Hybrid case-based reasoning for on-line product fault diagnosis

Abstract

This paper presents a hybrid case-based reasoning system for on-line technical support of PC fault diagnosis. HyCase consists of a natural language (keyword) input and the graph-theoretic constraint-net.Natural language or keyword inputs are parsed and then generated into a constraint-net. The constraint-net is validated and its links rationalized and standardized to minimize ambiguity. Cases that partially match either the keywords or the constraint-net are ranked according to their matching scores based on four different preferences, and then retrieved and presented to the user. The case base, which was developed in Microsoft Access 2000, is updated by means of a keyword and constraint-net manager. HyCase was implemented on a PC with a Pentium III processor running at 500 MHz and 128 MB of SDRAM.Twenty typical queries from customers were tested on a collection of 174 cases based on different cut-off overall matching scores. The effectiveness of case retrieval was measured by the proportion of relevant cases retrieved from the case base (recall) and, of these, the proportion directly applicable to the problem at hand (precision). The accuracy of the natural language parser was ascertained to range between 62.5% and 87.7%, while a parsing accuracy of 60% is sufficient to ensure a reasonable recall and precision. A minimum overall matching score of about 0.5 ensures a precision of 0.60. The parsing time gets noticeably longer when there are more than 15 words. Except for minimum overall matching scores exceeding 0.1, the parsing time is largely independent of the minimum score.The merits and limitations of HyCase are also discussed.

Text Case-Based Reasoning Framework for Fault Diagnosis and Predication by Cloud Computing

“…Hence, the fault diagnosis and predication problem can be quickly computed by a CaseBased Reasoning (CBR) module, which tries to infer from historical information the hidden relationship between system states and the corresponding historical solutions [8,9]. CBR is an effective technique for problem solving in the fields in which it is hard to establish a quantitative mathematical model, such as fault diagnosis, health management, or industrial systems [10]. Following this idea, He [11] proposes a framework to use text mining and Web 2.0 technologies to improve and enhance CBR systems for providing better user experience.…”

Fault diagnosis method for complex equipment using cased-based reasoning

“…Case-based reasoning (CBR),a popular problem solving methodology in data mining field, solves new problems by analyzing the solutions for similar past problems [14][15][16]. The many advantages of CBR include rapid self-learning, effective presentation of knowledge, and the ability to use unrestricted domains, et al And thus it has been widely used to diagnosis field [17][18][19][20][21][22][23][24].…”

Analyzing Massive Machine Maintenance Data in a Computing Cloud

“…The processes involved in CBR are 1. retrieving similar cases from case base, 2. reusing the information in the retrieved cases, 3. revising the solution, and 4. retaining a new experience into the case base [6]. CBR is an effective technique for problem solving in the fields in which it is hard to establish a quantitative mathematical model, such as fault diagnosis, health management, or industrial systems [7]. CBR systems utilize incremental learning in which new cases are added into the case base with time.…”