A three-layer architecture for reconciling compound actions in collaborative editing systems

In collaborative editing systems with replicated architecture, reconciliation of editing actions generated from the user interfaces of different editors for consistency maintenance is a critical issue. Consistency of compound editing actions cannot be maintained by simply maintaining the consistency of the corresponding primitive editing operations due to the semantic loss in representing compound editing actions with primitive editing operations. In this paper, we contribute a three-layer architecture to bridge the semantic gap between editing actions generated from the editor user interface and editing operations that are used to represent these actions. In this approach, editor-dependent editing actions are intercepted from the user interface and represented as editor-independent primitive and compound editing operations. These editing operations are then used to maintain the consistency of editor-dependent editing actions by leveraging operational transformation, and finally are translated into editor-dependent primitive editing operations for execution in individual editors. This approach can preserve the intentions of editing actions even in the presence of syntactic conflicts, while retaining the compatibility with existing single-user editors.