System and method for automatically generating answers to programming specification questions

Abstract

The invention discloses a system and a method for automatically generating answers to programming specification questions. The system comprises a subgraph matching module, a machine reading understanding module and a learning sorting module. The method comprises steps of a programming specification knowledge graph being constructed, and natural language questions being converted into statements conforming to an SPARQL protocol through the subgraph matching module; and querying to obtain candidate answers by using the structured information in the programming specification knowledge graph; selecting all nouns, nouns phrases and subjects from the natural language questions through a machine reading understanding module, marking the words as keywords, using an Apache Solr engine to search toobtain ten most relevant natural segments, using an algorithm based on TF-IDF to score the searched natural segments, and using a TF-IDF algorithm to obtain the most relevant natural segments; using the trained deep learning model for each natural segment to obtain candidate answers, combining the candidate answers through a learning sorting module to generate new candidate answers, and using a logistic regression classifier to sort the new candidate answers according to correct probabilities.

Description

technical field [0001] The invention relates to a technology in the field of artificial intelligence semantic recognition, in particular to a system and method for automatically generating answers to programming specification questions. Background technique [0002] In recent years, programming norms have played an increasingly important role in software quality assurance. Programming specifications are a series of code guidelines that help software developers improve code readability, maintainability, and reusability. When software developers encounter problems related to programming specifications, they can query programming specification documents. However, a programming specification may appear in multiple programming specification documents, and the content of each programming specification document may be different. Therefore, software developers need to query or browse multiple documents to obtain desired information, which is very cumbersome. The question-answerin...

AI Technical Summary

Problems solved by technology

[0003] The present invention aims at defects such as insufficient utilization of knowledge in the field of programming specification of the existing question-and-answer robot system, inaccurate understanding of user intentions, etc., and proposes a system and method for automatically generating answers to programming specification questions, and by combining knowledge graphs and machine reading comprehension techniques, fully Leverage programming specification domain knowledge to more precisely understand user intent, thereby improving the accuracy of automatically generating answers to programming specification questions

Images