Scholarlink

<p>With the exponential growth of social media platforms, user-generated content has emerged as a valuable resource for enhancing recommendation systems. However, the pervasive noise in user-centric tweets, stemming from irrelevant, ambiguous, or misleading information, continues to degrade the performance and reliability of such systems. To address this challenge, this paper proposes nuclear physics optimization algorithm (NPO) as the core contribution for effective noise filtering in tweet-based recommendation systems. The methodology begins with preprocessing the raw tweet data and extracting features that accurately reflect user preferences and interests. Central to our approach, the NPO algorithm is designed to dynamically adjust its parameters to identify and eliminate noisy tweets based on semantic similarity and domain relevance. By adaptively optimizing the filtering process, the NPO effectively tailors itself to the characteristics of the tweet dataset, leading to enhanced accuracy and reliability in movie recommendation outcomes. Experimental evaluations conducted on a real-world tweet dataset associated with the MovieLens dataset validate the superiority of the proposed approach. Notably, when integrated with a recurrent neural network (RNN) classifier, the system achieved impressive performance metrics, including an accuracy of 0.94, precision of 0.93, recall of 0.91, F1-score of 0.92, and a mean average precision (MAP) of 0.92. Comparative analyses against conventional baseline models highlight significant improvements in prediction accuracy and recommendation quality.</p>

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

Optimization-based noise filtering among user-centric tweets to improve predictions in recommendation system

  • 期刊论文

  • 发表日期：

    2025年6月10日

查看全文
  • Kirti Jain,
  • Rajni Jindal

摘要

With the exponential growth of social media platforms, user-generated content has emerged as a valuable resource for enhancing recommendation systems. However, the pervasive noise in user-centric tweets, stemming from irrelevant, ambiguous, or misleading information, continues to degrade the performance and reliability of such systems. To address this challenge, this paper proposes nuclear physics optimization algorithm (NPO) as the core contribution for effective noise filtering in tweet-based recommendation systems. The methodology begins with preprocessing the raw tweet data and extracting features that accurately reflect user preferences and interests. Central to our approach, the NPO algorithm is designed to dynamically adjust its parameters to identify and eliminate noisy tweets based on semantic similarity and domain relevance. By adaptively optimizing the filtering process, the NPO effectively tailors itself to the characteristics of the tweet dataset, leading to enhanced accuracy and reliability in movie recommendation outcomes. Experimental evaluations conducted on a real-world tweet dataset associated with the MovieLens dataset validate the superiority of the proposed approach. Notably, when integrated with a recurrent neural network (RNN) classifier, the system achieved impressive performance metrics, including an accuracy of 0.94, precision of 0.93, recall of 0.91, F1-score of 0.92, and a mean average precision (MAP) of 0.92. Comparative analyses against conventional baseline models highlight significant improvements in prediction accuracy and recommendation quality.