Requirements Elicitation

Where are we right now? Three ways to deal with complexity: Abstraction Decomposition (Technique: Divide and conquer) Hierarchy (Technique: Layering) Two ways to deal with decomposition: Object-orientation and functional decomposition Functional decomposition leads to unmaintainable code Depending on the purpose of the system, different objects can be found What is the right way? Start with a description of the functionality (Use case model). Then proceed by finding objects (object model). What activities and models are needed? This leads us to the software lifecycle we use in this class The identification of objects and the definition of the system boundary are heavily intertwined with each other.

Software Lifecycle Definition Set of activities and their relationships to each other to support the development of a software system Typical Lifecycle questions: Which activities should I select for the software project? What are the dependencies between activities? How should I schedule the activities? What is the result of an activity

Software Lifecycle Activities Requirements Elicitation Requirements Analysis System Design Object Design Implemen- tation Testing Implemented By Expressed in Terms Of Structured By Realized By Verified By class... ? class.... ? Application Domain Objects Use Case Model Solution Domain Objects SubSystems Source Code Test Cases

First Step in Establishing the Requirements: System Identification The development of a system is not just done by taking a snapshot of a scene (domain) Two questions need to be answered: How can we identify the purpose of a system? Crucial is the definition of the system boundary: What is inside, what is outside the system? These two questions are answered in the requirements process The requirements process consists of two activities: Requirements Elicitation: Definition of the system in terms understood by the customer (“Problem Description”) Requirements Analysis: Technical specification of the system in terms understood by the developer (“Problem Specification”) The identification of objects and the definition of the system boundary are heavily intertwined with each other.

Products of Requirements Process (Activity Diagram) Problem Statement Problem Statement Generation Requirements Elicitation system specification: Model Requirements Analysis analysis model: Model

System Specification vs Analysis Model Both models focus on the requirements from the user’s view of the system. System specification uses use cases and natural language to describe the behavior The analysis model uses formal or semi-formal notation They are both part of the Requirements Analysis Document

Requirements Analysis Document Introduction Current System Proposed System Functional requirements Nonfunctional requirements System Models Scenarios Use case model Object model Dynamic model User interface Glossary

Requirements Elicitation Overview and Challenges Problem Statement Functional and Nonfunctional Requirements Requirements Validation Criteria

Requirements Elicitation Very challenging activity Users don’t know what they need the system to do. They could not articulate their needs. Users don’t understand the limits of the available technology. Users express requirements using terms from their application domain. Users representing different stakeholders have requirements that vary to some degree. Political factors may influence how users express requirements. The economic and business environment is constantly changing and can affect the relative importance of certain requirements.

Requirements Elicitation Requires collaboration of people with different backgrounds Users with application domain knowledge Developer with solution domain knowledge (design knowledge, implementation knowledge) Bridging the gap between user and developer: Scenarios: Example of the use of the system in terms of a series of interactions with between the user and the system Use cases: Abstraction that describes a class of scenarios Starts with Problem Statement

Problem Statement The problem statement is developed by the client as a description of the problem addressed by the system Other terms for problem statement: Statement of Work A good problem statement describes The current situation The functionality the new system should support The environment in which the system will be deployed Deliverables expected by the client Delivery dates A set of acceptance criteria

Ingredients of a Problem Statement Current situation: The Problem to be solved Description of one or more scenarios Requirements Functional and Nonfunctional requirements Constraints (“pseudo requirements”) Project Schedule Major milestones that involve interaction with the client including deadline for delivery of the system Target environment The environment in which the delivered system has to perform a specified set of system tests Client Acceptance Criteria Criteria for the system tests

Categories of Projects Current situation depends on the category of the project Greenfield Engineering No prior system exists or existing system is being significantly extended with new functionality The requirements are extracted from the end users and the client Triggered by new user needs Example: A new online boardgame Re-engineering Re-design and/or re-implementation of an existing system using newer technology, essential purpose of system is unchanged The requirements are reverse engineered from the existing system Triggered by technology enabler Example: Replacing a mainframe COBOL system with SAP Interface Engineering Provide the services of an existing system in a new environment Requirements are dictated by users and existing system behavior Triggered by technology enabler or new market needs Example: Replacing curses-based forms with web-based forms

Types of Requirements Functional requirements: Describe the interactions between the system and its environment independent from implementation Examples: The system should alert the dispatcher of a new incident. Nonfunctional requirements: User visible aspects of the system not directly related to functional behavior. The response time must be less than 1 second within the reporting of the incident. Constraints (“Pseudo requirements”): Imposed by the client or the environment in which the system operates The implementation language must be Java

Requirements Validation Activity involving the client and user Requirements validation is a critical step in the development process, usually after requirements engineering or requirements analysis. Also at delivery (client acceptance test).

