Announcements

Scholars Data Visualization Challenge Results

March 1, 2017

Visualizing Scholars@Duke, the data visualization challenge open to the Duke community, concluded with six engaging visualizations capturing the richness and dynamism of Duke research. The datasets from Scholars@Duke encompassed five years of publications from researchers across Duke.

The data visualization challenge was part of the Duke Research Computing Symposium held on January 19, 2017 at the Technology Engagement Center (TEC), Duke's newest high-tech creative hub on campus. Each team creatively approached the challenge in ways that, as you can see below, captured Duke research very distinctively. Congratulations to all contestants who participated in the Visualizing Scholars@Duke challenge! Check out the posters below - PDFs are stored in the DukeSpace repository.

Scholars is a Duke institutional resource for data about faculty and researchers and it's used across Duke in school and department websites, Tableau reports and dashboards, and Duke's global map. These visualizations demonstrate the potential for representing the richness and variety of Duke faculty and research information, and could be displayed in interactive sites to capture recent updates in research.

First place: Mapping the geographic spread of collaborations across Duke University
Jeff MacInnes, Duke Institute for Brain Sciences, Duke University

This poster looks at how physical proximity of researchers affects collaborations by combining publications with space data from Duke’s Plant Accounting. Faculty were located by their department and co-authorships highlight collaborations across building and across departments.  

In addition to being visually appealing and easy to grasp, this visualization cleverly incorporates another data set to show cross-department collaborations and the interdisciplinary nature of roughly 48% of Duke publication collaborations within the five years of the publications data. An interactive version of this visualization lets you browse individual department collaborations.

Second place: A Glimpse Into Duke Research: Where Are You Now?
Avery Yuan (1), Liqi Feng (2), Xiaodi Qin (1), Yi Zhao (1); 1: Department of Biostatistics and Bioinformatics, Duke University 2: Duke Clinical Research Institute, Duke University

This team-based poster detect patterns in Duke's research networks by looking at the network from different perspectives--by topic, by department, and by collaboration frequency. Each team member contributed a different analysis, giving this poster a much broader, multi-faceted view of the data.

One of these visualizations, the pie chart, shows that 26% of researchers are not represented, which raises more questions: in what areas are collaborations not happening and why?  Which data sources might capture more collaborations?  Does this data unveil true opportunities for collaborations or would it be better to understand the culture of those disciplines? This visualization called out some of the inherent biases or gaps with only using publications data, and the need to address those. For example, perhaps data about Artistic Works, contributor roles on grants, or teaching roles would better reflect collaborations. 

Third place: Analyzing Scholars@Duke Data via Tensor Factorization
Changwei Hu, Department of Electrical and Computer Engineering Duke University

This work ingeniously infers research topics from publications and thus uncovers communities of Duke faculty who are working on similar research topics. The algorithm uses the frequency of a terms in the abstract and then distills the terms that were commonly used together into larger research areas as shown on the poster. Attaching authors, venues, and keywords to those inferred topics creates a process for aggregating keywords, which can’t be reliably done without hierarchical controlled vocabularies. 

For example, storing synonyms and related terms could create the foundation for a personalized recommendation tool that could suggest potential researchers, research topics, or journals of interest to an individual. Or search results could be expanded to look for these broader topics.

Honorable mentions

Towards an Intellectual Atlas of Scholars@Duke
James Moody (1,2), Peter Mucha (1,3,4); 1: Department of Sociology, Duke University, 2: Duke Network Analysis Center, Duke University, 3: Department of Mathematics, University of North Carolina, 4: DNAC Lab, Duke University

As a competition judge, Professor Moody was exempt from competing but contributed an incredible poster with multiple visualizations none the less.

This analysis uncovers a research landscape where collaborations are depicted as a topography. Using abstracts to find levels of similarity in papers, the shape of the landscape was calculated to place similar papers near one another. 

In one visualization, darker areas (hills) represent clusters of papers of similar topic, ridges represent papers that link topics together, valleys represent gaps in knowledge, and islands represent widely different topics disconnected from the mainland. “Islands” show isolated areas of collaboration.  Each point represents an author and the color of the point varies from blue to red presenting the publication volume. This poster does a great job of representing complex algorithms with a simple visual metaphor. An interactive version enables users to hover over certain areas to find more about researchers.

