Bill DeStefanis, Nuance
When watching old movies, viewers will often see antiquated scenarios and technology. Characters speaking on cell phones as large as bricks or flipping through encyclopedias to conduct research are a good reminder of how far technology has come in just the past few decades.
Another outdated scenario is administrative assistants manually filing and organizing documents. With the integration of artificial intelligence into document capture and management technology, businesses will no longer have to rely on manual distribution and organization of document-based data. Furthermore, artificial intelligence (AI) will evolve with more advanced natural language processing (NLP) which will enable a whole new level of efficiency and convenience to the task of managing “piles of paper.”
Identifying and Classifying Text
Beginning with paper documents that can be scanned and transformed into digital machine-readable files, today document management software can identify and classify information contained within the document. Today, documents are only “recognized” but not “understood.” This lack of understanding of content restricts how fast and efficiently documents can interact with business systems. Using artificial intelligence and natural language processing, we can mimic human behavior as to how documents are processed based on “understanding” of the content.
In the near future, such software systems will have the ability to identify whether a document is a contract or insurance claim. Once a document is identified by type, AI techniques can be used to identify content within the document, including social security numbers, monetary figures and terms like “classified.” These capabilities will be applied to many document processes such as invoice processing, where an AI-enabled system allows a document to be classified as an invoice, extract information it based on relevant parties, and then route the document to the appropriate destination.
In the future, as machine learning further evolves, it will allow document classification to derive even more meaning and textual understanding based on learned behavior. This will allow organizations to create smarter processes. For example, when a contract is captured and transformed to a machine-readable format, the software will identify dates of execution and classify the prioritization of the subsequent action based on how close the date of execution is to the date the contract is entered into the system.
Such capabilities give organizations the flexibility to structure their document workflows to optimize efficiency. These benefits will grow exponentially as machine learning and artificial intelligence technologies improve.
Prevent Sharing of Sensitive Information
According to the Ponemon Institute’s 2017 Cost of Data Breach study, malicious attacks are the leading cause of data breaches worldwide. Artificial intelligence can add another layer of security to safeguard document data. For instance, document capture systems with artificial intelligence capabilities can identify terms such as “confidential” and take action by watermarking the document, redacting information or simply not releasing the document to an email system or fax gateway.
As artificial intelligence technologies improve, these systems will evolve beyond simply identifying specific terms in a document to more comprehensively understand the context of the document through an ability to interpret “unstructured text.” For example, consider a document that references a company’s unannounced new product, but does not explicitly mention the new product by name. As machine learning evolves to more advanced AI, document capture software will recognize even loose references to the unnamed product, classify that language as relevant to the new product, and redact the reference.
Into the Future
As document capture and workflow software become increasingly equipped with artificial intelligence capabilities, organizations will be able to set up parameters to automatically route certain sections of documents to relevant parties. For example, a contract captured by an MFP can be routed to the legal department, while the financial information could be extracted and routed to the finance or accounting department, limiting the number of people who have access to potentially sensitive information by only passing on what is relevant to each department.
Someday, these systems will process massive amounts of information from piles of paper by extracting only the relevant data – a task that today takes hundreds of man-hours. Soon, the concept of manually filing and sharing documents will seem as antiquated as cell phones that are as large as bricks.
