Editor’s Note: Dr. Daniel Maxwell is president of KaDSci, a veteran-owned analytics boutique. Maxwell founded the USGIF Modeling & Simulation Working Group and has previously served on both the Defense Science Board Task Force on Intelligence and was a member of the Military Operations Research Society’s Board of Directors. Dan has a Ph.D. in Information Technology from George Mason University (GMU) and continues to teach analytics related topics at GMU.
The accelerated growth in high velocity, high resolution, high dimensional geospatial data is making machine reasoning an essential part of the intelligence analysis process. The above high-resolution visualization of Pittsburgh is extremely complex and only scratches the surface on all the types of data relevant for analysis at that location.
The Intelligence Community Information Technology Enterprise (IC ITE) and activity-based intelligence (ABI) are primary examples of how the Intelligence Community is recognizing and investing in this requirement. While this investment is necessary, it is so far not sufficient to achieve the machine learning capabilities the IC seeks. Achieving the envisioned capability requires the ability to—with minimal human intervention—meaningfully access, share, and process disparate data. Having access to and applying consistent data standards is key to achieve this outcome.
There is a delicate balance between a “one ring to rule them all” and a “laissez faire” approach to managing geospatial data. While analysts of all types require the flexibility to adapt their data and tools to meet continuously evolving requirements, the sheer magnitude and complexity of the data requires machine reasoning to present them with relevant data sets to analyze. Machine reasoning requires consistency across geospatial detail, time, and dimensionality that does not currently exist. The best-case outcome of this shortcoming is analysts perform below the optimal level because an analysis task takes longer to accomplish or is less than complete. The worst-case scenario occurs when key pieces of evidence are either not considered or are presented out of context. Decisionmakers are then presented flawed intelligence. The resulting risk of this shortcoming needs to be mitigated in a consistent and cost efficient manner.
The case for supporting open standards begins with recognition that most of the high velocity geospatial and geo-referenced data will come from open, commercial sources. The IC cannot (and should not attempt to) control the marketplace for these sources. Barriers across historic communities of technical interest are all but collapsing. Fields such as simulation, computational social science, economics, urban planning, and disaster response all have models and data relevant to meet intelligence analysis requirements. This growing interdisciplinary environment is now highly dependent on each contributor’s data and models—differing drastically from previously independent intellectual disciplines. The final point is that many vendors are using proprietary data standards and models as a tool to lock clients into their products and preserve a revenue stream.
Substantial progress is being made by the geospatial standards community to support interoperability between systems and data sources, and the international vendor community has broadly adopted these standards. Open geospatial standards have solved many of the basic geospatial data fusion challenges, and already support the integration of many types of data. However, these same standards have not been mandated and implemented consistently across the user community, leading to inconsistent and often incompatible exchanges across computer systems, agencies, and software applications. The challenge is exacerbated by the rapid pace of technology change, as well as the proliferation of analysis and visualization tools that in some cases ignore standards in favor of proprietary data formats.
It is in the U.S. government’s interest to actively engage in the development of open standards and to align on common, standards-based best practices for sharing consistent multi-resolution, high dimensional data.
There are efforts underway in multiple standards organizations to apply proven consensus methods to advance open geospatial data standards. One example is the Open Geospatial Consortium’s effort to advance the Common Database (CDB) standard, an open multi-resolution, high dimensional standard for geospatial data. This standard was initially created via a U.S. Special Operations Command investment to support their planning, rehearsal, and analysis requirements. CDB is a mature candidate for meeting the needs described above. However, the standard will need advancement informed by multi-disciplinary and multi-organizational collaboration as well as experimentation and engineering investments.
Achieving the machine reasoning the IC needs requires increased near-term government investment in the advancement of data standards. This investment is critical to reach the interoperability, cost efficiency, and culture of continuous innovation the IC seeks. These investments would be significantly less expensive than the ongoing investments in hardware and software. Socially and organizationally, these changes will be extremely challenging. However, with a clearly articulated senior leadership vision, dedicated investment, and near-term persistence the potential is there to simultaneously reduce long term data-related costs and increase the effectiveness of the IC’s information infrastructure.