Requirements Validation Criteria Completeness All possible scenarios through the system are described, including exceptions Consistency There are no contradicting requirements Clarity/Unambiguity The specification can only be interpreted one way Correctness Requirements represent accurately the system the client needs Realism The system can be implemented within constraints Verifiability Tests can be designed to demonstrate the system fulfills its requirements Traceability Requirements can be traced to system functions System functions can be traced to requirements Dependencies among requirements, system functions, and everything else in between can be tracked.

Requirements Elicitation Activities Identifying Actors Identifying Scenarios Identifying Use Cases Refining Use Cases Identifying Relationships between Actors and Use Cases Identifying Initial Analysis Objects Identifying Nonfunctional Requirements

Identifying Actors Actors – person or machine using the system in a particular role Actors usually correspond to existing roles within the client organization Guide Questions Which user groups are supported by the system to perform their work? Which user groups execute the system’s main functions? Which user groups perform secondary functions, such as maintenance and administration? With what external hardware or software system will the system interact? Watch out for confusion between actors and objects

Identifying Scenarios “A narrative description of what people do and experience as they try to make use of computer systems and applications” [Carrol, Scenario-based Design, 1995] Informal description of a single feature from the viewpoint of a single actor Types of Scenarios As-is scenarios – describes current situation Visionary scenarios – describes future system Evaluation scenarios – describes user tasks for evaluating the system (acceptance criteria) Training scenarios – introduces new users to the system

Heuristics for Identifying Scenarios Ask yourself or the client the following questions: What are the primary tasks that the system needs to perform? What data will the actor create, store, change, remove or add in the system? Who else can modify this data? What external changes does the system need to know about? What changes or events will the actor of the system need to be informed about? However, don’t rely on questionnaires alone. Insist on task observation if the system already exists (interface engineering or reengineering) Ask to speak to the end user, not just to the software contractor Expect resistance and try to overcome it

Scenario Example: Warehouse on Fire Bob, driving down main street in his patrol car notices smoke coming out of a warehouse. His partner, Alice, reports the emergency from her car. Alice enters the address of the building, a brief description of its location (i.e., north west corner), and an emergency level. In addition to a fire unit, she requests several paramedic units on the scene given that area appear to be relatively busy. She confirms her input and waits for an acknowledgment. John, the Dispatcher, is alerted to the emergency by a beep of his workstation. He reviews the information submitted by Alice and acknowledges the report. He allocates a fire unit and two paramedic units to the Incident site and sends their estimated arrival time (ETA) to Alice. Alice received the acknowledgment and the ETA.

Observations about Warehouse on Fire Scenario Concrete scenario Describes a single instance of reporting a fire incident. Does not describe all possible situations in which a fire can be reported. Participating actors Bob, Alice and John

Another Example Bob takes server out of service for software upgrades System detects that server is no longer active System identifies backup server manage.ist.unomaha.edu is notified to switch to IP address of backup server manage.ist.unomaha.edu replaces IP address in its lookup table A script running on a monitoring server determines when the primary server becomes unavailable. After verifying that the primary server is no longer responding, the monitoring server issues a command to the Web sites' DNS servers to change the 'A' records for monitored sites. This changes the primary IP address for the monitored sites to the IP address of the failover server. The monitored sites' DNS zones have a maximum time to live (TTL) of 5-10 minutes. This means that all Internet DNS servers that cache zone information for a site must update the cache every 5-10 minutes. When the monitoring server detects that the primary server is available again, it issues commands to the DNS server to change the primary IP address for the monitored sites back to that of the primary server.

Observations about Web Server Failover Scenario Concrete scenario Describes a single instance of failover incident. Does not describe all possible situations in which failover can occur (e.g., server crashed, denial of service, etc). Participating actors Network Administrator, DNS

Identifying Use Cases Use Case Motivations for use cases Specifies all possible scenarios for a given functionality Initiated by an actor Motivations for use cases Generalizing related scenarios help developers define the scope of the system The role of each user of the system is clarified Use Case Descriptions Entry and exit conditions Flow of events Quality requirements

Heuristics: How do I find use cases? Select a narrow vertical slice of the system (i.e. one scenario) Discuss it in detail with the user to understand the user’s preferred style of interaction Select a horizontal slice (i.e. many scenarios) to define the scope of the system. Discuss the scope with the user Use illustrative prototypes (mock-ups) as visual support Find out what the user does Task observation (Good) Questionnaires (Bad)

