Figure 4 Architecture of the deep fusion learning model.
Related Figures (10)
Fig. 2. Six kinds of caller-callee relationships among methods. Fig. 3. Syntactic feature embedding. Evaluation of precision. Table 2 It is computed by calculating the overall No. of identified CCs by the No. of accurately detected CCs (predicted). Table 2 compares extant and CCCD-DL methods for identifying all clone variants. Fig. 5. Evaluation of accuracy. Fig. 6. Evaluation of precision. Fig. 7 displays the results of recall achieved for CCCD-DL and existing Evaluation of recall. Table 3 methods to detect the clone types. For the given dataset Apache Maven 3.8.3, Appache ant 1.10.12 and Opennlp-master 1.9.1, it is observed that the recall of the CCCD-DL model is 13.91%, 12.58%, and 14.78% (LV- CCD), 9.20%, 9.31% and 9.22% (FCCA), 5.59%, 6.22% and 5.10% (TBCNN) is increased than the existing method respectively. According to the analyzed outcomes, the CCCD-DL framework has a greater recall than the emerging technique. Comparison of LV-CC types. Table 5 Comparison of Time (Processing) period and memory storage. Table 4 Fig. 7. Evaluation of recall.
