TL;DR
This paper proposes extensions to the Unified Modeling Language (UML) to better integrate human-computer interaction (HCI) practices, specifically work, process, and task analysis, into object-oriented systems development. The goal is to bridge the gap between how HCI practitioners understand user needs and how software engineers build systems, ultimately leading to more user-centered software.
What Problem Does This Paper Address?
The paper tackles a fundamental disconnect between two key communities in software development: Human-Computer Interaction (HCI) and object-oriented (OO) development. HCI practitioners excel at understanding user needs through task analysis and scenarios, while OO developers rely on methodologies like UML for system design. Historically, these two approaches have operated in parallel, leading to challenges in effectively translating user requirements into system designs. This paper aims to bridge this gap by proposing a way to integrate HCI-derived information directly into the UML framework, making it a more comprehensive tool for user-centered design.
Methods
The paper's core contribution is the proposal of extensions to UML to accommodate HCI concepts. This involves:
- Extending UML Semantics: The authors suggest that UML's existing constructs, particularly "Dependency" relationships, can be leveraged to link different models. They propose defining specific types of dependencies to represent the flow of information from task analysis to object models.
- Proposed Task Model: A new "User Task Model" is introduced, which is a UML-compliant extension. This model includes elements like "Task," "Task Instance (Scenario)," "Actor," and "Goal," aiming to capture the nuances of user activities.
- Traceability Mechanisms: The paper emphasizes the need for mechanisms to trace information from external sources (like documents and prototypes) into UML models, supporting "literate development."
- Bridging the Gap: The overarching method is to adapt and extend UML to directly incorporate the outputs of HCI analysis techniques, making them first-class citizens in the development process.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- UML as a Unifying Language: The paper posits that UML, with appropriate extensions, can serve as a common language and framework for both HCI and OO development, fostering better collaboration.
- The Importance of "Descriptions": The concept of "Descriptions" (from a previous workshop) is central. These are artifacts that capture different views of the user's world. In UML terms, these correspond to "Models." The relationships between these models (mappings) define the development process.
- Task Analysis is Crucial for OO Design: The authors argue that task analysis, a cornerstone of HCI, provides essential information for defining objects and system functionality in OO development.
- Use Cases are a Starting Point, Not the End: While acknowledging the widespread use of UML's "Use Cases" in OO development, the paper highlights their limitations for comprehensive task modeling and proposes a separate "User Task Model" to complement them.
- Ontological Drift is a Reality: The paper recognizes that the "world" of users and the "world" of the system can drift apart over time. This drift needs to be acknowledged and managed throughout the development process.
Key Results
The paper doesn't present quantitative results from empirical studies. Instead, its key results are conceptual and prescriptive:
- A Proposed UML Extension Framework: The paper outlines a conceptual framework for extending UML to better support HCI activities.
- A Proposed User Task Model Notation: A visual representation (Figure 1) of the proposed User Task Model is provided, demonstrating how task-related information can be structured within a UML-like notation.
- Recommendations for Tool Support: The authors strongly encourage CASE tool vendors to support these proposed extensions to enable practical application of the framework.
Limitations and Future Work
- Business Process Modeling: The authors acknowledge that they were unable to give sufficient attention to "work process modeling" due to time constraints. They suggest this is an important area for future exploration.
- UI Architecture: Similarly, "user interface architecture" was a tantalizing area of discussion that couldn't be explored in detail. The authors believe further discussion here could lead to significant improvements.
- Dissemination and Feedback: The immediate next step proposed is to disseminate this work to the OMG community and the broader object-oriented community to solicit comments and feedback.
- Tool Support: The authors emphasize the need for tool support to make these extensions practical for everyday development.
Summary updated