RELATING PREFERENCE TO BODY SIZE AND SHAPE THROUGH THE DATA MINING OF CONSUMER REVIEWS

Open Access
Author:
Chaklader, Rushtin
Graduate Program:
Mechanical Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
March 27, 2017
Committee Members:
  • Matt Parkinson, Thesis Advisor
  • Karen A. Thole, Committee Member
  • Conrad Tucker, Committee Member
Keywords:
  • Data Mining
  • Design for Human Variability
  • Consumer Reviews
Abstract:
In this thesis the merging of the fields of design for human variability and data mining consumer reviews are investigated to determine information on human preference for a product. Human preference unrelated to anthropometry is either difficult to obtain or completely ignored when designing a product. Determining human preference through data mining the review text of already released products may be a potentially less time consuming and costly method. Previously established methods of determining ergonomic information from consumer reviews are investigated, such as the Frequency and Accuracy Summation number and subsequent Manual Analysis. A new metric is also introduced, the Weighted Cue Phrase Product Rating, which can be an automated tool to quickly analyze consumer reviews, and compared to the Manual Analysis findings. These data mining techniques are then compared to human anthropometry measurements. The analysis is done on two different case studies. In the end it is determined that, though not an exact science, data mining consumer reviews can still be an avenue to gather ergonomic and consumer preference information about a product quickly and economically. It is also determined that the Weighted Cue Phrase Product Rating, though not without drawbacks, produces similar results to the Manual Analysis while being quicker and automated.