Can social media postings by consumers be a source of useful information about vehicle safety and performance defects for automobile manufacturers?
Yes, say Pamplin College of Business researchers who conducted what is believed to be the first large-scale case study confirming the value of social media for vehicle quality management.
“A lot of useful but hidden data on vehicle quality is embedded in social media that is largely untapped by auto manufacturers,” says Alan Abrahams, assistant professor of business information technology, who led the study together with Weiguo Fan, professor of accounting and information systems.
Abrahams says consumers rely heavily on the Internet for information about automobile safety and reliability, looking up vehicle consumer surveys, insurance industry statistics, manufacturer websites, and complaints filed with regulatory agencies. But in addition to being consumers of safety and reliability information, he says, automobile users are also producers of such information, using traditional Internet media (such as emails or online forms) and, increasingly, social media tools (such as bulletin boards, blogs, and Twitter).
Mining for valuable social content
Whether in public discussion forums, social networks, product reviews, visitor comments, wikis, or user-written news articles, user-contributed content is characterized by variable quality, says Fan.
Processing the “unstructured and dynamic” content of social media, in particular, is a daunting challenge for firms. And, in contrast to the numerous studies on the use of web mining for business intelligence, it is also a research question that has received little attention from scholars: how to detect the useful nuggets on vehicle defects that are buried among millions of unrelated or immaterial postings? “Our study sought to understand and prioritize the vast volume of consumer-produced automotive information and to ferret out and analyze the safety and performance issues," Fan says.
In their case study, the researchers focused on online discussion forums for owners of Honda, Toyota, and Chevrolet vehicles — the brands were chosen for the high use of their forums by consumers in the past decade. They employed three independent automotive experts to manually analyze and categorize thousands of posts from these forums, tagging them by defect existence and criticality (or significance for safety) and the component involved.
A new decision support system
Discovering shortcomings in the research technique typically used to analyze social media content, the researchers created an alternative set of marker words for the auto industry, which they called “automotive smoke words.” Using these linguistic markers, Abrahams, Fan, and their team designed a text analysis framework to distill the safety and reliability information into a digestible format for automakers. The researchers then implemented this framework in a new decision support system (a computer-based information system that helps managers make decisions) that they built and tested.
The result, Abrahams says, is a “robust” system for discovering vehicle defects from social media posts across multiple automotive brands.
“Vehicle quality management professionals would greatly benefit in terms of productivity by employing a vehicle defect discovery system like ours to sift defects from unrelated postings,” he says.
Their research shows that the existence of safety and performance defects is strongly predicted by the incidence of automotive smoke words, he says. Therefore, auto manufacturers can scan social media forums for vehicle defects by using an automotive smoke word list along with web crawlers and other tools.
Looking ahead, Abrahams says the researchers plan to extend their vehicle defect discovery system to determine whether new text mining techniques can be developed to enhance the defect detection and sorting process.
They would like to expand their analysis of postings to study Twitter and Facebook posts, additional linguistic features of text, and other vehicle brands. “We’ve just touched the tip of the iceberg in Social CRM,” says Fan, referring to customer relationship management through the use of social media technology — a growing practice among businesses as well as an emerging field of study. “Our method is very generic and can be extended easily to other areas. We look forward to working with other consumer brands and products for total quality management.”
Says Abrahams: “With the volume of social media posts expanding rapidly, we expect that the need for automated business intelligence tools for the exploration of this vast and valuable data set will continue to grow.”
Abrahams and Fan co-authored the study with G. Alan Wang, assistant professor of business information technology, and Jian Jiao, a Virginia Tech doctoral student in computer science and a software design engineer at Microsoft. Their study, “Vehicle Defect Discovery from Social Media,” will be published in Decision Support Systems journal.