The American Geophysical Union, in July 2017, opened up its Fall Meeting database through their OpenAPI Challenge, asking the community to develop apps to facilitate

"serendipitous discovery of relevant research, discovery of new collaboration opportunities, and identification of emerging areas of science".

As enthusiastic AGU members we were delighted at the chance to write software built on the Fall Meetings data. Our app submission, AGU Explorer, is intended as a tool to help scientists explore collaboration networks, learn about recent research, and discover opportunities for interdisciplinary work. We are thrilled that AGU Explorer won first prize; and we hope it helps promote discovery and collaboration.

The landing page of AGU Explorer answers one of the questions that has always intrigued us as we have wandered through the hallways of Moscone Center, hearing snatches of conversations in at least a dozen languages: where in the world do AGU authors come from? The landing page answers that question. But, it does more than that - it also shows trends, for example, how the number of AGU contibutors from a given country has increased or decreased over the years. It can also uncover outreach and collaboration opportunities - for example, countries with few AGU participants.

The Authors page allows the user to interactivley explore the network of abstracts and co-authors for any given AGU Meetings author. The initial display shows the co-author graph: the selected author, their AGU abstracts, and their co-authors on these abstracts. Clicking on any author in the graph will add their network to the existing graph. Additionally, the co-co-authors button allows the user to expand the network to include all abstracts by the co-authors as well as the co-authors of these co-authors.

Abstracts are color-coded by the AGU Program they are part of; for instance, green for Biogeosciences and sky blue for Atmospheric Sciences. Thus, a quick glance at an author's network reveals the author's disciplinary interests and connections. We hope that AGU Explorer, by enabling the exploration and visualization of authorship connections, will enable authors to discover collaboration opportunities, especially interdisciplinary opportunities, within the AGU community.

The AGU Fall Meeting is the meeting place for the AGU Community. Yet, navigating the meeting can be overwhelming for high school and undergraduate students, as well as first time attendees.

The Programs page provides a visual guide to help navigate AGU meetings - past and present. Meeting data is organized in a set of hierarchical pie charts, going from Meetings at the top level, to Programs, to Sessions, and to abstracts (or events) at the leaf level. Data for each level is progressively displayed as the user navigates the hierarchy, allowing the user to navigate the 80,000+ abstracts and events without getting overwhelmed.

Behind the scenes we're using a Neo4j graph database, which is populated using AGU's OpenAPI interface to their Fall Meetings database. Neo4j's advantage is that as a native graph database it handles the network of authors and abstracts in a way which makes queries compact and efficient. The visualizations for the Authors and Programs pages are done using D3, specifically using a force-directed graph to lay out the author network. The interactive map on the landing page uses Leaflet. AGU Explorer would not have been possible without these open source tools.