Celaton's Chief Technology Officer provides some clarity for and defines
some of the most commonly used terms in the industry
There are a lot of different terms being banded about within the software and technology industries at the moment and it can be confusing to make sense of it all. The below is a list of essential terms that organisations need to know in order to succeed in their digitial transformation journeys.
Business Process Management is the practice of optimising processes and procedures to allow teams to become as efficient as possible by focusing on the tasks that matter, conducting them in the correct order and eliminating any that are unnecessary.
Digital Process Automation is an evolution of Business Process Management. It takes the appropriate manual tasks within a process and making use of computer systems helps to organise and perform them more efficiently, eliminating unnecessary repetitive tasks altogether by having the computer system carry them out instead. Digital Process Automation relies upon structured data and rules based decision making.
Intelligent Process Automation is an evolution of Digital Process Automation. Instead of simply using rules based decision making, which relies upon structured data, computer systems that implement approaches such as Machine Learning, Natural Language Processing or Intelligent OCR are used to additionally eliminate tasks within a process that normally rely upon a human’s intelligence.
Intelligent Document Processing is a specific form of Intelligent Process Automation (IPA). It refers to the IPA of business processes specifically concerned with processing a document. A document can refer to a printed page that has been scanned, an email or a soft copy office document such as a Word, PDF or spreadsheet file.
Robotic Process Automation is a type of business process automation. RPA is designed to automate the repetitive tasks business users find themselves performing on a daily basis because the software applications they use don’t integrate well with each other. Similar to how macros can be recorded in applications such as Excel and Word, RPA can be thought of as a macro recorder and executer for the entire operating system and all applications installed on it. RPA also includes many of the tools expected of a scripting language having the ability to read and write structured data from files and the ability to make calls to web services. Unlike a scripting language, however, RPA robots are configured using a drag and drop Graphical User Interface (GUI) and aimed at technical business users, although in reality having some sort of software development experience is advantageous in configuring them. RPA helps address the issue of IT departments having multiple demands on their time along with the business users requirements to streamline the manual tasks they perform which require little or no skill by allowing them to carry out the configuration themselves.
Artificial Intelligence is not a new term, it has been utilised for decades in computer science. It is the ability of a machine to mimic a human’s ability for learning based on experience and problem solving behaviour. Humans exhibit intelligent behaviour in many different ways, from being able to recognise peoples faces, driving a car or finding key data and information within an unstructured document by reading the entire document and inferring meaning. Horizontal AI can be thought of as a computer system that mimics many of the aspects of human intelligence using the same core AI computer system. What has been seen more recently is the adoption of Vertical AI where one computing system is developed to learn and mimic one specific human intelligence; it is unlikely you will see a driverless car that is also capable of processing your invoices for example!
Machine learning is a subset of Artificial Intelligence. It is the use of computer algorithms and statistical models implemented to have the ability to automatically learn and improve from experience without being explicitly programmed. There are differing approaches to machine learning; Supervised Learning, Unsupervised Learning and Reinforcement Learning, the specific problem to be solved using machine learning will dictate which method should be used.
Natural Language Processing is a subset of Artificial Intelligence that deals with the interaction between computers and humans using natural language. It is the use of computer algorithms to read, decipher, understand and make sense of human languages and often relies upon a document having been “read” using OCR or speech having been converted to text using speech recognition.
Cognitive Computing is the ability for a computer to reason with purpose and interact with humans naturally. By mimicking thought and reasoning like a human it can be thought of as the ultimate Horizontal Artificial Intelligence. The ability to perform many tasks we would consider only previously possible with human intelligence.
OCR has been around for many decades and is the ability for a computer to take a typed document which has been scanned or photographed into an image and convert that image into text. Taking our definition of a Vertical AI as the ability to mimic a specific human behaviour, recognising characters and words on a page, then OCR can be considered to be a very specific form of AI. However, OCR itself infers no meaning or understanding of the text it has extracted from an image, it simply converts the printed page into a text document which requires further processing to make use of it.
ICR is a superset of OCR and is the ability for a computer to take a hand written document which has been scanned or photographed into an image and convert that image into text. It specifically refers to the recognition of hand written characters, more recently however the term ICR has had a 2nd definition introduced.
With the increased use of Machine Learning algorithms being used on text that has been OCR’d from an image to not only recognise the printed text but also make some human like intelligent use of that text the term Intelligent Character Recognition has increasingly been used to describe such computer systems.
Average Handling Time is an important metric when measuring the effectiveness of Digital Process Automation; it is the average amount of time a human takes to process one transaction or document. Often there is the need to keep a ‘human in the loop’ of a process, having them still perform higher functioning tasks which cannot yet be addressed using computerised methods such as rules based decision making, workflow, Robotic Process Automation (RPA) or even Intelligent Process Automation (IPA). By reviewing the end-to-end manual process and breaking it down into tasks, there will be many tasks that can be computerised and those that cannot. The time spent by an operator carrying our manual tasks will therefore reduce dramatically by leaving them only with those best suited to a human operator.