Related Experiment Video
Updated: Dec 24, 2025

Applying an eMASS Customization Program as a Research Tool to Evaluate Consumer Benefits
Published on: September 27, 2019
Feature-based recommendations for one-to-one marketing
1Department of Information Management, Fu-Jen Catholic University, HsinChuang City, Taipei 242, Taiwan, ROC.
This study introduces a new recommendation system that analyzes customer purchasing behavior and product features. It effectively recommends new or rarely purchased items by understanding customer interest profiles, overcoming limitations of traditional methods.
Area of Science:
- Information Science
- Computer Science
- Marketing Science
Background:
- Traditional recommendation systems struggle with time-sensitive products (popular, seasonal) and new or rarely purchased items due to limited purchase/rating data.
- Market basket analysis and collaborative filtering methods fail to recommend products lacking sufficient customer interaction history.
- Low ratings for infrequently bought items (e.g., furniture, appliances) reduce their visibility in existing recommendation engines.
Purpose of the Study:
- To develop an improved recommendation system capable of suggesting new and infrequently purchased products.
- To analyze customer purchasing behaviors by integrating transaction records with product feature databases.
- To create customer interest profiles based on preferences for specific product features.
Main Methods:
- Customer purchasing behaviors are analyzed using transaction records and product feature databases.
- Customer preferences for product features are identified to construct customer interest profiles.
- A two-stage clustering technique is employed to identify customer segments with similar interests.
Main Results:
- The proposed system successfully recommends new and rarely purchased products by matching them to customer interest profiles.
- Customer interest profiles provide explainability for recommendation results.
- Analysis of feature preferences offers insights for product development and targeted marketing strategies.
Conclusions:
- This research presents a novel approach to recommendation systems that overcomes the cold-start problem for new and unpopular products.
- The method enhances recommendation accuracy by focusing on product features and customer preferences.
- Insights gained can inform product development and enable more profitable one-to-one marketing strategies.
Related Concept Videos
Persuasion Strategies
Factorial Design
Routes of Persuasion
Introduction to One-to-one Functions
Impression Management Techniques IV: Altercasting
Relationship Formation

