A STUDY ON THE ENHANCEMENT OF COMMUNICATION STRATEGIES IN VIRTUALSOFTWARE PROJECT TEAMS THROUGH A SYSTEMATIC LITERATURE REVIEW

Abstract

The study is called A Study on the Enhancement of Communication Strategies in Virtual Software Project Teams and examines how artificial intelligence and deep learning would improve communication in distributed software teams. The proposed research is based on the use of the Kaggle Virtual Team & Collaboration Dataset, where it is possible to distinguish such tools of virtual collaboration as GitHub (82.8%), Slack (72%), Jira (54%), and Google Workspace (41.5%), which are the most popular platforms in the world of software development. To examine the patterns of team communication, decision synchronization, and task coordination, three models were implemented, including Long Short-Term Memory (LSTM), Convolutional neural network (CNN), and Swin Transformer. Both of the models included important functions such as ReLU, Softmax, Layer Normalization, and Multi-Head Self-Attention to learn effectively. The dataset has been divided into 70 percent training, 15 percent validation and 15 percent testing to have a balanced assessment of the performance of models. The error reduction measures were statistical analysis and optimization, i.e. normalization and dropout regularization, to reduce overfitting and enhance generalization. The LSTM scored 94 percent, CNN scored 97 percent, and Swin Transformer scored highest at 98.1, which proves that it has better contextual awareness of communication data. The study adds a framework supported by data to the software management literature as it proves that the deep learning models have the potential to close the communication gaps and improve the quality of collaboration, which is needed more and more as part of the intelligent communication strategies in projects located over the globe.