Function Point Method Based on Hierarchical Convolutional Neural Network
Jie Guo · 2018
Function Point method is vulnerable to personal knowledge and experience in software requirements assessment. Combined with JIRA-the work item tool, CNN method is used to extract the feature value of each item, through pooling layer compression and fully connected layer classification. The function classification and function point count of the item are finally determined, and the total function point count are obtained by weighting the item level. It is shown that this method is not easy to produce excessive deviation and can be used to assist prediction.