Scholarlink

In this paper, we give out an approach towards utilisation of Large Language Models (LLMs) for extracting key and relevant information from exercise logs generated by people. We demonstrate the efficacy of six popular LLMs in handling exercise logs while classifying multiple dimensions of exercise type, intensity, duration, and feeling post-exercise. We compare the performance of LLMs with traditional models such as BERT classifiers and XLNet for few-shot classification tasks. We go on to suggest that we can use the power of LLMs to analyze exercise log data without the need for training. This can then be used for personalized recommendations related to fitness and also for marketing analysis while measuring the effectiveness of a coach/application. Our conceptual framework proposes LLM based classification and analysis resulting in an intelligent recommendation system.

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

An Approach for Extraction of Relevant Contextual Information from Exercise Logs Using Large Language Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Anand Arumilli,
  • Amit Oberoi,
  • Ayesha Rifa

摘要

In this paper, we give out an approach towards utilisation of Large Language Models (LLMs) for extracting key and relevant information from exercise logs generated by people. We demonstrate the efficacy of six popular LLMs in handling exercise logs while classifying multiple dimensions of exercise type, intensity, duration, and feeling post-exercise. We compare the performance of LLMs with traditional models such as BERT classifiers and XLNet for few-shot classification tasks. We go on to suggest that we can use the power of LLMs to analyze exercise log data without the need for training. This can then be used for personalized recommendations related to fitness and also for marketing analysis while measuring the effectiveness of a coach/application. Our conceptual framework proposes LLM based classification and analysis resulting in an intelligent recommendation system.