Task man-hour estimation model training method, task man-hour estimation method and device

By constructing a directed weighted task dependency graph and a Stacked GAT model, the problem of accuracy in task time estimation in software development projects is solved, and the quantification of dependencies and full-process automation are achieved, improving the accuracy and applicability of time estimation.

CN121543639BActive Publication Date: 2026-07-21WUHAN FIBERHOME TECHNICAL SERVICES CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
WUHAN FIBERHOME TECHNICAL SERVICES CO LTD
Filing Date
2026-01-21
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing technologies cannot accurately predict task duration in software development projects, especially in agile development and cross-domain projects, where they cannot effectively capture the topology of task dependencies and the propagation effect of dependency chains, resulting in a significant deviation between the estimated duration and the actual time consumption.

Method used

A task time prediction method based on the Stacked GAT model is adopted. By constructing a directed weighted task dependency graph, integrating task attributes and dependencies, and utilizing multi-head attention mechanism and global feature aggregation, the impact of dependency strength on time is quantified, thereby achieving accurate prediction of task time.

Benefits of technology

It improves the accuracy of time estimation in software development projects, adapts to different types of software project scenarios, supports incremental model training and interpretability analysis of prediction results, and reduces prediction bias rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121543639B_ABST
    Figure CN121543639B_ABST
Patent Text Reader

Abstract

The application discloses a task working hour estimation model training method, a task working hour estimation method and device, and belongs to the cross technical field of software development project management and artificial intelligence. The training method comprises the following steps: based on a historical project database, a directed weighted task dependency graph, a feature matrix and an actual working hour vector of each project are constructed, a training set and a test set are further constructed, a stacked graph attention convolution network model based on an attention mechanism is trained and evaluated by using the training set and the test set, and a trained stacked graph attention convolution network model based on the attention mechanism is obtained; wherein, the stacked GAT model comprises a stacked graph attention convolution layer, a global feature aggregation layer and a full connection prediction layer. A stacked graph convolution network based on an attention mechanism is designed, global features within a K-hop dependency range are captured through multi-layer message passing, a modeling problem of remote dependency influence is solved, and the accuracy of working hour estimation in a software research and development project can be improved.
Need to check novelty before this filing date? Find Prior Art