Combining conceptual graphs with the task-based specification method to specify software requirements helps capture richer semantics, and integrates requirements specifications tightly and uniformly.

Conceptual modeling is an important step toward the construction of user requirements. Requirements engineering is knowledge-intensive and cannot be dealt with using only a few general principles. Therefore, a conceptual model is domain-oriented and should represent the richer semantics of the problem domain. The conceptual model also helps designers communicate among themselves and with users.

To capture and represent a conceptual model for the problem domain, we need

We propose the task-based specification methodology as the mechanism to structure the knowledge captured in conceptual models. TBSM offers four main benefits for constructing conceptual models:

First, incorporating the task structure provides a detailed functional-decomposition technique for organizing and refining functional and behavioral specifications.

Second, the distinction between soft and rigid conditions lets us specify conflicting functional requirements.

Third, with TBSM, not only can we document the expected control flow and module interactions, but we can also verify that the behavioral specification is consistent with the system's functional specification.

Fourth, the state model makes it easier to describe complex state conditions. Terminology defined in the state model can easily be reused for specifying the functionality of different tasks. Without such a state model, describing the state conditions before and after a functional unit of an expert system is too cumbersome to be practical.

We propose conceptual graphs as the formalism to express task-based specifications where the task structure of problem-solving knowledge drives the specification, the pieces of the specification can be iteratively refined, and verification can be performed for a single layer or between layers. We chose conceptual graphs for their expressive power to represent both declarative and procedural knowledge, and for their assimilation capability--that is, their ability to be combined.