IvoryTowers
David P. Bradway, Department of Biomedical Engineering, Duke University

This analysis looked at the Erdős number or the "collaborative distance" within a community, similar to the six degrees of Kevin Bacon phenomenon. David Bradway revealed that the longest “chains” at Duke to connect people was the distance of 14 people through publication co-authorships.  He found a chain of three frequent co-authors that occurred frequently in the Duke publication chains.

Bradway also developed an interactive version that functions like a prototype for identifying friends of friends. In the tool, the user enters the name of a Duke researcher and it returns the co-authors that haven’t published with the researcher but have published with the researchers’ co-authors. David jokingly referred to it as "who is publishing without you," but this idea could be re-purposed as a way to recommend new colleagues who may share interests.

Shift of Interests Among Duke Scholars
Tai-Lin Wu, Economics and Computation, Duke University

This fascinating analysis focused on publication titles in Wu’s own department, Computer Science, creating word clouds for two time periods to capture a shift in research trends.  From 2011 - 2016 he saw a move away from words like "network, privacy, and DNA" and the inclusion of words like "parallel, data, and molecular."

This engaging visualization demonstrated the advantages of straightforward approaches to mapping research interests. Scholars data can tell a lot of stories about research, including simple depictions like a word cloud. Sometimes complexity can obscure a message that can be told effectively through a simpler medium.

Summary:

We were thrilled to meet all of the competitors and take a deeper dive into their work. All of these great ideas could easily provide a springboard to thematic visualizations, interactive sites, or even new functionality in Scholars@Duke. The Scholars team is seeking others at Duke who are interested in using Scholars data to help support and represent the missions and activities of their units. To contact us, email scholars@duke.edu.

 

Did you know that Scholars@Duke has a CV?

January 4, 2017

Scholars can display your publications, grants, courses and other profile information in a Word document formatted as a CV. The Word document can be easily edited to add, move, or format information as you'd like. The Scholars CV is useful for faculty who maintain their publications in Elements and other profession activities in Scholars@Duke.

Note: More data fields to support a complete CV are planned for February 2017.

To create a CV:

  1. In Profile Manager, click the Generate CV button. This button saves a Word file called “cv.docx.zip” in your Downloads folder.
  2. Open the file in Word to edit and save your CV with a different filename.   

 

To make your CV available publically, you can store your CV in your Duke Box location and create a webpage link on your Scholars page. For more about how to use Duke Box, see box.duke.edu.

Scholars Report Simplifies Annual Reporting

December 5, 2016

The Scholars@Duke report makes it easy to copy information from your Scholars profile into a CV, an email, or any document used for annual reviews or reporting. The Scholars report can display publications, grants, and other activities produced this year or during any date range.

To create a Scholars report, take these steps:

  1. Search for the person's name in Scholars@Duke and on the People tab of the Scholars search results page, click the individual's name.
  2. On the profile, go to the editing interface by clicking "Manage This Profile" in the upper right corner. Login with your NetID and password.
  3. In Profile Manager, click the "Generate Report" button on the right. The report window appears.
  4. At the top of the report window, you can select or remove the sections you don't want to display by de-selecting each check box. For example, to create a current publications list, you can deselect all sections except "Publications."
  5. To display only the output produced this year, add a start date and an end date, for example “2016-01-10 to 2016-12-31” to display the information produced during the 2016 calendar year. Note: Outputs and activities that are entered without dates always appear, regardless of the
  6. To change the format of your publication citations, select your preferred style, such as MLA, from the drop-down at the top of the report. Your publication citations will be displayed in your preferred style.

 

The Scholars report is intended for copying and pasting information into other systems rather than for printing. To print the report or save it as a PDF, paste the report into a Word document.

  • To copy the report, click the "Select all" button and select "Copy" from your browser’s Edit menu, or press the Ctrl + C buttons together. Then paste the text into a Word document, an email, or other application window with the Ctrl + V buttons.

If you need to add information to your Scholars report that is hidden from your public profile, follow these steps:

  1. Unhide the section or items that you want to include in your Scholars report.
  2. Generate the report, and copy the text to another document.
  3. Re-hide the information that you don’t want to display on your Scholars profile.

For more help, contact your Power User or create a Scholars Help ticket by emailing scholars@duke.edu.