The internet of things (IoT) is one of the most rapidly growing technologies. Therefore, the interest in industry and academia has been increasing. The published research data have evolved in IoT because of scientific advances in this field. Since science plays a vital role in decision-making, this study examines the thematic landscape of research on IoT, which may contribute to understanding the research field’s structure allows for critical reflections and the identification of blind spots for advancing this field. The current study applies a text mining approach on 25966 Scopus-indexed abstracts and titles published from 2008 to 2020 on a latent Dirichlet allocation-based topic model. In this study, various models in the range of 1–100 topics were created. Examination of coherence scores was combined with manual analysis; the 25-topic model was chosen as an optimal one. The statistical methods employed highlight the timely trends of the extracted topics, intellectual topic structure and resulting communities in the topic network. The study carpingly depicts the quantitative results from an IoT perspective. The statistical analysis depicts that IoT publications has exponential growth rate. The hotspot of the IoT research can be concluded as ‘intrusion attack detection’, ‘cloud and edge computing’, ‘energy consumption’, ‘access channels’, ‘algorithm optimization’ and ‘healthcare and medical’. The topics that reflect the wireless sensor networks, security and privacy, high-range signal, devices and context aware computing and sensor control and monitoring have stable trends. This study identifies research focus on the development of low-energy consumption systems (Green IoT), application of high-range signals and their performance in tracking and identification, and data analytics (Big data IoT). Furthermore, the research focuses on industrial solutions towards diseases diagnosis and its treatment in health sector. Finally, in agriculture sector for intelligent manufacturing, research focuses on the application of image recognition for plant and food analysis.

中文翻译：

---

使用主题建模分析物联网的研究趋势

物联网 (IoT) 是发展最快的技术之一。因此，工业界和学术界的兴趣一直在增加。由于该领域的科学进步，已发布的研究数据在物联网中得到了发展。由于科学在决策中发挥着至关重要的作用，因此本研究考察了物联网研究的主题景观，这可能有助于理解研究领域的结构，允许进行批判性反思和识别推动该领域发展的盲点。当前的研究在基于潜在狄利克雷分配的主题模型上对 2008 年至 2020 年发表的 25966 篇 Scopus 索引摘要和标题应用了文本挖掘方法。在这项研究中，创建了 1-100 个主题范围内的各种模型。一致性分数的检查与人工分析相结合；选择了 25 个主题的模型作为最佳模型。所采用的统计方法突出了主题网络中提取的主题、智能主题结构和由此产生的社区的及时趋势。该研究从物联网的角度描绘了定量结果。统计分析表明，物联网出版物呈指数增长。物联网研究的热点可以归结为“入侵攻击检测”、“云和边缘计算”、“能耗”、“接入渠道”、“算法优化”和“医疗保健”。反映无线传感器网络、安全和隐私、远距离信号、设备和上下文感知计算以及传感器控制和监测的主题有稳定的趋势。本研究确定了低能耗系统（绿色物联网）的开发、远距离信号的应用及其在跟踪和识别中的性能以及数据分析（大数据物联网）的研究重点。此外，该研究侧重于卫生部门疾病诊断和治疗的工业解决方案。最后，在智能制造的农业领域，研究重点是图像识别在植物和食品分析中的应用。