A member of the Mid-American Conference Council of Presidents Executive Committee and the boards of the Council for Higher Education Accreditation and the Digital Preservation Network, he was appointed by Governor Cuomo as co-chair of the Regional Economic Development Council for Western New York.

Mining Large Graphs

Tuesday, November 4, 2014

Abstract:
Given a large graph, like who-calls-whom, or who-likes-whom, what behavior is normal and what should be surprising, possibly due to fraudulent activity? How do graphs evolve over time? How does influence/news/viruses propagate, over time? We focus on three topics: (a) anomaly detection in large static graphs (b) patterns and anomalies in large time-evolving graphs and (c) cascades and immunization.

For the first, we present a list of static and temporal laws, including advances patterns like 'eigenspokes'; we show how to use them to spot suspicious activities, in on-line buyer-and-seller settings, in FaceBook, in twitter-like networks. For the second, we show how to handle time evolving graphs as tensors, how to handle large tensors in map-reduce environments, as well as some discoveries such settings.

For the third, we show that for virus propagation, a single number is enough to characterize the connectivity of graph, and thus we show how to do efficient immunization for almost any type of virus (SIS - no immunity; SIR - lifetime immunity; etc)

We conclude with some open research questions for graph mining.

Bio:
Christos Faloutsos received the Presidential Young Investigator Award by the National Science Foundation in 1989, the Research Contributions Award in ICDM in 2006, the SIGKDD Innovations Award in 2010, and two Test of Time awards. He is an ACM Fellow, and has served as a member of the executive committee of SIGKDD. He holds eight patents and he has given over 35 tutorials and over 15 invited distinguished lectures. His research interests include data mining for graphs and streams, fractals, database performance, and indexing for multimedia and bio-informatics data.

Learning from Demonstration Technology: A Tale of Two Applications

Tuesday, November 18, 2014

Abstract:
Learning from demonstration technology has seen increased focus in recent years as a means to endow computers with capabilities that might otherwise be difficult or time-consuming for a user to program. This talk describes two efforts that employ learning from demonstration technology to quite distinct ends. The first is to provide a capability that supports users with no programming experience in the creation of procedures that automate repetitive or time-consuming tasks. This capability has been operationally deployed within a collaborative planning environment that is used widely by the U.S. Army. The second is to support automated performance evaluation of students as they seek to acquire complex procedural skills through training in virtual environments. In this second case, instructional content developers employ learning from demonstration technology to create solution models for training exercises. An automated assessment capability employs soft graph matching to align a trace of a students response to an exercise with the solution models for that exercise, providing a flexible basis for evaluating student performance. In contrast to intelligent tutoring systems that force students to follow a pre-specified solution trajectory, our approach enables meaningful feedback in domains where solutions can have significant variability.

Bio:
Karen Myers is a Principal Scientist within the Artificial Intelligence Center at SRI International, where she leads a team focused on developing intelligent systems that facilitate man-machine collaboration. Myers has led the development of several AI technologies that have been successfully transitioned into operational use in areas that span collaborative systems, task management, and learning from demonstration. Her research interests include autonomy, multi-agent systems, automated planning, personalization, and mixed-initiative problem solving.