WP5: Techniques for Supporting Cooperation/Coordination in Software Processes

Duration: M17 – M26

Description of the approach: This WP will examine the application of analysis techniques from the Social Network Analysis (SNA) domain, to promote the collaboration within members of a SW organization which are involved in a SPI project as well as facilitate communication and enable smooth information flow within the participants in the process improvement. The application of SNA techniques will be examined in case studies (SPIRE projects) which will take place in the involved Greek SW SMEs.

As far as research innovativeness is concerned, following the latest strands of social networks research (e.g. [35]), we propose to extend SNA to cover both quantitative and qualitative methods for capturing knowledge in SPI projects. This can be achieved by using agent-based modelling and simulation [36, 55, 56] together with static SNA methods. The SPI domain explored at a social level will provide the required concepts to study patterns of social networking and behaviour and find important criteria to evaluate the targeted SPI objectives. Subsequently, we will be interested in incorporating more intelligence, emotional aspects and capturing the trust between stakeholders (actors) involved in a SPI project. Incorporating such factors in the SNA framework will enable to consider criteria, such as trust, emotion and actor relations for characterizing SPI projects and devising best SPI practices by taking into account the particular needs of the individual actors of a SW organisation.

In current SNA software, it is common to calculate indicators of collaboration of actors and groups within a communication network. For example, a metric like the “contribution index” is defined as the number of messages sent minus the messages received, normalized by the total number of all messages. One of our research goals is to extend this and other social network indicators by adding “weighted factors” represented by the role of people in the organization or other meta-information of social nodes, such as emotions and capabilities. In this respect, we plan to use a validation system (an algorithm) to compare actor nodes based on such meta-information. For example, an actor will have a higher weight if he/she has a higher role, tenure or expertise in the SW company. Furthermore, a higher weight may be assigned to actors who have more relationships with important clients or suppliers.

Several agent engineering toolkits, such as JADE, and agent-based modelling software frameworks, such as SWARM, RePast and Ascape, are particularly suitable for constructing research-quality models. We plan to evaluate available tools and select the most suitable one for implementing and carrying out the necessary simulations. Furthermore, we will consider existing agent-based modelling methodologies, such as the one described in [57], and we will build realistic models reflecting the software engineering group behaviour at social level and the environment. The rules determining agent behaviours will be based on both qualitative and quantitative knowledge and will result in complex interactions and emergent collective behaviour.

Deliverables:

D5.1: Techniques for Supporting Cooperation/Coordination in Software Processes (M17-M26)

WP5 Leader: Anthony Karageorgos

WP5 Members: V.C. Gerogiannis, L. Anthopoulos, E. Rapti

Top