Judgment document structured partitioning method and device

Abstract

The invention provides a judgment document structured partitioning method and device, and relates to the technical field of data mining and data analysis. The method comprises the following steps: firstly, simplifying paragraphs of a cleaned to-be-processed judgment document; then, inputting the simplified paragraphs into a hierarchical sequence labeling model for calculation; and finally, takinga label with the maximum output probability of the hierarchical sequence labeling model as a block label. According to the scheme, the hierarchical sequence labeling model is adopted, semantic information between paragraphs of the judgment document can be captured, semantic features of the paragraphs can also be captured, paragraph partitioning is carried out from the perspective of semantics, andthe partitioning effect of the judgment document can be improved. By adopting the simplified paragraphs, the overhead of model calculation can be reduced, and the processing efficiency is improved. Meanwhile, the scheme does not need excessive manual participation, the development is simple, and the maintenance cost is low. Newly-added structural blocks of the judgment document can be accuratelyrecognized through partitioning based on semantic features, and good expansibility is achieved.

Description

technical field [0001] The present invention relates to the technical field of data mining and data analysis, and in particular, to a method and device for structured block division of referee documents. Background technique [0002] Judgment documents, as judicial data published online, are very important for data analysis and information mining in various industries such as lawyers, finance, and government. However, referee documents are generally very long, which is very inconvenient for users to read and locate information. Therefore, the structural processing of the judgment documents and the structured division of the content can facilitate the mining and analysis of the judgment document data, and also facilitate the reading and analysis of relevant users such as lawyers and judges. [0003] Most of the existing referee documents are structured using regular expressions. Due to the massive data and diverse expressions of referee documents, many regular expressions ne...

AI Technical Summary

Problems solved by technology

[0003] Most of the existing referee documents are structured using regular expressions. Due to the massive data and diverse expressions of referee documents, many regular expressions need to be developed. At the same time, the existing regular expressions may not be able to handle new Increase the structure of judgment documents

Another problem with the use of regular expressions in the structural processing of referee documents is that due to the existence of a large number of regular expressions at the same time, it is necessary to manually sort the application order of the regular expressions to eliminate conflicts between different regular expressions, which makes the segmented results Accuracy is heavily dependent on manual operation

To sum up, the current structured processing of referee documents has technical defects such as troublesome development, high maintenance costs, and poor scalability

Images