Software requirements activity, in the organizational context, is about addressing the business information problem; discover the needs for improving the situation and consequently specify the software requirements. Goal-Oriented Requirements Engineering (GORE), aims to better understand the information problem by looking at organizational actors' intentionality (goals and softgoals) first. Eliciting goals and softgoals within an organizational context is a difficult task: since, among other things, it demands skills and time. This paper describes one strategy for eliciting goals and softgoals that still relies on software engineers' skills and time, but it simplifies the process. We propose the use of a software tool to support a systematic process to mitigate the chances for goals to be missed regardless of the experience and skills of the software engineers involved in the project.
illustrate the proposed strategy. Section 2 describes the AGFL (Actor Goals from Lexicon) strategy concepts using the AGFL tool prepared to facilitate the job, its concepts, and it shows, in a simple way, the central ideas of the AGFL strategy for perceiving the intentionality and how the process is carried out. An in-vitro experimentation run by UERJ students is portraited. Section 3 concludes stressing the continuity of the requirements process.
The AGFL Strategy provides activities to guide goals and softgoals elicitation. Figures and examples illustrated in this Section were extracted from an in-vitro experiment conducted with undergraduate students. They were divided into 4 groups of 3 students for preparing this experiment work of modeling TRC System.
The first activity of the strategy is “A - Build Lexicon”. The strategy adopts the
Language Extended Lexicon (LEL) [
The activity “B - Extract Goals” requires that the RE recognizes goals and softgoals and organize them by actors.
For recognizing goals, we build on Eric Yu´s observation: “A goal is a condition
or state of affairs in the world that an actor would like to achieve” [
Behavioral Responses (BRs) in LEL symbols mention actions which happen in the organizational context. Two kinds of actions can be observed: concrete actions and flexible actions. A concrete action changes one state into another, and a flexible action adds a quality attribute to a state.
Oliveira [
When one concrete action is found ➔ the action will define a goal.
When one flexible action is found ➔ the action will define a softgoal.
This activity (B - Extract Goals) is connected to the C&L Tool [
Figure 2 shows one example portraying the classifying actions. Usually, on the screen, we use the field RATIONALE to describe the flexible actions justification. 1 In our context states are interpreted as “desired states”. For defining goal elements (exemplified in Figure 3), the RE must select for each BR that denotes a concrete action one LEL symbol element (subject or object) and fill in one verb in a passive voice. Furthermore, if the current actor (see “AUTHORIZED DRIVER”) depends on another actor (“operator”) to achieve the goal, RE must indicate this by adding “ ” a new line, defining a second actor’s goal. We call this case a reflexive goal when one actor has a goal but depends on another actor for the goal achievement.
The activity “C - Refine Goals”, requires that the RE organizes goals and softgoals
as a list sorted in chronological order. The RE should recognize when one goal comes
before another one. Long-time goals should be placed at the end of the list. The
method proposes two activities to refine the actor’s goals: merge goals (concrete and
softgoals) by actor and set them in chronological order. Chronological order means
long term goals first (the most abstract before and the less abstract after). This order is
important on modeling according to the ERi*c method [
For example, explaining Figure 5, ADMINISTRATION goals chain in chronological order is: toll BE charged is important for toll BE paid which is required for toll BE computed, and toll BE computed is necessary for road BE maintained, and so on.
Figure 5 shows the final list of AGFL of Toll Road System goals. The final report shows two new elements: “DEPENDER” and “DEPENDEE”. DEPENDER is the first actor, the LEL subject of the actions, and DEPENDEE is the second actor who appears in the elicitation process as an actor from whom the subject (“DEPENDER”) depends on to achieve one goal. This idea of “DEPENDER” and “DEPENDEE” is the same used by iStar Framework models.
The aim of this work is to propose a strategy to help the RE in the intentionality dimension of the elicitation process.
The AGFL presented in this work is an extension of the first step of the ERi*c
Method [
For modeling goals and softgoals before the application of iStar Models, the ERi*c
Method uses a diagram language similar to state charts that are a simpler view of iStar
SR model, to represent chains of goals and softgoals (states) relationships. These
diagrams are called “Intentionality Panels” [
The AGFL Tool was developed using PHP, Javascript and MySQL, it has almost 2000 lines of code and required a 9 man-months effort. AGFL Tool will be available on the i* wiki. Future work is aimed at integrating the AGFL and IP Diagram tools. Our contribution relies on proposing a set of heuristics supported by a tool to help the discovery of goal and softgoals.