The increasing concern of climate change and unstable oil prices induce the development of technological fuel in automobile industry. To investigate such a rapidly changing path, researchers apply bibliometrics and topic modeling to patent data. These commonly used methods, however, have several drawbacks such as considering macro-level trend only and focusing on high probable terms. To avoid these weaknesses, we propose the two-layer trend analysis based on Time country topic model (TCT) and Dirichlet compound multinomial model (DCM) that enable to detect both macro-level and micro-level trend and identify bursty terms in automotive industry. Experimental results show rising, falling and fluctuating trend topics on condition of countries using TCT model. We also find path of automotive technology based on bursty terms from the analysis of DCM model. Specifically, electric vehicle, aluminum in lightweight material and diesel engine are considered as rising topics in the automobile fuel. Our proposed framework can be applied to analyze the trajectory analysis in various other fields.

中文翻译：

---

基于两层的汽车燃料行业研究趋势轨迹分析

对气候变化和油价不稳定的日益关注促使汽车行业技术燃料的发展。为了研究这种快速变化的路径，研究人员将文献计量学和主题建模应用于专利数据。然而，这些常用的方法有几个缺点，例如只考虑宏观层面的趋势和关注高概率项。为了避免这些弱点，我们提出了基于时间国家主题模型（TCT）和狄利克雷复合多项式模型（DCM）的两层趋势分析，能够检测汽车行业的宏观和微观趋势并识别突发术语. 实验结果显示了使用TCT模型的国家条件下的上升、下降和波动趋势主题。我们还从DCM模型的分析中找到了基于突发项的汽车技术路径。具体而言，电动汽车、轻质材料铝和柴油发动机被认为是汽车燃料中的新兴话题。我们提出的框架可用于分析其他各个领域的轨迹分析。