2nd Quarter 2019, Volume 33
Message from the Director
This has been an active Spring at VACCINE, with team members engaging in researching, solving, and delivering new solutions to our partner community. One area that has grown in prominence has been work in human-computer collaborative decision-making. Areas of active projects include explainable AI and explainable machine learning, human-guided interactive machine learning, human-computer teamed decision making, and trustable information (e.g., reliability, misinformation, deep fakes). I am encouraged to see many agencies, sponsors, and researchers realizing that automated deep neural networks have limits and the human-in-the-loop is necessary to create understandable, trustable, and actionable information for most real-world problems with non-repeatable patterns. We have been actively engaging in discussion at the policy level and societal impact level in additional to the traditional technology research, development, and deployment level. The interdisciplinarity of the needed solutions and teams we are forming is exciting and we expect great progress over the next year. If these topics interest you, please let us know. 

Below, we highlight some of the activities of team members. Please feel to contact us and them for more information.

Summer is always an active period for our Center with an influx of summer students working on interesting problems. We hope you have a great summer and if your travels bring you to West Lafayette, we are happy to show you some of the exciting work that will be occurring. 

SMART Showcase at PVAMU Research Week Symposium

VACCINE’s partner researchers at Prairie View A&M University, led by Dr. Louis Ngamassi, presented two posters at PVAMU’s inaugural Research Week Symposium, including one featuring the Social Media Analytics and Reporting Toolkit (SMART) and one featuring Dr. Ngamassi’s work on the use of mobile apps for disaster communication in underserved communities. The Research Week Symposium is an opportunity for PVAMU faculty and students to showcase their projects. 
Dr. Ngamassi presenting SMART
Graduate research assistant Tracy Jones with one of the Research Week posters
GeoTxt and GeoAnnotator Publications
VACCINE post-doctoral researcher Dr. Morteza Karimzadeh co-authored the papers “GeoTxt: A scalable geoparsing system for unstructured text geolocation,” which appeared in Transactions in GIS’s February 2019 issue, and “GeoAnnotator: A Collaborative Semi-Automatic Platform for Geo-Annotation”, which appeared in the ISPRS International Journal of Geo-Information’s March 2019 special issue on human-centered visual analytics and visuospatial display design. GeoTxt and GeoAnnotator were created and authored in collaboration with VACCINE partner Dr. Alan MacEachren from Pennsylvania State University.

GeoTxt Abstract—In this article we present GeoTxt, a scalable geoparsing system for the recognition and geolocation of place names in unstructured text. GeoTxt offers six named entity recognition (NER) algorithms for place name recognition, and utilizes an enterprise search engine for the indexing, ranking, and retrieval of toponyms, enabling scalable geoparsing for streaming text. GeoTxt offers a flexible application programming interface (API), allowing for customized attribute and/or spatial ranking of retrieved toponyms. We evaluate the system on a corpus of manually geo‐annotated tweets. First, we benchmark the performance of the six NERs that GeoTxt provides access to. Second, we assess GeoTxt toponym resolution accuracy incrementally, demonstrating improvements in toponym resolution achieved (or not achieved) by adding specific heuristics and disambiguation methods. Compared to using the GeoNames web service, GeoTxt's toponym resolution demonstrates a 20% accuracy gain. Our results show that places mentioned in the same tweet do not tend to be geographically proximate.
VACCINE at the American Association of Geographers’ Annual Meeting

Dr. Morteza Karimzadeh, VACCINE post-doctoral researcher, and Gourav Jhanwar, one of VACCINE’s graduate student researchers, presented two papers at the American Association of Geographers’ annual meeting in Washington, DC on April 3-7: “Interactive Deep Learning for Identifying Relevant Social Media Posts in Crisis Monitoring” and “Irrigation Management Using Deep Learning of Soil Moisture”.

