Method for enhancing classification-based software demand tracking link recovery through knowledge learning and electronic device

A technology for software requirements and knowledge learning, applied in the computer field, can solve problems such as limited application, lack of requirements and code semantics, time-consuming and labor-intensive, etc., to ensure accuracy, reduce workload, and reduce recovery costs.

CN113011461AActive Publication Date: 2021-06-22INST OF SOFTWARE - CHINESE ACAD OF SCI
5 Cites 3 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2021-06-22

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention provides a method for enhancing classification-based software demand tracking link recovery through knowledge learning and an electronic device. The method comprises the steps: after text and structure information contained in a software demand and code file with a tracking relation to be determined is preprocessed, constructing a demand-code knowledge graph and a code dependency graph; modeling the structure and text information of the demand-code knowledge graph, and performing learning to obtain vectors of demand and code entities; modeling the relation of the triple in the demand-code knowledge graph to obtain a relation feature vector; and mining the code dependency graph, extracting inference rules for discovering R2C links between potential demands and codes, and expanding the training data scale. According to the method, an effective prediction model can still be obtained when the training data is less, so that the recovery accuracy of the R2C link is ensured, the recovery cost is reduced, and the workload of manually annotating the data is reduced.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the field of computer technology, and in particular relates to a classification-based software requirement tracking link recovery method and an electronic device enhanced through knowledge learning. Background technique

[0002] Requirements-to-Code (requirements to code, R2C) is a tracking relationship between software requirements and codes. It builds a logical abstraction bridge between requirements written in natural language description and source code written in programming language. R2C links can help developers better understand the logic and purpose of source code, and help locate codes affected by changes in requirements, thereby greatly reducing the maintenance cost of software projects. R2C links can also be used to automatically select relevant test cases for execution, which makes automated testing projects more efficient. However, R2C links are often lost or wrongly linked in actual work, and most R2C links are ...

Examples

Embodiment approach

[0115] 1. It is possible to construct a demand-code knowledge map similar to the definition of the present invention, and add / delete different entity relationships; and then use the TransR+DKRL expanded knowledge representation learning algorithm and other knowledge representation learning algorithms to construct relationship feature vectors;

[0116] 2. The inference rules based on intimacy analysis can be used to expand the training set with different threshold settings.