基于多维鲁棒数据质量分析模型的中文专利法律状态信息清洗研究

发布时间：2018-04-25 20:28

本文选题：专利法律状态 + 数据清洗　；参考：《南京理工大学》2017年硕士论文

【摘要】：专利法律状态信息具有时序特征,能够体现专利的生命周期和特定时间内权利归属等特征,在专利分析中具有独特价值。其清洗工作要求保证法律状态个体的正确性和状态之间时序关系的合理性。但是,现有数据清洗研究集中于对单个数据的清洗,很少涉及到时序关系数据的清洗。因此,构建并实现能够对时序关系进行清洗的专利法律状态信息清洗框架势在必行。多维鲁棒数据质量分析模型能够方便有效的完成对时序关系数据的清洗。该方法首先进行时序数据库建模,即抽象出时序数据中的状态集及状态转移关系,并形式化表示,然后通过模型检测的方式将时序数据清洗任务转换为模型检测任务,从而识别出时序数据中所隐藏的时序逻辑问题。该方法实现了数据问题的自动化发现,提高了时序数据清洗的效率,降低了清洗成本。为了完成法律状态个体以及状态之间时序关系的清洗工作,本文利用多维鲁棒数据质量分析模型对法律状态信息数据库建模,并对其进行模型检测,实现法律状态信息时序关系的清洗,并以此为核心构建并实现法律状态信息清洗框架。通过对实际数据的清洗验证了框架的科学性和有效性,发现了专利法律状态信息在时序关系方面的常见错误模式,并对这些错误进行了分析。本文将多维鲁棒数据质量分析模型应用到专利法律状态信息时序关系的清洗中,为时序数据清洗提供了新思路,同时本文构建的数据清洗框架,为同类型数据清洗工作提供了实证参考。
[Abstract]:Patent legal status information has the characteristics of time series, which can reflect the characteristics of the life cycle of the patent and the rights belonging to a certain time. It has the unique value in the patent analysis. The cleaning work requires the rationality of the timing relationship between the correctness and the state of the individual in the legal state. However, the current research of data cleaning is concentrated on the individual. The cleaning of data is rarely involved in the cleaning of temporal relational data. Therefore, it is imperative to build and implement a patent legal status information cleaning framework that can clean the temporal relationship. The multidimensional robust data quality analysis model can easily and efficiently complete the cleaning of temporal relational data. The model, namely, abstracts the state set and state transfer relation in the time series data, and formally expresses it, and then transforms the time series data cleaning task into the model detection task through the model detection method, thus identifying the temporal logic problems hidden in the time series data. This method has realized the automatic discovery of the data problem and improved the time sequence. The efficiency of data cleaning reduces the cost of cleaning. In order to complete the cleaning of the time sequence relationship between the legal state individual and the state, this paper uses the multidimensional robust data quality analysis model to model the legal state information database, and carries out the model detection to realize the cleaning of the temporal relation of the legal state information, and takes this as the core. Construction and implementation of the legal state information cleaning framework. Through the cleaning of actual data to verify the scientific and effective framework of the framework, the common error patterns in the temporal relationship of patent legal status information are found, and these errors are analyzed. This paper applies the multidimensional robust data quality analysis model to the patent legal state. In the cleaning of the temporal relation of information, a new idea is provided for the cleaning of time series data. At the same time, the data cleaning framework constructed in this paper provides an empirical reference for the cleaning of the same type of data.

【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：D923.42;G306

【相似文献】