Enhanced evolutionary computing based artificial intelligence model for web-solutions software reusability estimation

Ensuring the aging resilient software design can be of paramount significance to enable faultless software system. Particularly assessing reusability extent of the software components can enable efficient software design. The probability of aging proneness can be characterized based on key OO-SM like cohesion, coupling and complexity of a software component. In this paper, aging resilient software reusability prediction model is proposed for object oriented design based Web of Service (WoS) software systems. This work introduces multilevel optimization to accomplish a novel reusability prediction model. Considering coupling, cohesion and complexity as the software characteristics to signify aging proneness, six CK metrics; WMC, CBO, DIT, LCOM, NOC, and RFC are obtained from 100 WoS software. The extracted CK metrics are processed for min-max normalization that alleviates data-unbalancing and hence avoids saturation during learning. The 10-fold Cross-validation followed by outlier detection is considered to enrich data quality for further feature extraction. To reduce computational overheads RSA algorithm is applied. SoftAudit tool is applied to estimate reusability of each class, while binary ULR estimates calculates (reuse proneness) threshold. Applying different classification algorithms such as LM, ANN algorithms, ELM, and evolutionary computing enriched ANN reuse-proneness prediction has been done. The performance assessment affirms that AGA based ANN model outperforms other techniques and hence can be used for earlier aging-resilient reusability optimization for WoS software design.