Order of steps when formulating use cases First step: name the use case Use case name: ReportEmergency Second step: Find the actors Generalize the concrete names (“Bob”) to participating actors (“Field officer”) Participating Actors: Field Officer (Bob and Alice in the Scenario) Dispatcher (John in the Scenario) Third step: Then concentrate on the flow of events Use informal natural language

Use Case Example: ReportEmergency Use case name: ReportEmergency Participating Actors: Field Officer (Bob and Alice in the Scenario) Dispatcher (John in the Scenario) Exceptions: The FieldOfficer is notified immediately if the connection between her terminal and the central is lost. The Dispatcher is notified immediately if the connection between any logged in FieldOfficer and the central is lost. Flow of Events: on next slide. Special Requirements: The FieldOfficer’s report is acknowledged within 30 seconds. The selected response arrives no later than 30 seconds after it is sent by the Dispatcher.

Use Case Example: ReportEmergency Flow of Events The FieldOfficer activates the “Report Emergency” function of her terminal. FRIEND responds by presenting a form to the officer. The FieldOfficer fills the form, by selecting the emergency level, type, location, and brief description of the situation. The FieldOfficer also describes possible responses to the emergency situation. Once the form is completed, the FieldOfficer submits the form, at which point, the Dispatcher is notified. The Dispatcher reviews the submitted information and creates an Incident in the database by invoking the OpenIncident use case. The Dispatcher selects a response and acknowledges the emergency report. The FieldOfficer receives the acknowledgment and the selected response.

Identifying Use Cases Writing Guide Choose proper name – use verb phrases; indicate user’s objective Name actors with noun phrases Clearly distinguish actors’ actions from system’s actions Use active voice to phrase steps in flow of events The causal relationship between steps should be clear Describe complete user transaction Describe exceptions separately Do not describe the user interface Use cases should not exceed 2-3 pages – break up using <<include>> and <<extends>> relationships

Refining Use Cases Goal: completeness and correctness Refining use case descriptions leads to other use cases and clarifies system boundaries Entry and exit conditions – additional use cases are identified as entry and exit conditions are refined Flow of events – discussing flow of events clarifies system boundaries Quality requirements – elicit nonfunctional requirements in the context of this particular functionality Refinements Details of objects in the system Low-level interactions between actors and system Access rights Missing exceptions Common functionality among use cases

Refining Use Cases Heuristics Use scenarios to communicate with users and validate functionality Refine a single scenario to understand user’s assumptions Define many high-level scenarios to determine scope of the system Use mock-ups or prototypes for visual support Present user with a range of alternatives Detail a broad vertical slice when scope of system and user preferences are well-understood

Relationships Between Actors and Use Cases <<initiate>> <<participate>> Determines access rights Who can initiate a functionality Who else is involved in this functionality Relationships between use cases Heuristics for making use cases shorter and simpler to understand <<include>> For factoring out common functionality Explicitly invoked from the including use case <<extend>> For specifying exceptions Entry conditions of the extending use case determine when it is used Caveat: use discretion when applying these decompositions (a few longer use cases are sometimes easier to understand than many short ones)

<<Include>>: Functional Decomposition Problem: A function in the original problem statement is too complex to be solvable immediately Solution: Describe the function as the aggregation of a set of simpler functions. The associated use case is decomposed into smaller use cases ManageIncident CreateIncident HandleIncident CloseIncident <<include>>

<<Include>>: Reuse of Existing Functionality Problem: There are already existing functions. How can we reuse them? Solution: The include association from a use case A to a use case B indicates that an instance of the use case A performs all the behavior described in the use case B (“A delegates to B”) Example: The use case “ViewMap” describes behavior that can be used by the use case “OpenIncident” (“ViewMap” is factored out) OpenIncident <<include>> ViewMap AllocateResources <<include>> Base Use Case Supplier Use Case Note: The base case cannot exist alone. It is always called with the supplier use case

<Extend>> Association for Use Cases Problem: The functionality in the original problem statement needs to be extended. Solution: An extend association from a use case A to a use case B indicates that use case B is an extension of use case A. Example: The use case “ReportEmergency” is complete by itself , but can be extended by the use case “ConnectionDown” for a specific scenario in which the user cannot communicate with the dispatcher ReportEmergency FieldOfficer f ConnectionDown <<extend>> Note: The base use case can be executed without the use case extension in extend associations.

