Building a Narrative-Based Requirements Engineering Mediation Model

Abstract

This thesis presents a narrative-based Requirements Engineering (RE) mediation model (NREMM) to help RE practitioners to effectively identify, define, and resolve conflicts of interests, goals, and requirements in the RE process. Conflict has received increasing attention in the Software Engineering (SE) literature. However, in the current RE literature, conflict has received fairly superficial attention. Previous research views conflict as an issue causing "inconsistency" in a requirements specification. As a result, current research has proposed a wide range of inconsistency checking and modelling tools, which address the result of conflict rather than fundamental causes and roots of the conflict. Little attention is given to the socio-psychological aspect of conflict. Furthermore, previous work views the resolution of conflict in RE as a negotiation-based process, in which a requirements engineer acts as a representative of a developer site and negotiates with users. This thesis differentiates itself from previous work by recognising conflict is more than an issue of inconsistency. This thesis argues that conflict is a social and organisational phenomenon. This thesis also argues that the process of resolving conflict in RE can be viewed as a socially mediated process, in which a requirements engineer can act as a mediator rather than a representative of the developer/user site. The fundamental difference between negotiation and mediation is that, negotiations often only involve parties themselves reaching an agreement. Mediation then involves a mediator to lead the process and help parties to reach an agreement. However, there is a distinct gap in the RE literature, in which the role of a requirements engineer as a mediator has not been explicitly explored. To address the socio-psychological aspect of conflict, Wins lade and Monk (2000)'s narrative mediation model is introduced, justified and translated into the context of RE by following a systematic and transparent methodological approach. This leads to a new RE specialised mediation model (NREMM), which includes three phases: conflict identification, conflict definition and conflict resolution. The new model aims to be not only theoretically robust but also practically useful. It builds on the storytelling metaphor advocated by narrative mediation theory and also integrates well-established and practical RE specialised techniques. The NREMM was empirically assessed and evaluated. I used an expert panel interview survey to empirically assess whether the NREMM is theoretically robust. The experts' feedback indicated that the NREMM is capable of helping RE practitioners to resolve conflict in the RE process. The experts also indicated that the use of story-telling as a theoretical underpinning is a strength, and matches well with current state of the RE practice. I also used a quasi-experiment to empirically evaluate whether the NREMM is practically useful in a simplified real-world scenario (University of Hertfordshire's StudyNet) by using real users of a real system. The experimental results indicated that the NREMM is a useful model to help RE practitioners to identify, define and resolve conflict in practice. This is evidenced by significantly higher satisfaction results and a better perception of the mediator's performance obtained from workshops where the mediator implemented the NREMM. The key contribution of this thesis is the NREMM, which is a useful model to not only help RE practitioners resolve conflicts among different stakeholders but also improve stakeholders' cooperativeness and satisfaction.