What an EDMS is: a complete guide to holding originals, values, and history together

An EDMS (electronic document management system) keeps the originals, attribute information, and processing history of an organization's documents in one place, and manages searching and viewing according to permission.

Extracted data needs a destination too

Extracted data needs somewhere to arrive.

Discussion of document automation concentrates on recognition and extraction. Yet the point where the flow actually breaks frequently arises from the question of where the extracted values go.

Put only the values into the business system and manage originals separately, and the grounding cannot be checked later. Keep only originals with no values flowing, and the operator types them again. That is why production configurations commonly load extracted data into the EDMS and then register it into core systems through an automation tool. Work continues only when originals, values, and history connect in one place.

What an EDMS actually is: how it differs from a file server

File server and EDMS across three axes

First, the unit of management differs. A file server manages files and folders. An EDMS manages documents. Multiple versions of the same document, its attributes, and who viewed or edited it and when all travel with the document.

Second, the objective differs. A file server aims at storage and sharing. An EDMS aims at controlled retention and traceability. Restricting access by permission and retaining processing history are basic functions.

Third, retrieval differs. A file server is searched by name and path. An EDMS is searched by attribute. Once fields extracted by document automation become attributes, an original can be found directly by contract number or business registration number.

The two are not in opposition. But where audit response and history tracing are required, a file server alone struggles to meet the requirement.

Four conditions for real integration

Four conditions that hold in integration

First, define the attribute fields in advance. Decide which of the extracted values become EDMS attributes. That list is exactly what can be searched on later.

Second, store originals and results together. Binding the original file, the extraction result, and the source position of each value to the same document makes review and audit response straightforward.

Third, retain processing history. Whether a case was processed automatically or corrected by a person, and what was changed and how, has to be recorded.

Fourth, align the permission scheme. Viewing scope frequently differs by document type, and the same standard has to apply to documents loaded by automation.

How an EDMS is applied in practice

Decide when loading happens

Settle whether values are loaded immediately after extraction or only after passing validation. Loading pre-validation values mixes them with later corrections; loading only post-validation leaves the originals of exception cases nowhere.

In practice, storing the original immediately at intake and attaching extracted values as attributes once validation passes is common. The arrival of the original and the confirmation of the values are handled separately.

Choose attributes by working backwards from search

When defining attribute fields, work backwards from how they will be searched. Searching by contract number, by business registration number, or by intake date range each calls for different attributes.

Not every extracted value needs to become an attribute. Loading values that will never be searched adds management burden and nothing else.

EDMS in the Korean environment

Retention periods for documents are frequently fixed by regulation in Korean finance and the public sector. Retention period and disposal procedure differ by document type, so that classification has to apply to documents loaded by automation as well.

Audit requirements add a history of who viewed and modified what and when. For automatically processed documents, retaining which rule determined them normal is the appropriate configuration.

Frequently asked questions