Smart systems that support increasingly incisive decision making and data-driven insight are the future of business. These systems unlock the much-trumpeted potential of Big Data, transforming it from a raw material with a great deal of potential into something that can be actively utilized and wielded.
This is where cognitive systems have a significant role to play — cognitive systems that are supported at every turn by AI-based technology.
What are Cognitive Systems?
Cognitive systems work along the same lines as the cognitive processes in the human brain. They gather data, store this data, and then use it to execute decisions and actions. While the human brain is adept at reaching qualitative and interpretative conclusions, cognitive systems work in a slightly different way.
These systems are designed to process far greater volumes of information. Working with immutable records and performing high-level calculations and procedural operations, these cognitive systems form the “brain” of a business, providing the framework of data insight required to achieve more in the modern landscape.
Different Types of Cognitive System
Cognitive systems can be divided into three separate types — record, engagement, and insight. However, these three types operate together, creating a symbiotic relationship that forms the basis of the overarching cognitive system.
System of Record
A system of record is designed to be a comprehensive and complete data resource for a business. This is the centralized touchpoint that software applications and human users draw upon as they make business-critical decisions.
At a micro-level, this might not seem so revolutionary. After all, human users and smart software pieces recognize the need for a unified data source. Providing that there are no fundamental errors, this unified data source should be relatively easy to achieve. However, when we expand the scope and begin looking at mission-critical functions on a macro-level, the system of record becomes more crucial, as it is easy for inconsistencies to arise at this level.
When data becomes siloed, processes become siloed too. Without a centralized data resource provided by the system of record, source data inconsistencies quickly become operational inconsistencies.
To combat this, the system of record must adhere to four key principles:
- Completeness — The system must be a unified source of the data required by all processes and programs.
- Accuracy — Data must be correct.
- Timeliness — Data must be up-to-date or as close to up-to-date as possible.
- Consistency — There must be no inconsistencies or disparities within the data source.
System of Engagement
Another key aspect of cognitive systems is the system of engagement. These systems operate differently from systems of record and occupy a different position within the business’s operational landscape.
While systems of record align organizational processes via a unified, back-end data source, systems of engagement exist further forward and are used directly by teams operating in the field. For instance, a team on the shop floor will need to be able to oversee and audit the material inventory available to them, as well as executing commands and operations based on this knowledge. This will all be handled via a process application that forms the system of engagement.
This represents an important distinction between systems of record and systems of engagement. The system of record is a business-wide resource — a consistently accurate monolith of data designed to unify operations. Systems of engagement, on the other hand, are geared towards specific operations, which means your shop floor teams may use a specific set of systems of engagement, while your marketing personnel may use a very different set.
All of these systems of engagement, however, will draw their data from the same centralized system of record.
System of Insight
While the system of record and the system of engagement perform very different roles, there is already a point of connection that bridges the gap between them. This point of connection is data. In simple terms, systems of engagement need timely and accurate data to function properly, and the system of record provides this data.
To continue these simple terms, it may appear that the relationship between these systems is complete — data is stored in the system of record and acted upon in the system of engagement. There is no need for a third system.
In practice, however, there is a need for a third system — this is where the system of insight comes into play.
The system of insight creates a more sophisticated and meaningful connection between the systems of record and engagement. Systems of engagement need to be able to do more than simply draw upon data from a unified source. They need to be able to draw upon the right kind of data — data that has already been processed and translated from its raw form into actionable insight.
This role is fulfilled by the system of insight. When data is created or written into the system of record, the system of insight begins its process. It draws upon this data directly and refines it into a form that can then be executed as an operational action. This occurs across six primary phases.
The Consumption Phase
In the consumption phase, the system of insight determines what data is important and pertinent. Modern systems of data storage are vast, and systems of engagement struggle to process these data stores efficiently. With the system of engagement in place, the right datasets — at the right volume — are delivered to the system of engagement.
The Collection Phase
With the correct datasets determined and identified, the system of insight begins to collect the required data. Systems such as IGM DB2 and IBM InfoSphere BigInsights work to narrow the data parameters in a meaningful way, creating manageable data structures ready for analysis and reporting.
The Analytic Phase
All data is analyzed to some extent. This analysis can take many forms — perhaps involving decisions on relevant or irrelevant data or identifying fraudulent activity through anomalous transactions or events. However, in the system of insight, this analytical capability is taken a step further, leveraging automated functions to process and analyze data at incredibly high volumes, high speeds, and across a variety of data types.
The Reporting Phase
Even once data has been analyzed, it still needs to be translated into actionable insight. While automated functions built into the system of engagement can achieve this to an extent, reporting enables real-time understanding of data structures from human operators. The system of insight needs to deliver this, drawing upon stored data, data in the process of transfer, and data that is currently in use to provide an up-to-the-minute snapshot.
The Detection Phase
Detection takes analytics and reporting to the next level, actively identifying situations and occurrences that are of specific interest to the business. For instance, converging datasets that represent business opportunities or anomalous results that indicate risk.
The Decision Phase
The decision phase provides the logical basis upon which a decision can be taken. In many cases, this will be very simple — perhaps identifying a customer need for a specific product or service — while in others, it will be more complex. It is in this more complex process of decision logic formation that the system of insight will be required.
The Role of Artificial Intelligence in Cognitive Systems
The cognitive system — formed of the systems of record, engagement and insight — cannot exist without artificial intelligence. Data volumes have grown too large, and processing speeds have grown so rapidly that human operators cannot keep track. This is where AI needs to take up the slack.
It is in systems of insight that AI is leveraged to the greatest effect.
Artificial Intelligence in Consumption and Collection
At a very basic level, AI is used to organize data as it is consumed and collected. For example, data is classified as pertaining to when an event occurred, and to the geographical location the event took place in. AI recognizes these data sources and assigns them the appropriate tags.
However, as volumes increase, AI needs to work in a more sophisticated manner. One example of this is found in manufacturing, where ebbs and flows in the supply chain can result in falls in production or overloading of equipment respectively. The system of insight will need to recognize this relationship.
Artificial Intelligence in Analysis and Reporting
Analysis takes the data relationship identification function mentioned above and evolves it. The AI-based system will cross-reference real-time data streams with historical data, identifying new relationships and trends. In addition, the system of insight will need to work in a cognitive fashion, recognizing differences between correlation and causality and making value judgements based on this.
The AI will also need to identify the relevant datasets for automated reporting and draw upon historical reports to build an enhanced understanding of today’s data landscape.
Intelligence in Detection and Decision Logic
Before a decision can be taken, teams first need to recognize if there is any need for a decision at all. In this sense, the system of insight’s AI-based function works to detect anomalies, discrepancies and deficiencies, before identifying how an operational change or other alteration can improve the situation for the better.
With the logic established, the system of insight can then automate key elements of the decision-making process to improve efficiency and effectiveness. Platforms such as IBM’s Operation Decision Manager work on rule-based procedures to execute responses based on real-time data streams and data stores.
Artificial Intelligence in Cognitive System Integration
While AI can help systems of insight to identify and make decisions, integration with the system of engagement is still required to actively execute this decisions. Similarly, the system of insight needs to integrate with the system of record to draw upon unified data sources.
Artificial intelligence again comes into play, bridging the gap between these systems and managing their interactions. Human team members still have their roles to play in assessing and analyzing data, but it is AI-based automation that does the majority of the heavy lifting.
Build Your Cognitive Systems with IBM
IBM provides a range of smart, AI-enabled products designed to help business owners build their cognitive systems. As an IBM Platinum partner, we are perfectly positioned to help you connect with these all-important products. Reach out to our team today and discover more.