We are developing a Visual Analytics Digital Library (VADL) to support instructors building new visual analytics courses and students and others who simply seek to learn more about the area. The VADL includes articles, course notes, lecture slides, video lectures, homework exercises and other useful items. By accumulating all these materials at one site, we provide a convenient and easy-to-access collection.
]]>We are developing a Visual Analytics Digital Library (VADL) to support instructors building new visual analytics courses and students and others who simply seek to learn more about the area. The VADL includes articles, course notes, lecture slides, video lectures, homework exercises and other useful items. By accumulating all these materials at one site, we provide a convenient and easy-to-access collection.
Together with researchers at the other RVACs and at NVAC, we developed a taxonomy that serves as the organizational framework for the materials. Generalized search capabilities are provided, and viewers also can search by material type such as PowerPoint slides or documents. Our goal is to provide a flexible set of exploration possibilities to help people with differing objectives to use the VADL.
The VADL home page, shown in the figure, includes a conspicuous SHARE link. We hope that you will contribute. Our goal is to make this site the place to go for accessing visual analytics focused documents and materials. This project is led by Jim Foley.
]]>
We are developing computational techniques for recognizing story plots connecting (apparently isolated) events. The goal is to recognize story plots early enough to make useful predictions about future events. Work is currently done on the VAST dataset synthesized by PNNL, which contains news stories about numerous events, including some 120 events pertaining to illegal or unethical activities in a hypothetical community.
Parallel Sets is a powerful technique for the interactive visual analysis of categorical data. In addition to visualizing the existing data, parallel sets allow the user to create new data dimensions by combining existing ones, thus tailoring the data to the analysis process and the user’s model. We are currently adding new capabilities, such as including the display of textual dimensions, and also building a Bayesian model of the data directly from the visualization.
]]>
The above video analysis produces rich and complex patterns over time. We are developing a variety of analysis techniques to attack this ever-changing stream of topics, allowing the user to quickly get an overview of what is being discussed and then home in on the most relevant stories associated with topics of interest. The top figure shows clusters of topics over a period of time, in this case one month. For this time period, the pixel grid for each topic looks like a monthly calendar with importance of the topic for a given day given by saturation of the color.
]]>
The integration of intelligent image content analysis and interactive visualization has produced a complete visual analytics tool. Image content analysis is automatically applied to a set of images with unknown content. Concepts within the images can be found from a pre-trained set or added by user selection on a training set of images. Images are then clustered according to their properties and according to the concepts they contain so that like images will be close to one another.
]]>
We are developing a system that will help analysts browse through large collections of intelligence reports. The system begins by parsing plain text reports and constructing XML nodes with fields for the important report items such as people mentioned, dates, places, and activities. From there, we construct a graph visualization of the reports with a node for each report and for all of the entities in a report (see figure). Specific colors represent the different types of entities.
]]>
A multimodal data analysis is applied to concurrent visual signals, auditory signals, and, when available, closed caption text. The analysis is general and unstructured; it can be applied, for example, to broadcast video in any language. We have applied the analysis to automatically identify and segment news broadcasts. However, the methods can be applied to identify and segment other broadcast types as well. The methods break the news broadcast stream into separate stories and also determine key frames for each story. When closed captions are available, each story can be labeled with its topic. The story segmentation is robust and has been applied to broadcasts in both Japanese and English. For the latter we have assembled a very large and rich collection over multiple months.
]]>
Real world knowledge gathering and investigative tasks are very complex because the problem-solving context is constantly evolving, and the data may be incomplete, unreliable and/or conflicting. We are developing a mixed-initiative reasoning agent that will assist homeland security analysts to choose from and reason about enormous databases of text, imagery, video and webcast. This agent will leverage an AI blackboard system and resource-bounded control mechanisms to support hypothesis tracking and validation in a highly uncertain environment. Interactive visualizations are being used to enable analysts to gather and sift large amounts of evidence and to collaborate with and, where necessary, to control the agent.
]]>
We have developed visual analysis methods to investigate hundreds of thousands or more bank transactions over extended periods of time. Transactions are clustered and reclustered according to their similarities and then explored for patterns of keywords, distribution within and between clusters, amount of activity or amount of money transferred, and other factors. Multiple interlinked views are maintained so that users can rapidly navigate from overviews to particular views of keyword correlations, transactional activity, or even the details of specific transactions. These methods are being applied to investigations of wireless transfer transactions for possible money laundering. The process here is similar in several ways to intelligence analysis. The tools are general and can be applied to other types of transactional analysis over time.
]]>
