| 庞庆华,姚玉康,张丽娜.基于 LDA-ARIMA 的我国智能手机关键技术主题识别与演化分析[J].情报工程,2024,10(3):049-062 |
| 基于 LDA-ARIMA 的我国智能手机关键技术主题识别与演化分析 |
| Identification and Evolution Analysis of Key Technologies Topics of Smartphones in China Based on LDA-ARIMA |
| |
| DOI：10.3772/j.issn.2095-915X.2024.03.004 |
| 中文关键词: 关键技术；主题识别；LDA 模型；主题演化；ARIMA 模型；智能手机 |
| 英文关键词: Key Technology; Topic Recognition; LDA Model; Topic Evolution; ARIMA Model; Smart Phone |
| 基金项目:河海大学2022年度“优秀专业硕士学位论文选拔培育”（422003524）。 |
| 作者 | 单位 | | 庞庆华 | 河海大学商学院 常州 213022 | | 姚玉康 | 河海大学商学院 常州 213022 | | 张丽娜 | 河海大学商学院 常州 213022 |
|
| 摘要点击次数: 353 |
| 全文下载次数: 620 |
| 中文摘要: |
| [目的/意义］专利是技术能力的表现形式，包含关键技术主题信息。以智能手机专利数据为基础，提出关键技术主题的识别和演化分析方法，帮助企业获取行业内的技术信息，调整专利研究的成本与精力投入。［方法/过程］首先，选择专利数据库和高级检索内容，下载和导出专利标题和摘要数据，并对数据进行去停用词和jieba分词等处理；其次，构建困惑度求解模型，确定最优主题数，再将已经分好词的文本导入LDA模型进行主题挖掘，得到每个技术主题下关键词语的分布；再次，将主题热度转化为时间序列，进行平稳性检测和白噪声检验，确定ARIMA模型参数后应用预测；最后，依据每个技术主题下提取的特征词确定关键技术主题并进行解读，通过时间序列预测结果对关键技术主题进行演化分析。［结果/结论］以智能手机为研究对象，成功识别出屏幕、电池与充电、生物识别系统等15个关键技术主题，挖掘出各主题不同的演化发展特征，根据演化趋势分析提出建议，验证了本文主题识别与演化分析方法的可行性与有效性。 |
| 英文摘要: |
| [Objective/Significance] Patent is a manifestation of technological capability and contains information on key technological topics. Based on smartphone patent data, the identification and evolution analysis method of key technological topics is proposed to help enterprises obtain technological information in the industry and adjust the cost and effort invested in patent research. [Methods/Processes] Firstly, the patent database and advanced search content are selected, and the patent title and abstract data are downloaded and exported. The data is then processed, including deactivating words and jieba-splitting.Subsequently, a perplexity solution model is built to determine the optimal number of topics. The preprocessed text is then imported into the LDA model for topic mining, resulting in the distribution of key words under each technical topic. The topic’s heat degree is transformed into a time series, followed by smoothness and white noise tests. The parameters of the ARIMA model are determined and applied for prediction. Finally, the feature words extracted under each technical topic are used to identify and interpret the key technical topics. The evolution of these topics is analyzed through the results of the time series prediction. [Results/Conclusions] By focusing on smartphones as the research subject, this study successfully identifies 13 key technological topics such as screen, battery and charging, and biometric system. The different developmental characteristics of each theme are analyzed. Moreover, suggestions are provided based on the analysis of evolutionary trends, thus confirming the feasibility and effectiveness of the theme identification and evolutionary analysis method proposed in this paper. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
