Leverage AI to unlock data’s potential during electronic records transition

Agencies are at various stages of their journey to meet the National Archives and Records Administration electronic records deadline at the end of 2022. In fact, at the most recent Digital Government Institute E-Discovery, Records & Information Management Virtual Conference, event organizers surveyed the audience to get a sense of how far along participants were in their digital transformation efforts. Unsurprisingly, the results came back rather even divided among “making progress,” “we have a long way to go” and simply, “don’t ask.” In our discussions with government agencies, we found similar sentiments.

No matter where agencies are in their electronic records management journey, they should take advantage of the opportunity to put into place capabilities that can unlock insights from that information. That is easier said than done, however, as most agencies struggle to uncover the value of their information. They have too much unstructured and unclassified information, or they lack the internal resources and the skills to analyze it -- or both.

Integrating artificial intelligence and machine learning into records management

NARA’s deadline mandates that all agency records must be 100% digital by the end of 2022. Today’s distributed agency environment is characterized by diverse systems with multiple data types from various data sources, including physical and digital information. This is causing headaches for agency records managers responsible for delivering on the NARA electronic records management requirements. But with AI- and ML-enabled content analytics, data management and information governance tools to classify, extract and enrich physical and digital content, agencies can successfully navigate these pain points.

The integration of AI and ML capabilities into an agency’s information management program during digital transformation efforts will enable them to automatically classify, extract and enrich physical and digital content. At a high-level, the three-step process would look something like this:

1. Input records – The system uses optical character recognition, image recognition and video processing capabilities to ingest information and save it in an electronic format.
2. Apply AI/ML – Using classification processes and entity extraction combined with AI/ML tactics, the system will automatically train a model, apply it, capture feedback and prepare the data for the next phase.
3. Output – Once the data is prepared, it is exported into a data visualization function that enables agencies to search, analyze, vie, and generate deeper insights to make better decisions from that information.

Benefits of an AI/ML approach

In terms of digital transformation and streamlining of the records management process, specific benefits of incorporating AI/ML include:

• Metadata for search and tracking. Metadata tagging is part of the digitization process that helps agencies properly sort, search and manage information. Metadata is vital to managing, accessing and eventually tracking information throughout its lifecycle. When used in conjunction with AI/ML, metadata can provide vital information about agency records, such as what the content is and what characteristics it possesses.
• Data ingestion from multiple sources. AI/ML systems help agencies ingest information from siloed sources and classify each piece of information by type with the associated metadata.
• Data classification. The combination of supervised and unsupervised learning in AI/ML solutions helps to avoid the need to know all document classes as these systems find, separate and learn about any unexpected document classes in the data automatically. This process trains the neural networks connected to the AI/ML platform, so that once document and metadata indexing is complete, a baseline library is established, which can then be searched for patterns and trends.
• User interface for visualization. Adding in the ability to view/visualize information in a common format provides more accountability and transparency across an agency, while offering more valuable data to decision-makers.
• Continuous improvements. AI/ML systems continually use human signals such as manual corrections to improve the electronic learning to deliver more accurate results.

Although the process of moving all records to a digital format can be daunting -- especially given the volume and variety of records that agencies manage -- it does provide an opportunity for agencies to derive value from that information. By incorporating AI/ML technologies into their records management program, agencies can unlock that potential within their information to help make better, more informed decisions and enhance the services they are providing across the board.