“Interactive Deep Learning for Identifying Relevant Social Media Posts in Crisis Monitoring” was co-authored by Dr. Morteza Karimzadeh and Dr. David Ebert as well as VACCINE graduate researchers Yi-Shan Lin and Luke Snyder, and presented by Dr. Karimzadeh.  
Abstract—Many first response groups or emergency operation centers are tasked with real-time monitoring of events or day-to-day activities to identify potential disruptions. In recent years, such organizations have been increasingly using public social media posts for situational awareness, essentially leveraging social media users as human sensors on the ground. Systems built for this purpose rely heavily on filtering and classification of social media posts to eliminate large amounts of noise (e.g. irrelevant posts, robot posts, or job advertisement) to provide relevant information to the analysts for the type of event or disaster under study. Linguistic (semantic) ambiguities, however, make this filtering and classification a challenging task. For instance, “My team is on fire tonight” may trigger a false positive alarm for “fire event” classifiers. In this paper, we present a novel interactive deep learning-based classification system for identifying social media posts (tweets) that are relevant to the analysts’ tasks in situational awareness. We evaluate our approach on a test dataset. Our results show significant improvement over baseline filtering approaches. Further, we present the integration of our deep classifiers within SMART (“Social Media Analysis and Reporting Tool”), our geovisual analytics platform that has been in use by more than 20 organizations across the U.S. for daily use or event monitoring. This integration allows analysts to train the system’s classifiers in real-time for any event of interest.

“Irrigation Management Using Deep Learning of Soil Moisture” was co-authored by VACCINE graduate researchers Gourav Jhanwar, Calvin Yau, and Kirubel Tadesse along with Dr. Morteza Karimzadeh and Dr. David Ebert, and presented by Gourav Jhanwar. 
Abstract—There are many critical decisions that can affect the environmental and financial outcome of farmers’ business. Among such decisions, one that stands out is the decision of knowing the irrigation requirement of crop and return on investment in irrigation systems. Irrigation can play a huge role in determining how successful a crop is; however, irrigation carries its own costs, and at the scales many farmers work at, using simplistic, inefficient methods of determining when to irrigate can cause them to waste quite a lot of money and water. To properly estimate the cost, the benefit, and the environmental impact we must first have a clear and comprehensive estimation of the water usage that considers soil properties. Traditional models rely on the FAO-56 method for estimating evapotranspiration, which has common issues of the presence of confounding factors and geographic variability. Hence, in this paper, we present a novel approach which will integrate deep learning for predicting evapotranspiration, use remote sense along with weather station data and soil properties in an integrated system to enable growers to make accurate irrigation decisions. We evaluate our model using test data showing better RMSE value than traditional baselines. We later present a visual analytics platform which uses this approach and deploys it to growers, which will help them to make better irrigation decisions.
Oxford-Purdue Collaboration: Systematic Approaches to Designing and Evaluating Visual Analytics Workflows

Since 2014, VACCINE has collaborated with the UK Visual Analytics Centre (UKVAC), analyzing the challenges in evaluating many visual analytics applications and searching for a systematic approach to designing and evaluating visual analytics workflows. Dr. Min Chen (a member of UKVAC) and VACCINE director Dr. David Ebert have recently completed a joint paper, “"An Ontological Framework for Supporting the Design and Evaluation of Visual Analytics Systems", outlining a potential solution to this systematic approach. In this paper, for the first time, they relate the problem-solution space in visual analytics workflows to symptoms, causes, remedies, and side-effects in medicine, and describe an abstract reasoning methodology for analyzing the causal relationships among symptoms, causes, remedies, and side-effects in terms of the four primary components (statistics, algorithms, visualization, and interaction) in visual analytics workflows. This work represents a transformation from practice to theory, which will hopefully benefit many applications of visual analytics in the coming years.

This paper will be presented at EuroVis 2019, and will be published in Computer Graphics Forum, volume 38, issue 3.  It is also supported by a web site: IVAS (Improving Visual Analytics Systems).
CERIAS Annual Symposium and Seminar Series

On April 10th, VACCINE Director Dr. David S. Ebert participated in the Center for Education and Research in Information Assurance and Security (CERIAS)’s Annual Security Symposium. He served as the moderator for the panel “Health & Longevity and Security”, with participants George Bailey (Senior Advisor, Purdue Health IT Security), Vijay Raghunathan (Professor of Electrical and Computer Engineering), Chris Reed (Eli Lilly Director of Product Cybersecurity), and Jim Routh (CVS Health Chief Security Officer). 

Dr. Ebert also gave a presentation as part of CERIAS’s Information Security and Cyber Crime Seminar Series on February 6th, on the topic “Trustable Information for Security Applications: Visual Analytics for Reliable, Effective Decision Making”.  
The Future Impact of Technology and Automation on Perennial Crop Decision Making and Sustainability
VACCINE Director Dr. David S. Ebert attended the Unified Wine & Grape Symposium, held in Sacramento, CA from January 28-31, and participated in their panel discussion on technology, big data, and human interaction with technology in the wine industry. The discussion focused on the role of AI and digitization in winemaking, and the expanding possibilities for growers in light of new technological developments.
powered by emma
Subscribe to our email list.