Mitsubishi Fuso combines machine learning and advanced natural language processing (NLP) techniques to improve product experience

•Mitsubishi Fuso is integrating cognitive search solutions for higher efficiency and accuracy in quality management processes.
•New technology has improved the FUSO product experience by reducing feedback time on product issues reported by customers by 30%.

Kawasaki, Japan – Mitsubishi Fuso Truck and Bus Corporation (MFTBC), one of Asia’s leading commercial vehicle manufacturers, is pleased to announce improvements in quality management processes through the application of cognitive search solutions combining natural language processing techniques and machine learning. This initiative comes as a further push within MFTBC to support customer uptime by leveraging big data.

Cognitive search allows for the indexing of large bodies of information and helps clarify interpretations of data through clustering and comparison. MFTBC has taken advantage of this technology to better respond to quality reports from customers. Before the upgrade, all reports received through dealers had been manually read, analyzed and matched to other known quality issues to identify trends, as well as ascertain the scale and severity of reported cases. While this step relied heavily on the expertise of each individual overseeing the task, the process now integrates a cognitive search based analysis of texts from quality reports and searches for potentially related issues in a cloud-based library of past and running scenarios to aid the responsible employees. Natural language processing capabilities, which improve the “comprehension” of data through linguistic analytics, were added to boost search accuracy. As a further step to boost the precision of analyses, machine learning was also integrated to continuously refine the relevancy of suggested information.

To construct the system, MFTBC quality management team started by creating a detailed library of past cases to extract important information out of the quality reports. The system now continuously builds upon this library by analyzing quality reports submitted through an online portal by dealers in English or Japanese. The process is currently compatible with all KD and BU FUSO vehicles produced in Japan and Tramagal (Portugal), including those distributed in international markets.

Since the implementation of these technologies started January of this year, the lead time for reports processing has been reduced by 30%. This means that the feedback to customers can be completed more quickly compared to last year, minimizing the potential for extended downtime scenarios. As a customer-interfacing initiative, the new system complements the real-time vehicle monitoring and telediagnosis features supplied by the Truckonnect telematics platform. The use of cognitive search in quality management at MFTBC is also part of a larger digitalization movement within operations at the Kawasaki headquarters. Digitalization has been a major pillar of process improvement in a wide variety of work areas, ranging from the production line to supplier relations.