A problem solving approach to developing product support systems
R.M. Setchi, N. Lagos
Manufacturing Engineering Centre, Cardiff University, Cardiff , UK
Introduction
Product support is everything needed to support the continued use of a product
Interactive Electronic Technical Manuals
Intelligent Product Manuals
Electronic Performance Support Systems
Each user query can be represented as a problem
A problem solving approach can be used throughout the development of a product support system
Background
Problem solving and reasoning techniques
Case-based reasoning: Reuse and modify parts of previous solutions, represent specific knowledge
Model-based reasoning: Reuse previous solutions, capture general knowledge
Rule-based reasoning: Follows an iterative cycle, series of progrssive inferences, represent specific knowledge
Reasoning in product support
Rule-based reasoning is the most common for troubleshooting
Case-based reasoning used in diagnosis and help-desk applications
Model-based reasoning less used technique, employed in adaptive interfaces
The inference mechanism needs to be studied in a structured manner and the advantages of different reasoning techniques can be exploited within the same system
Problem solving approach
Process starts with problem identification/recognition, deciding the course of actions that should be followed and reach a product support solution, which is delivered as a document
Documents are constructed from Information Objects and Information Object Clusters
Information Object: A data structure that represents an identifiable and meaningful instance of information in a specific presentation form
Information Object Cluster: A set of Information Objects, which share a common property and are arranged in a structure S
Document: A set of Information Object Clusters logically structured.
Problem solving approach-2
Problem characterisation: Two main factors are the domain and environment of the system
Domain: Products and tasks supported
Environment: User, location, time, purpose
Product Support Problem is a 4-tuple PSP := (MOD, HYP, CON, OBS)
MOD represents the domain knowledge
HYP is a finite set of combinations of MOD (hypotheses)
CON stands for the knowledge about the application environment
OBS is a finite set of observations from the query
A hypothesis can be a Product Support Solution only if when combined with MOD and CON can infer OBS or a subset of OBS
Problem solving approach-3
Knowledge reuse for previously solved problems
Case-based reasoning
Knowledge adaptation for similar problems
Case-based reasoning
Knowledge creation for new problems
Case-based reasoning and model-based reasoning
Problem solving approach-4
Implementation
Conclusions and future work
The approach presented segments the development phase of the system's inference mechanism into small manageable steps
According to the product support problems different reasoning strategies are utilised
Multi-modal reasoning enables the system to respond to a large variety of user queries
A co-relation between problem solving and product support has been identified
The problem solving strategy can be integrated with knowledge models to provide a product support knowledge engineering framework
