[Paper Review] A Model for Interactive Scores with Temporal Constraints and Conditional Branching

Mauricio Toro, Myriam Desainte‐Catherine+1 more|arXiv (Cornell University)|Oct 9, 2015

Constraint Satisfaction and Optimization4 references17 citations

TL;DR

This paper proposes an extended model for Interactive Scores (IS) that integrates conditional branching with temporal relations by introducing Timed Conditional Relations (TCRs), enabling dynamic control over multimedia scenarios. The model uses temporal points and conditions to manage execution flow, supporting choice, loops, and random durations—demonstrated through the Mariona multimedia installation—while showing that rigid durations cannot always be preserved in such scenarios.

ABSTRACT

Interactive Scores (IS) are a formalism for the design and performance of interactive multimedia scenarios. IS provide temporal relations (TR), but they cannot represent conditional branching and TRs simultaneously. We propose an extension to Allombert et al.'s IS model by including a condition on the TRs. We found out that in order to have a coherent model in all possible scenarios, durations must be flexible; however, sometimes it is possible to have fixed durations. To show the relevance of our model, we modeled an existing multimedia installation called Mariona. In Mariona there is choice, random durations and loops. Whether we can represent all the TRs available in Allombert et al.'s model into ours, or we have to choose between a timed conditional branching model and a pure temporal model before writing a scenario, still remains as an open question.

Motivation & Objective

• Address the limitation of existing Interactive Score models in simultaneously supporting conditional branching and temporal relations.
• Enable dynamic, interactive multimedia scenarios involving choice, loops, and random durations—common in installations like Mariona.
• Develop a formal model that maintains temporal consistency while allowing runtime decisions based on conditions.
• Investigate the feasibility of preserving rigid and semi-rigid durations in conditional branching scenarios.
• Demonstrate the model’s applicability through a case study on the Mariona multimedia installation.

Proposed method

• Extend Allombert et al.'s IS model by introducing Timed Conditional Relations (TCRs), where temporal relations are guarded by conditions.
• Represent temporal relations using temporal points (start/end) and encode Allen’s interval relations (e.g., meets, during) as TCRs with condition checks.
• Use a constraint-based execution model where conditions are evaluated in the local environment to determine relation activation.
• Model conditional branching via condition guards on TCRs, allowing dynamic selection of execution paths during performance.
• Support trans-hierarchical relations by evaluating conditions in the context of the parent object’s environment.
• Preserve flexibility in durations by allowing durations to be variable unless explicitly constrained, with analysis of cases where rigid durations can still be maintained.

Experimental results

Research questions

• RQ1 Can conditional branching and temporal relations be unified in a single formal model for Interactive Scores?
• RQ2 What are the constraints on preserving rigid and semi-rigid durations when combining conditional branching with temporal relations?
• RQ3 How can temporal relations such as 'meets' or 'during' be encoded as conditionally guarded relations in a dynamic execution model?
• RQ4 In what scenarios can rigid durations coexist with conditional branching, and what are the structural or logical conditions enabling this?
• RQ5 How can complex multimedia behaviors—such as loops, random delays, and user-driven choices—be formally represented and executed using the proposed model?

Key findings

• The proposed model successfully integrates conditional branching with qualitative temporal relations (e.g., 'A must be played during B') via conditionally guarded Timed Conditional Relations (TCRs).
• Rigid and semi-rigid durations cannot be preserved in all scenarios involving conditional branching, as the model requires flexible durations for consistency across all execution paths.
• The model can represent all qualitative temporal relations from Allombert et al.’s original IS model by encoding them as TCRs with appropriate conditions.
• The Mariona multimedia installation was successfully modeled using the proposed framework, demonstrating support for choice, loops, random durations, and trans-hierarchical relations.
• While quantitative temporal relations (e.g., 'duration of A is 3 seconds') cannot be universally preserved under conditional branching, specific cases allow for their retention when constraints are carefully structured.
• The model enables dynamic execution behaviors such as non-deterministic loop termination and user-driven path selection, validated through the Mariona case study.