Scholarlink

Movie recommendation systems often rely on a vast array of tags to categorize films, but the high dimensionality of these tags can complicate and reduce the effectiveness of the recommendation process. This paper explores an approach to enhance recommendation accuracy by reducing the dimensionality of the tags in the MovieLens dataset. We utilize the embeddings obtained from a sentence-transformer model to identify and merge semantically related tags using clustering approaches, thereby reducing redundancy in the tag space, thereby reducing the number of dimensions. This approach not only mitigates the challenges associated with high-dimensional data but also improves the quality and relevance of movie recommendations.

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

Tag Dimensionality Reduction for Enhanced Movie Recommendations in the Long Tail Using Semantic Clustering

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Soanpet Sree Lakshmi,
  • T. Adilakshmi,
  • Bakshi Abhinith

摘要

Movie recommendation systems often rely on a vast array of tags to categorize films, but the high dimensionality of these tags can complicate and reduce the effectiveness of the recommendation process. This paper explores an approach to enhance recommendation accuracy by reducing the dimensionality of the tags in the MovieLens dataset. We utilize the embeddings obtained from a sentence-transformer model to identify and merge semantically related tags using clustering approaches, thereby reducing redundancy in the tag space, thereby reducing the number of dimensions. This approach not only mitigates the challenges associated with high-dimensional data but also improves the quality and relevance of movie recommendations.