Geodata Analytics-as-a-Service
By Ben Conklin; Barry Tilton; Kevin Hyers; and Cordula A. Robinson, Ph.D.
This article was originally published in USGIF’s State & Future of GEOINT Report 2017. Download the full report here.
Anyone with access to data wants to exploit and understand it. They want to draw conclusions and make decisions, predict outcomes, and identify patterns and trends based upon data. As more location-based data is collected, tools that make it easy to analyze geodata need to be provided to the entire analytic community. Currently, geospatial analysis tools are only provided to GIS users in the form of desktop software. To unlock the full potential of geospatial data, the GEOINT Community needs to provide Geodata Analytics-as-a-Service (GaaS). GaaS offers a web-based platform where any analyst can have access to powerful geospatial analytic tools to support their analytic workflow.
The application of geodata analytics allows analysts to gain a better understanding of their environment, access timely answers to their questions, and apply operational knowledge to the information. Providing GaaS will give analysts better access to tools and information while simultaneously offloading simple GEOINT analytic problems from the analyst’s desk, offering them more time to explore complex problems and define new methodologies.
GaaS is much more complicated than providing data as a service. Geodata services have well-defined and broadly adopted standards. Geodata analytics does not have the same maturity of open standards. Additionally, analysis is fundamentally a loosely defined discipline with many potential pitfalls if exposed in the wrong way to inexperienced users. GaaS needs to provide a unique set of capabilities across a continuum of problems, from the simple to the complex.
From Simple Visualization to Solving the Impossible Puzzle: The Analytic Continuum
In thinking about analysis-as-a-service, it is useful to consider analytic methods from their simplest to most complex forms. Each level of complexity can have a defined solution and an appropriate audience for use. The simplest of analytics can be leveraged by the widest variety of users, while a small group of highly trained individuals can properly use sophisticated tools. These lines aren’t hard and fast, but many types of analytic services can be categorized using this method.
Simple visualization is the first level of analytics, and it is part of the delivery of GaaS. In simple visualization, analytic tools are applied on top of the data, and filter and aggregate information to help a user gain knowledge. In GEOINT, there are three key aspects to the visualization of data: time, space, and relationships. Data services expose these three variables in the data and use tools to query and visualize information based on these factors. Simple visualization adds in situational understanding and helps to make GEOINT data digestible.
The next level of GaaS is self-service analytics. These algorithms are developed using predefined models and applied to the data with user-defined input. The user can select operational parameters, such as location, equipment characteristics, or limiting condition, and execute the tool. The user views the results as a new data set. A typical example of this kind of analytics is visibility analysis, in which a user defines an observer location and other characteristics, then receives information showing the visible and non-visible areas on a map. Traditionally, this type of analysis is built directly into software applications; now GEOINT platforms are allowing this same analysis to be shared as standard web services and accessible to a wide variety of clients.
Expert analytics is needed to solve more complex problems. Predefined algorithms cannot be used to answer every analyst’s question, but users can combine tools in new ways to process data and create robust information products. This type of workflow requires a trained analyst who can apply analytic rigor to problems and can also select the right tool for the job. In this case, the analytic service needs to be richer. Rather than only exposing single interface tools, analysts need access to an analytic workbench from which they can combine analytic tools and algorithms in new and useful ways. In addition to sharing their analytic results, they can share their analytic methodology as models for other experts to use. GaaS facilitates this kind of sharing and helps analysts connect with one another through the creation and curation of analytic tool libraries.
The final level of GaaS strives to solve the enigma problem, where there are too many unknowns to leverage expert models and existing algorithms — thus stretching the limits of analytic services. With an enigma problem, the analytic workbench becomes more of a lab environment. Raw tools and data are made available for experimentation and hypothesis testing. First level principles are applied to develop new methods. Exploitation of raw data occurs, and analytics will deterimine if new data needs to be collected or if new software capabilities need to be developed. The outputs of this advanced process are new analytic models. Once the models are validated and time-tested, they are made available to experts or published as new self-service analytic tools. GaaS provides the publication environment for these newly developed tools, and can also provide a way for analysts to request this advanced support.
