WIPE: A Novel Web-Based Intelligent Packaging Evaluation via Machine Learning and Association Mining

Abstract

This paper introduces the Web-Based Intelligent Packaging Evaluation (WIPE) platform, a novel approach to assess the performance of product and packaging systems within the e-commerce distribution sector. Unlike traditional methods that primarily rely on laboratory evaluations under controlled conditions, WIPE addresses the unique challenges posed by e-commerce distribution, such as increased handling points and unforeseen hazards that standard physical tests may not capture. Leveraging advanced machine learning algorithms and association rule mining, WIPE extracts insights about packaging defects directly from customer reviews on e-commerce platforms. Analyzing both images and text from these reviews establishes connections between frequently used words and the predicted damages, causes, and effects. This innovative approach was exemplified in two case studies involving laundry detergent liquid bottles and pods sold on the Amazon. The findings from these studies demonstrate WIPE’s capability to extract pertinent information from customer feedback and identify specific packaging defects and predict their potential causes. This integration of sentiment analysis and association rule mining into the packaging evaluation process marks a significant advancement in the field. The introduction of WIPE represents a transformative step in packaging evaluation, offering a more dynamic, real-world analysis that can significantly enhance product and packaging design, ultimately leading to improved customer satisfaction in the rapidly evolving e-commerce landscape.

Keywords

• Packaging evaluation
• E-commerce distribution
• machine learning algorithms
• association rule mining
• sentiment analysis
• damage prediction