Knowledge Builder Software
Knowledge Builder is the XpertRule Rules Authoring Studio used to develop, test, deploy and maintain rules based applications. The main value proposition of Knowledge Builder is the speed of development of rules based applications and the associated web user interface. This is a result of 26 years of investment into R & D reacting to graphical development environments, graphical rules representations, building web user interfaces, and high performance deployment architectures.
Below are examples of the decision tree / flow representation in Knowledge Builder for a diagnostics application:
Knowledge Builder also allows rules to be expressed as If/Then statements or decision tables which are easy to maintain within Knowledge Builder or can be imported from external databases or spreadsheets.
Decision/flow trees in XpertRule can contain 'procedures' which are scripting blocks (in VB script or XpertRule script) for calculations, string manipulations and for calling external programs, web services and data sources. Similarly, data capture forms and reports can be embedded in the tree flows. Below is an example of a flow tree with an embedded procedure:
The rules authoring environment within XpertRule has a graphical user interface with an object explorer showing the attributes, trees, tables, procedures, forms and other components of the application.
The Knowledge Builder Authoring Studio also allows the design of customised user interface forms for capturing attribute/question data. A WYSIWYG (what you see is what you get) graphical screen editor is supported which can be used by business or IT users to define and position what data needs to be captured on each screen. For example, the following design view:
The above form design will result in the following runtime user interface :
In addition to the rules automation technology, XpertRule offers advanced technology for Data Mining which is the automatic discovery of operating/ process rules from historic performance data of asset operation. XpertRule offers two data mining add on technologies:
XpertRule® Process Data Analyser This is a full Data mining environment supporting data preparation, visualisation and knowledge discovery. Discovered trees / rules can be exported to the rules authoring environment for deployment.:
Knowledge Builder Tree Mining add-on: This is an optional module to Knowledge Builder which allows for the discovery of decision trees from within the knowledge builder environment. Discovered trees can be used in rules inference and mixed freely with expert authored trees.
Rule or decision tree induction is the most established and effective data mining technologies in use today. It is what can be termed 'goal driven' data mining in that a operating goal is defined and rule induction is used to generate patterns that relate to that goal. The operating goal can be the occurrence of an event such as 'on target production' or the magnitude of an event such as 'energy use' or 'efficiency'.
Below is an extract from a data mining tree related to a dataset from a crude stabilization terminal. The objective of applying rule induction data mining was to discover patterns relating to total power usage taking into account factors such as plant throughput and column temperatures and pressures. Each leaf represents a mode of operation defined by a set if rules eg leaf 12 below Column Press < 13.22 and Crude Inlet Temp > 22.56 Then Avg. Energy Use = 18.68.