文献类型：期刊文献

中文题名：一种基于LDA模型的新兴主题识别与探测方法

英文题名：An emerging topic identification and detection method based on LDA model

作者：吴东雪[1];沈桂兰[2]

第一作者：吴东雪

机构：[1]北京联合大学应用文理学院,北京100191;[2]北京联合大学商务学院,北京100191

第一机构：北京联合大学应用文理学院

年份：2024

卷号：52

期号：2

起止页码：72-80

中文期刊名：河南师范大学学报（自然科学版）

外文期刊名：Journal of Henan Normal University(Natural Science Edition)

收录：CSTPCD;;北大核心:【北大核心2023】；

基金：国家社科基金(21BXW057);北京联合大学人才强校优选项目(BPHR2019CZ03)。

语种：中文

中文关键词：主题识别;最优主题数;新兴主题识别指标;Prophet模型

外文关键词：topic identification;optimal topiccount;emerging topic identification indicators;Prophet model

摘要：新兴主题识别是科技研究领域识别新兴技术的重要方式,高效精准地识别新兴主题是早期辨识新兴技术研究方向的前提.提出一种基于LDA模型的新兴主题识别与趋势预测方法,通过LDA模型提取科技文献中的研究主题,构建主题强度、主题新颖度和复合主题关注度的指标体系识别新兴主题,采用Prophet模型预测新兴主题的主题强度,探测未来发展趋势.以智慧农业领域最近14年的科研文献为数据集,对提出的识别和探测方法进行验证,识别出了5个新兴主题,并预测了未来3年的发展趋势,同时验证所提方法的有效性.
Emerging topic identification is an important way to identify emerging technologies in the field of science and technology research,and efficient and accurate identification of emerging topics is the premise of early identificating emerging technology research direction.An emerging topic identification and trend prediction method based on LDA model is proposed.It extracts research topics from scientific literature by LDA model,constructs an index system of topic strength,topic novelty and composite topic attention to identify emerging topics,and uses Prophet model training to predict topic strength of emerging topics.Based on the data set of scientific research literature in the field of smart agriculture in the last 14 years,the proposed recognition and detection methods are verified.Five emerging topics are identified,and the development trend in the following three years is predicted.The validity of the proposed methods is verified.