Understanding and modeling user behavior for recommendation systems

E-commerce has revolutionized consumer-business interactions through global accessibility and tailored services. However, existing session-based, sentiment-based, and location-based recommendation systems encounter challenges, such as accuracy issues in Machine Learning (ML) and Deep Learning (DL) c...

Full description

Saved in:
Bibliographic Details
Main Authors: Bhogan, Snehal, Rajpurohit, Vijay S., Sannakki, Sanjeev S.
Format: Conference Proceeding
Language:English
Subjects:
Data mining
Deep learning
Electronic commerce
Machine learning
Recommender systems
Sentiment analysis
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:E-commerce has revolutionized consumer-business interactions through global accessibility and tailored services. However, existing session-based, sentiment-based, and location-based recommendation systems encounter challenges, such as accuracy issues in Machine Learning (ML) and Deep Learning (DL) classification, leading to potential inaccuracies and reduced adaptability. Another framework, integrating location-centric and time-centric transitions, faces limitations in complexity, customization, and scalability. In this work, we address these challenges by presenting a survey on these systems and two existing frameworks. To tackle these issues, we propose an innovative architecture encompassing five phases for an effective e-commerce recommendation system. This proposed framework aims to resolve the drawbacks of existing systems by focusing on data preprocessing, feature extraction, and personalized recommendations based on session, sentiment analysis, location, and user-item relationships. The results show that the location and time-centric methods provided an increase in terms of Hit Rate and Mean Reciprocal Rate.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0224821