Dnyanesh Rajpathak

A generic ontology development framework for data integration and decision support in a distributed environment

Data inconsistency and data mismatch are critical problems that limit data interoperability and hinder smooth operation of a distributed business. An ontology represents a semantic model that explicitly describes various entities and their properties of a domain of discourse and acts as a vehicle for seamless data integration and exchange. The existing methodologies for ontology development fail to provide a comprehensive coverage for different steps, e.g. pre-development, development and post-development, which are necessary for successfully developing ontologies. We propose a generic and comprehensive methodology that puts ontology engineering on a firm scientific foundation and at the same time provides a collaborative environment for effective knowledge sharing and reuse. Furthermore, our approach also provides a way for automatically extracting frequent terms from the data to construct an ontology in a bottom-up fashion. The performance of our methodology has been evaluated by developing different ontologies to solve the real life applications, e.g. fault diagnosis and root cause investigation and spare parts maintenance.

An ontology based text mining system for knowledge discovery from the diagnosis data in the automotive domain

Abstract In automotive domain, overwhelming volume of textual data is recorded in the form of repair verbatim collected during the fault diagnosis (FD) process. Here, the aim of knowledge discovery using text mining (KDT) task is to discover the best-practice repair knowledge from millions of repair verbatim enabling accurate FD. However, the complexity of KDT problem is largely due to the fact that a significant amount of relevant knowledge is buried in noisy and unstructured verbatim. In this paper, we propose a novel ontology-based text mining system, which uses the diagnosis ontology for annotating key terms recorded in the repair verbatim. The annotated terms are extracted in different tuples, which are used to identify the field anomalies. The extracted tuples are further used by the frequently co-occurring clustering algorithm to cluster the repair verbatim data such that the best-practice repair actions used to fix commonly observed symptoms associated with the faulty parts can be discovered. The performance of our system has been validated by using the real world data and it has been successfully implemented in a web based distributed architecture in real life industry.

An integrated framework for effective service and repair in the automotive domain: An application of association mining and case-based-reasoning

In this paper, we propose a novel integrated framework combining association rule mining, case-based-reasoning and text mining that can be used to continuously improve service and repair in an automotive domain. The developed framework enables identification of anomalies in the field that cause customer dissatisfaction and performs root cause investigation of the anomalies. It also facilitates identification of the best practices in the field and learning from these best practices to achieve lean and effective service. Association rule mining is used for the anomaly detection and the root cause investigation, while case-based-reasoning in conjunction with text mining is used to learn from the best practices. The integrated system is implemented in a web based distributed architecture and has been tested on real life data.

A domain-specific decision support system for knowledge discovery using association and text mining

We propose a novel association and text mining system for knowledge discovery (ASTEK) from the warranty and service data in the automotive domain. The complex architecture of modern vehicles makes fault diagnosis and isolation a non-trivial task. The association mining isolates anomaly cases from the millions of service and claims records. ASTEK has shown 86% accuracy in correctly identifying the anomaly cases. The text mining subscribes to the diagnosis and prognosis (D&P) ontology, which provides the necessary domain-specific knowledge. The root causes associated with the anomaly cases are identified by discovering frequent symptoms associated with the part failures along with the repair actions used to fix the part failures. The best-practice knowledge is disseminated to the dealers involved in the anomaly cases. ASTEK has been implemented as a prototype in the service and quality department of GM and its performance has been validated in the real life set up. On an average, the analysis time is reduced from few weeks to few minutes, which in real life industry are significant improvements.

An Ontology-Based Text Mining Method to Develop D-Matrix From Unstructured Text

Fault dependency (D)-matrix is a systematic diagnostic model [7] to capture the hierarchical system-level fault diagnostic information consisting of dependencies between observable symptoms and failure modes associated with a system. Constructing a D-matrix from first principles and updating it using the domain knowledge is a labor intensive and time consuming task. Further, in-time augmentation of D-matrix through the discovery of new symptoms and failure modes observed for the first time is a challenging task. Here, we describe an ontology-based text mining method for automatically constructing and updating a D-matrix by mining hundreds of thousands of repair verbatim (typically written in unstructured text) collected during the diagnosis episodes. In our approach, we first construct the fault diagnosis ontology consisting of concepts and relationships commonly observed in the fault diagnosis domain. Next, we employ the text mining algorithms that make use of this ontology to identify the necessary artifacts, such as parts, symptoms, failure modes, and their dependencies from the unstructured repair verbatim text. The proposed method is implemented as a prototype tool and validated by using real-life data collected from the automobile domain.

Ontology-driven data collection and validation framework for the diagnosis of vehicle healthmanagement

The current warranty data collection processes exhibit several data quality issues – the level of detail and precision is missing in the collected data, the semantic heterogeneity is observed and no systematic data quality validation mechanism to automatically certify the data quality. Such data cannot be translated seamlessly into the knowledge assets to perform business functions, for example, fault diagnosis. An ontology-driven structured data collection framework is proposed to acquire the necessary data in the warranty domain. The proposed framework uses the integrated vehicle health management ontology as an information model to populate necessary data acquisition fields of the framework. A novel three-dimensional data quality metric is proposed to validate the completeness, correctness and relevance of newly collected data. We also evaluate the performance of the tool by using the real-life data. The data accuracy precision after using the framework has been improved from 0.30 to 0.80, whereas the recall is improved from 0.28 to 0.70. Furthermore, the precision and recall of the tool is evaluated for the 500 real-life field failure cases and it was greater than 90% for data completeness and relevance. Throughout this paper we will use the words ‘correctness’ and ‘accuracy’ interchangeably.