Knowledgebase Development

Contracted Knowledgebase Development
The Value of Good Information
The Evolution of a SpotLight Knowledgebase
Data Source - FIM
Data Source - FMEA
Data Source - Actual Field Experience
Subject Matter Experts (SME) and Solution Prototype Tables (SPT)
Solution Finishing and Hierarchy
Quality Assurance
Conclusion

Contracted Knowledgebase Development

CaseBank has a decade of experience in the development of valid, effective knowledgebases for use in field service software and guided diagnostics. Most customers recognize the value of that experience and choose to contract with CaseBank for their knowledgebase development.

1. Analyze your historical maintenance records, existing fault isolation trees, and reference material.
2. Ensure that sufficient information exists to create an effective knowledgebase with
   • enough coverage to be useful and provide value on primary systems
   • enough coverage to be useful and provide value on specified subsystems and functional areas,
3. Extract maintenance and systems data from existing sources for inclusion in the knowledgebase
4. Sanitize any information that might identify the organization or specific individuals
5. Edit and repair solutions for maximum troubleshooting effectiveness
6. Finalize the "seed" knowledgebase

At this point, the seed knowledgebase is deployed to the customer's server, and knowledgebase authoring is transitioned from CaseBank to designated customer staff with appropriate training. Over time, the knowledgebase grows and becomes increasingly effective, as new information from the field is incorporated through system usage.

Benefits of this approach:

• Allows you to leverage CaseBank's experience and expertise for timely, cost-effective development of your seed knowledgebase
• Reduces your technical risk since CaseBank develops the initial domain structure and your in-house authors simply flesh it out rather than designing it from scratch.
• Simplifies future solution authoring because CaseBank-authored solutions offer "best practice" examples and templates

The Value of Good Information

The existence of a sound knowledgebase is essential for the operation of SpotLight and SpotLight Mentor, and is an area where CaseBank has pioneered many core techniques and abilities. CaseBank has developed tools and created processes designed to make the establishment of the initial or "seed" SpotLight knowledgebase ("KB") as painless as possible, minimizing time and risk. Importantly, CaseBank's active approach to the creation of the initial knowledgebase is aimed at relieving customers of the need to develop or maintain their own expertise in this highly specialized field, while also offering a proven path to self-sufficiency at the chosen time.

The Evolution of a SpotLight Knowledgebase

The initial knowledgebase ideally contains all known and available failure knowledge at the time the database is created. CaseBank assembles a symptom/solution correspondence from as many sources as possible, although three broad categories of source data are typically used for knowledgebase development:

• Fault Isolation Manuals ("FIMs") from
  • The primary equipment OEM's documentation
  • The major system vendor documentation
  • Other platforms for similar systems that CaseBank has encountered
• The OEM Engineering Failure Modes and Effects Analysis ("FMEA")
• Actual field experience, comprising
  • Expert tacit knowledge
  • The OEM and vendor's own reliability analyses
  • CaseBank's analysis of equipment records

This information is then extracted, sanitized and incorporated into representative solutions that the SpotLight reasoning engine accesses during troubleshooting sessions.

Fault Isolation Manuals (FIM)

Existing FIMs are troubleshooting procedures, identifying failures that the OEM's design engineers have decided are likely to occur. Clearly, these contain valid information for the knowledgebase. SpotLight Solutions are built using the tests and observations requested in the path through the FIM by CaseBank's knowledgeable and expert staff. CaseBank analyses the fault isolation trees, starting with the initial symptom, and associated "descriptors", which are derived from measurements and tests to differentiate fault causes. The result is a set of derived "solutions", which are loaded into the initial knowledgebase.

Failure Modes and Effects Analyses (FMEA)

The OEM FMEA analyses identify failure modes and symptoms, which can be extracted into the SpotLight solutions format for incorporation in the knowledgebase. If the data supplied is in the form of a FMEA database, it may need to be pre-processed before performing a FMEA import into the SpotLight knowledgebase, using specialized CaseBank software tools.

Actual Field Experience

Where field experience is available in the form of maintenance records, CaseBank has tools, such as Rondo^TM, that assist developers in mining those records to extract and cluster related records. The tools used by CaseBank typically reduce hundreds of thousands of historical maintenance records into several hundred, manageable and unique solutions that finally enter the knowledgebase. This is the most difficult source of knowledge because those records seldom contain all the symptom information needed by SpotLight, yet it is also the most valuable source of knowledge because it reflects absolutely what is failing on the equipment.

Subject Matter Experts (SME) and Solution Prototype Tables (SPT)

That issue is resolved through Subject Matter Expert ("SME") interviews and Solution Prototype Tables ("SPTs")

SPTs are the final step in rounding out the knowledgebase. SPTs guide interviews with experienced technicians. In order for SpotLight to provide thorough troubleshooting guidance, the knowledgebase needs to contain a solid complement of failure modes. With SPTs the existing knowledgebase content is organized into a format conducive to eliciting SME confirmations of root causes and any further information required to enhance the knowledgebase. This process allows the human experts to ponder the causes of key symptoms without having to worry about the troubleshooting steps. Data is also thoroughly checked against equipment, system design, and/or maintenance documents, by CaseBank's own expert staff.

Solution Finishing and Hierarchy

After the consultation and data validation phase, the table is converted into solutions using knowledgebase development software tools. "Solution Finishing" is done by enriching the solutions with detailed explanations, descriptions of symptoms, and references to design and links to maintenance documents.

Data elements in a knowledgebase are organized in a convenient domain hierarchy that is familiar to the end users. Typically this mirrors the subject breakdown found in the equipment technical manuals, or a process functional flow, or a combination of both.

Quality Assurance

CaseBank has a proven effective and efficient methodology for building knowledgebases, supported by a suite of semi-automated tools. To maintain standards among developers, a project style guide is developed at the beginning of each project. Technical authors, peer developers, knowledgebase administrators, and subject matter experts are involved at various phases of the knowledgebase development process in order to assure the highest level of solution quality. Solution authoring tools enforce the workflow and quality review processes and also provide tracking and resource management. Domain modification and tracking processes ensure that any modifications to the domain are approved by authorized personnel and tracked to ensure data integrity and adherence to design standards.

Conclusion

CaseBank routinely builds effective SpotLight knowledgebases for very complex equipment on time and on budget with low risk. The very nature of SpotLight avoids the pitfalls suffered by other approaches to intelligent diagnostics tools that rely on deep functional models, or on encoded troubleshooting processes. For a SpotLight knowledgebase, you need only to describe failures, the symptoms that would differentiate them from other failures in the knowledgebase, and some simple cost and time data associated with tests and repairs - SpotLight itself generates the troubleshooting steps during use. That means the knowledgebase can easily handle cross-system effects, operator errors, maintenance errors, mis-configurations, and anything that would be reported as a problem. And SpotLight's tools keep that whole process on track over the lifetime of the knowledgebase.