Technology Advancements Enabling Analytics-as-a-Service
Technology advancement enables new ways for analytic services to deliver value to the GEOINT Community. New kinds of GEOINT data are being collected all the time. New sensors are creating integrated GEOINT data, and open-source and crowdsourced data are creating enormous volumes of information. Spatiotemporal big data systems yield the ability to store and access this data for exploratory analytics. These systems leverage new cloud infrastructure to provide access to everyone in an organization and across the Intelligence Community.
For the simplest form of analytics, new visualization techniques are beginning to take hold. Business intelligence and data science techniques are being applied to GEOINT data to create new statistical views. Linking these tools to the traditional, map-based view of GEOINT further extends this simple form of analytics. Advanced user interfaces allow exploration of more dimensions in the data. 3D technology in browsers makes data visualization richer. Maps, charts, and timeline tools are integrated in simple applications for rich data exploration. In the future, immersive technology such as virtual reality, audio, and haptic feedback could add even more dimensions to analysis.
GEOINT analytic platforms are being implemented and rest on the foundation of big data and cloud computing, adding critical services to the enterprise. They provide environments for combining GEOINT data in predefined visualizations. They allow for the simple sharing of data and analytic tools in easy to configure apps. They create an environment for analysts to share tradecraft and to publish new analytic services. These open platforms provide application programming interfaces (APIs) for embedded developers in an organization to create new analytic tools and applications. As these platforms evolve, they will connect to deep learning and artificial intelligence environments for more in-depth, real-time analysis of data.
Implications of Analytics-as-a-Service
GaaS enables technology that will extend the reach of GEOINT services to the entire analytic workforce. Analysts will be able to draw conclusions and make decisions, predict outcomes, and identify patterns and trends based upon data and the application of analytics. This will increase the demand for geospatial data and tools. The growth of GEOINT to new audiences will require new ways of doing business. Trained GEOINT analysts will need to be placed at different levels to support the demand for advanced methods. Training methods must evolve to train GEOINT consumers in new analytic tools. Policies need to be developed to ensure authoritative models are leveraged and prevent the use of incorrect methods.
Analytic services also require the development of new standards. The nature of analytic tools makes them difficult to standardize, but a fundamental architecture needs to be enforced to ensure logical separation of data, analytic process, and user interfaces. These must interoperate as part of the intelligence enterprise. In addition, analytic tool-sharing standards need to be developed. New metadata definitions will be required to capture the analytic qualities of the service. Metadata definitions need to address standards of quality and reliability, represent uncertainty in the data and model, and describe their fitness for purpose. Competing tools can be categorized and, when necessary, standards enforced.
As the workforce becomes more literate in GaaS, they will require additional training. The so-called “geo-native” is familiar with consumer mapping and understands how to interact with maps and perform simple analytic functions such as navigating to a location. The efficient use of GEOINT analytics goes beyond consumer mapping technologies and requires additional broad level training for a non-geo workforce to leverage the tools correctly.
Professional analysts will also require additional training; rather than simply performing analytics, some analysts will need to be trained as “tool makers.” These tool makers will support the rest of the analytic workforce by developing new tools to exploit new data sources or to assist in answering new questions. They will need to understand basic principles to develop useful and reusable tools. Analysts should be rewarded for sharing their tradecraft knowledge. The tool developers will connect directly to the end users of their tools so they can refine and improve processes over time.
GaaS is the next logical step for providing GEOINT services, moving beyond simple sharing and visualization of data. NGA’s GEOINT Services already provides the foundation technology to enable GaaS and simply needs to formalize the capability to share and develop analytic tools and tradecraft. NGA has hosted application challenges that have proven it is possible to deliver GaaS. The next step is to formalize a program to address the standards and training needs to extend this capability to the entire workforce. GEOINT services can connect to the work performed at other agencies to capture GEOINT tradecraft. This would help to create a rich set of initial capability.
Gaas will be a force multiplier for the GEOINT workforce. The reach of geospatial analysts will extend as new users pick up GEOINT tools. Analysts will also have more time for complex problems as they offload routine requests to simple apps. They will be able to collaborate and share their tools and techniques with like-minded professionals. New data can be collected and structured to support emerging analytic processes.
