Machine-Readable Documents

Machine-readable documents are documents whose content can be readily processed by computers. Such documents are distinguished from machine-readable data by virtue of having sufficient structure to provide the necessary context to support the business processes for which they are created. Data without context (language use) is meaningless and lacks the four essential characteristics of trustworthy business records specified in ISO 15489 Information and documentation -- Records management:

The vast bulk of information is unstructured data and, from a business perspective, that means it is "immature", i.e., Level 1 (chaotic) of the Capability Maturity Model. Such immaturity fosters inefficiency, diminishes quality, and limits effectiveness. Unstructured information is also ill-suited for records management functions, provides inadequate evidence for legal purposes, drives up the cost of discovery (law) in litigation, and makes access and usage needlessly cumbersome in routine, ongoing business processes.

There are at least four aspects to machine-readability:

As early as 1981, the U.S. Government Accountability Office (GAO) began reporting on the problem of inadequate record-keeping practices in the U.S. federal government. Such deficiencies are not unique to government and advances in information technology mean that most information is now "born digital" and thus potentially far more easily managed by automated means. However, in testimony to Congress in 2010, GAO highlighted problems with managing electronic records, and as recently as 2015, GAO has continued to report inadequacies in the performance of Executive Branch agencies in meeting records management requirements. Moreover, more than two decades after a major and formerly highly respected auditing firm, Arthur Andersen, met its demise due to a records destruction scandal, record-keeping practices became a central issue in the 2016 Presidential election.

...
Wikipedia