Generalization association in use cases Problem: You have common behavior among use cases and want to factor this out. Solution: The generalization association among use cases factors out common behavior. The child use cases inherit the behavior and meaning of the parent use case and add or override some behavior. Example: Consider the use case “ValidateUser”, responsible for verifying the identity of the user. The customer might require two realizations: “CheckPassword” and “CheckFingerprint” ValidateUser CheckPassword CheckFingerprint Parent Case Child Use Case

Identifying Initial Analysis Objects Le v el 1 Top Level Use Case Level 2 Use Cases Le v el 2 Level 3 Use Cases Le v el 3 Operations Le v el 4 B A A and B are called Participating Objects

Use Cases can be used by more than one object Le v el 1 Top Level Use Case Le v el 2 Le v el 2 Level 2 Use Cases Le v el 3 Le v el 3 Le v el 3 Level 3 Use Cases Operations Le v el 4 Le v el 4 A B Participating Objects

Identifying Initial Analysis Objects Identify the participating objects to create the initial analysis object model Maintaining glossary of objects minimizes potential confusion in terminology between users and developers Heuristics Terms the needed clarification (by developer or user) Recurring nouns in use cases Real-world entities and resources that system must track Use cases Data sources or sinks Artifacts with which user interacts Use application domain terms Cross-check Eliminate ambiguity: verify that objects with the same name refer to the same concept Maintain consistency: verify that objects do not refer to the same concept using different names Eliminate objects not involved in any use cases

Identifying Nonfunctional Requirements (FURPS+ Classification Scheme) Quality Requirements Usability Reliability/Dependability Safety Security Survivability Performance Response Time Throughput Availability Accuracy Supportability Adaptability Maintainability Portability Pseudo Requirements Implementation Interface Operations Packaging Legal

Identifying Nonfunctional Requirements Heuristics Use a taxonomy (e.g., FURPS+) to generate checklists Give different checklists to users in appropriate roles Checklists vary depending on application domain

How to Specify a Use Case (Summary) Name of Use Case Actors Description of Actors involved in use case) Entry condition “This use case starts when…” Flow of Events Free form, informal natural language Exit condition “This use cases terminates when…” Exceptions Describe what happens if things go wrong Special Requirements Nonfunctional Requirements, Constraints)

Managing Requirements Elicitation Negotiating specifications Maintaining traceability Tool support

Negotiating Specifications (JAD) Use case modeling is useful in requirements elicitation, but it is not the only activity Requirements have to be identified and negotiated between different stakeholders JAD – Joint Application Design A moderated meeting with all stakeholders participating Users, clients, developers + trained facilitator Leverages group dynamics of face-to-face meetings Developers get to understand application domain Users get to understand potential solution domain tradeoffs

JAD Activities Project definition Research Preparation Facilitator determines objectives and scope of project through interviews with project manager and client Research Facilitator interviews present and future users Facilitator gathers information about application domain Facilitator creates initial high-level use cases Facilitator creates initial list of problems Preparation Facilitator creates Working Document, agenda and presentation materials Facilitator forms team with adequate representation from all stakeholders

JAD Activities Session Final document preparation Facilitator guides team in creating the requirements specification Discover new requirements Classify and organize requirements Prioritize requirements Validate requirements Derive use cases Activities are repeated until closure is achieved Final document preparation Facilitator prepares Final Document Team reviews and approves Final Document

JAD Facilitator Qualifications of JAD facilitator is crucial Must keep the discussion within the scope of the project Discern wants from needs Keep the discussion within the application domain to avoid prescribing requirements that restrict the solution space unnecessarily (pushing specific technology, methodology or language) Mediate disputes before they get out of hand Watch out for political influences and hidden agendas

Maintaining Traceability Traceability the ability to follow the life of a requirement as it is translated into design and then implementation and test cases The system is complete when all requirements can be traced to implementation Traceability also enables developers to uncover the rationale behand certain requirements and design decisions Traceability is harder for nonfunctional requirements Traceability is difficult to maintain manually Need to maintain cross-references between different artifacts (requirements, design documents, code, test plan, user documentation) Need tool support

Tool Support Requirements for Managing Requirements Store requirements in a shared repository Provide multi-user access Automatically create a system specification document from the repository Allow change management Provide traceability throughout the project lifecycle

Summary The requirements process consists of requirements elicitation and analysis. The requirements elicitation activity is different for: Greenfield Engineering, Reengineering, Interface Engineering Scenarios: Great way to establish communication with client Different types of scenarios: As-Is, visionary, evaluation and training Use cases: Abstraction of scenarios Pure functional decomposition is bad: Leads to unmaintainable code Pure object identification is bad: May lead to wrong objects, wrong attributes, wrong methods The key to successful analysis: Start with use cases and then find the participating objects If somebody asks “What is this?”, do not answer right away. Return the question or observe the end user: “What is it used for?” Use case modeling is a part of the requirements elicitation process, not the entire process itself.