The Freedom of Information Act is a win for free speech and transparency. But what happens when you fail to obtain a timely response to your request? You may not realize how grueling and stressful the processes involved in sourcing the requested information are, particularly given the explosion of data growth and hoarding in the federal government.
Even though many government agencies have digitized their FOIA request data sourcing and response processes, it hasn’t resulted in any significant increase in response time, nor has it lessened FOIA request burdens. So, what does this tell you?
The advantages of digitizing FOIA request data search, sourcing and response time haven’t been fully realized. Instead, the mandate for digital records only increased the digital load on federal agencies.
Pathways to Lessening the Burden
It is essential to lessen the FOIA burden on government agencies. Using a framework that includes automated thematic analysis and associated aspects of machine learning, natural language processing and statistical clustering allows agencies to more effectively manage FOIA requests.
Agencies can overhaul their approach in four steps:
- Data Asset Inventory: Analyze information at rest in its original location to ensure only necessary information is curated into intuitive data clusters. This builds an inventory of metadata facets as well as the relevant in-file data for effective conduction of a targeted pre-search, without exhausting agency resources.
- Data Cleanse: The cleaning process involves determining what data constitutes as high-value or sensitive from that of lesser value.
- Model and Auto-categorize: Modeling typically involves the agency subject matter experts sorting out the collected information inventory by applying various policies and rules involved in their dispensation, depending on the specific FOIA request.
- Report: A final report will then be delivered. The key is working smarter and with the FOIA request process.
When taking these steps, it’s important to note that search isn’t enough. A typical data search within a large agency can bring up 200,000 to 2 million results with tens of terabytes of unstructured/semi-structured data. It is very easy to get overwhelmed by such a massive number of results. As a reviewer, this does not help you optimize the response time.
And leveraging artificial intelligence is still limited in capability. Artificial intelligence can have a role in automating and increasing the efficiency of FOIA processes. However, identifying the correct datasets to properly train AI models is more time-consuming and resource-intensive than it is effective.
But agencies can start at the ground level bringing FOIA into policy review. Review agency directives and policies containing approved information management controls and reporting requirements that may need to involve a FOIA program review against any business products generated as the policy is executed. Listing these business products as “exempt” or “responsive” helps to prepare offices responsible for executing policy requirements.
Agencies also can create a lightweight metadata index of thematic properties. An unmanaged data organization system leads to heavy backlogs. Therefore, creating an index involving mapped data, with obtained and available core associated metadata, leads to more effective and efficient decisions made from the resulting lightweight metadata. But agencies should avoid retaining a full-text index, which is resource-intensive and not scalable, holding close to 100% of the total volume under management.
Auto-categorization is the best approach to lessening the burden is to incorporate human subject matter experts. This should done alongside policy agnostic data modeling tools against an inventory of high-value, pre-discovered metadata facets to quickly discover relevant responsive records for the incoming request.
Although agencies are awash in data, by spending time reviewing and curating your data, FOIA requests can be handled in a much more efficient manner.
Tom Jacobs is a senior federal sales executive and James Jones is a senior federal sales engineer at ActiveNav.
source: NextGov
