Integrated Collaboration Transaction Model

Collaboration is about sharing objects, a concept that has been explored in-depth by research in database transaction models. Database transaction models, however, are considered too strict for supporting collaboration - the main goal of databases is to isolate the transactions of different users while the goal in collaborative systems is to couple these transactions. Moreover, collaborative systems can be less strict about automatically enforcing consistency since users can be expected to communicate with each other to maintain consistency and/or go through a special merge phase to reconcile inconsistencies.

We have completed initial projects on coupling, concurrency control, recovery (undo/redo), and merging. Each of these projects has independently defined a collaboration model and associated parameters to support flexible collaboration. The goal of this work will be to build on previous projects by extending and integrating these separate models into an a single, integrated next-generation transaction model. It will have three main aspects: First, it will extend one or more of the individual models to address issues that were not considered by it. In particular, it will consider the impact on these models of mobile computing and also address versioning support for collaboration. Second, it will explore dependencies among the various collaboration parameters defined in the individual models and define global parameters that set consistent values of dependent parameters. The driving applications for this project will be traditional software development and document processing environments. Other projects, listed below, will examine transaction models for non-traditional environments such as virtual worlds and multimedia systems.

This project will give you experience with research in versioning, concurrency control, and mobile computing. For more information, contact Prasun Dewan.