In recent years, software project managers compare actual completion of activities against the progress reports filled by project members to identify significant deviations from the estimated schedules and manage software project risks. However, quantitative measurements are limited due to the format of project documents, which are mostly natural languages. In this paper, we propose an intelligent system for software project monitoring and control by using natural language processing techniques to recognize textual entailment of progress reports to further evaluate the level of project fulfillment in a qualitative manner. Our experimental results demonstrate that the proposed method can recognize entailment from text efficiently and outperform other textual entailment approaches. Moreover, we successfully apply the textual entailment technique to project monitoring and control, which not only reduces the project cost and human's effort but also provides a basis for project managers to qualitatively evaluate the performance of each project member.
ASJC Scopus subject areas
- Control and Systems Engineering
- Electrical and Electronic Engineering