SmartCloud Workbench is the next generation tool for situational reasoning application implementation. Workbench enables interactive design, specification, deployment and on-going maintenance of artificial neural network applications. It centers on development of Cognitive Maps, a goal-directed situationally-aware knowledge representation inspired by work in cognitive modelling, decision logic, and reinforcement learning. Cognitive Maps are frameworks able to leverage multiple technologies for data storage and manipulation, system modelling, and artificial intelligence, while retaining a focus on symbolic knowledge representations that fully support explainability through natural language interfaces.

CHALLENGE

• To make creation and managing of cognitive map easy for user on different devices.

• To make application architecture scalable and flexible for implementation of new features/different users requirements/systems.

• To make creation and managing of decision table easy and clear for user.

SOLUTION

• Usage of best practices for mind maps, user interfaces and continuous integration.

• Usage of flexible app architecture solutions, development tools and methodologies, rich preferences for customers personalisation.

• Usage of best design practices for managing huge amounts of structured data.

Cognitive map

• Creation and management of Cognitive Maps is a highly interactive task for users. Cognitive Maps evolve through feedback from visualizations that highlight relationships between goals and situations, show where the map needs to be expanded, changed or reduced. The Workbench is designed to enable interactive creation with intuitive displays and ease-of-use features.

Bayesian Network

• The Workbench Bayesian Network (BN) user interface enables a user to: Manually create BNs by creating Nodes and Links, and entering BN Node/Link data. Automatically learn a BN structure from user-supplied data (via CRex API). Automatically learn BN node probability tables from user-supplied data (vis CRex API). Manually modify the BN node structure, parameters (states, probability tables), and connections. In a Decision Table, use evaluation of a valid BN (via CRex API) to assess part of a Condition Clause and display the results of evaluation.

• Simulation Capabilities When developing Cognitive Maps with the Workbench, users often simulate a running application so that they can examine the results and modify the Cognitive Map to achieve desired reasoning behavior. To do this, they use Simulation capabilities of the Workbench. These capabilities allow the user to do the following for a single Cognitive Map that is complete and can be Published:

• Connect to a Microsoft SQL database containing data that are the inputs to a CRex application.

• Edit the input data.

• Visualize results of the simulation in an embedded Microsoft Power BI display.