Scholarlink

The rapid surge of the electronic commerce (Ecomm) industry has ushered in intensified competition, with platforms vying for customer attention and loyalty. Many Ecomm sites have been employing various tactics to stand out, with Recommendation Systems (RS at the forefront of their strategies. However, traditional RS faces challenges, mainly when interactions occur at irregular intervals. This article introduces the DTW + LSTM framework, a methodology combining the merits of Long Short-Term Memory Networks and Dynamic Time Warping to improve this work. This method requires consideration of DTW's time series data alignment capabilities to account for interaction interval variability. The LSTM is then designed to accurately record the key elements of user interactions in sequential order. Combining the outcomes of these two methodologies in the DTW + LSTM model presents a more flexible and extensive approach to generating online shopping recommendations. The design guarantees that forecasts continue to be reliable and accurate despite the inherent temporal deviations that can be observed in user data. This paper investigates the context of this approach, its benefits, and prospective developments, demonstrating its possibilities in the dynamic Ecomm market.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient Temporal Data Mining Technique Using Dynamic Time Warped LSTM for E-Commerce Recommendation Systems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Dinesh Rajassekharan,
  • Jayasundar Subramanian,
  • Arokia Jesu Prabhu Lazar,
  • Srinivas Pichuka Veera Venkata Satya,
  • Samrat Ray,
  • Viswanathan Ammasai,
  • Sudhakar Sengan

摘要

The rapid surge of the electronic commerce (Ecomm) industry has ushered in intensified competition, with platforms vying for customer attention and loyalty. Many Ecomm sites have been employing various tactics to stand out, with Recommendation Systems (RS at the forefront of their strategies. However, traditional RS faces challenges, mainly when interactions occur at irregular intervals. This article introduces the DTW + LSTM framework, a methodology combining the merits of Long Short-Term Memory Networks and Dynamic Time Warping to improve this work. This method requires consideration of DTW's time series data alignment capabilities to account for interaction interval variability. The LSTM is then designed to accurately record the key elements of user interactions in sequential order. Combining the outcomes of these two methodologies in the DTW + LSTM model presents a more flexible and extensive approach to generating online shopping recommendations. The design guarantees that forecasts continue to be reliable and accurate despite the inherent temporal deviations that can be observed in user data. This paper investigates the context of this approach, its benefits, and prospective developments, demonstrating its possibilities in the dynamic Ecomm market.