| 基于主题模型的科技文献主题演化及优化方法研究综述 |
| 修订日期：2023-03-25 点此下载全文 |
| 引用本文：于诗睿,李爱花,林紫洛,等.基于主题模型的科技文献主题演化及优化方法研究综述[J].医学信息学杂志,2023,44(8):31-36 |
| 摘要点击次数: |
| 全文下载次数: |
|
| 基金项目:中国医学科学院医学与健康科技创新工程重大协同创新项目(项目编号:2021-I2M-1-033)。 |
|
| 中文摘要:目的/意义 梳理主题演化分析方法研究进展,改善科技文献主题识别和趋势研判效果,支撑文献信息服务,为未来深入研究和实践应用指明方向。方法/过程 根据隐含狄利克雷分布主题模型的缺陷归纳其衍生模型,针对现有科技文献主题演化分析方法存在的不足对相应改善方案进行总结,最后结合现有研究局限性提出展望。结果/结论 通过全面调研,为主题演化研究提供启示和借鉴。 |
| 中文关键词:主题模型 主题识别 主题演化 隐含狄利克雷分布主题模型 文本挖掘 |
| |
| A Review of Research on the Improvement of Topic Model Based Topic Evolution Analysis Methods for Scientific Literature |
|
|
| Abstract:Purpose/Significance The research progress of topic evolution analysis method is sorted out, so as to improve the effect of topic identification and trend judgment of scientific literature, support literature information service, and point out the direction for future in-depth research and practical application. Method/Process The derivative model of latent Dirichlet allocation (LDA) is summed up according to the defects of the topic model, and the corresponding improvement plan is summarized according to the shortcomings of the analysis methods of the topic evolution of the existing scientific literature. Finally, the prospect is put forward according to the limitations of the existing research. Result/Conclusion Through comprehensive investigation, the paper can provide inspiration and references for the study of topic evolution. |
| keywords:topic model topic identification topic evolution latent Dirichlet allocation (LDA) text mining |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
