Standard index extraction method based on rule and neural network model fusion

Abstract

The invention provides a standard index extraction method based on rule and neural network model fusion. The method comprises the following steps: S1, selecting a quantitative standard text and carrying out data processing; s2, establishing a data extraction rule; s3, performing standard index extraction on the standard text; s4, performing deep learning neural network model training; s5, performing standard index extraction by using the model trained in the step S4, and guiding iteration of the data extraction rule in the step S2; s6, repeating the steps S2-S5, and iteratively extracting the rule and the model; and S7, performing standard index extraction by using the combination of a data extraction rule and the deep learning neural network model. According to the method, three indexes including parameter names, parameter values and constraint conditions can be extracted from standard texts, and the method aims at converting unstructured text data related to aviation field standards into structured knowledge information and ensuring the extraction accuracy so as to realize carding and effective utilization of standard index knowledge in the aviation field.

Description

technical field [0001] The invention relates to the field of standard indicators in the aviation field, in particular to a standard indicator extraction method based on the fusion of rules and neural network models. Background technique [0002] In aviation encyclopedia knowledge, standard index extraction is very important for content service in aviation field. By transforming the unstructured text data in the aviation field into structured knowledge information, we can sort out and effectively utilize the fragmented knowledge in the aviation field. [0003] The standard index extraction method is to extract three indexes of parameter name, parameter value and constraint condition from standard text, which is essentially a named entity recognition task in sequence labeling. Named entity recognition refers to identifying entities with specific meanings in text, such as person names, place names, institution names, time, currency, proper nouns, etc. It is a very basic and im...

AI Technical Summary

Problems solved by technology

Although the method of named entity recognition based on deep learning has achieved certain results, named entity recognition in professional fields still relies on specific rules and professional domain knowledge, and the effect of entity recognition is limited by specific field situations

Images