Generating Annotated Behavior Models From EndUser Scenarios

1
Generating Annotated Behavior Models From
End-User Scenarios

• By
• Christophe Damas
• Bernard Lambeau
• Pierre Dupont
• Axel Van Lamsweerds

Presented By Rajesh Dilip Bathula
2
Aim

• To develop tool-supported techniques that
  overcome the problems like
  
  
• 1. Additional input like state assertions or
  flowcharts on episodes which are difficult to be
  used by non-expert end-users.
• 2.Validation of the generated state machines,
  which is difficult for analysts.
• in synthesizing behavior models inductively
  from scenarios.

3
Introduction

• Scenarios
• Labeled Transition System (LTS)
• Message Sequence Charts (MSC)
• Agent
• Fluents

4
Basic Idea

• A global LTS for the entire system is synthesized
  first, this covers all scenario examples.
• The procedure is incremental on training
  examples.
• The synthesized system LTS is then projected on
  local LTS for each system component.
• For model validation by analysts, the tool
  generates state invariants that decorate the
  nodes of the local LTS.

5
Example

• A simple train system is composed of three
  agents
• i. Train Controller
• ii. Train actuator/sensor
• iii. Passengers

6
Generating annotated state machines from end-user
scenarios
7
3 Steps

• Step 1 Submitting an initial set of positive and
  negative scenarios.
• Step 2 Generating scenario questions and
  synthesizing agent LTS.
• Step 3 Generating state invariants to document
  the generated state machines.

8
Step 1 3 positive and 1 negative scenarios
9
Step 2
10
Step 2 (contd)
11
Step 2 is attained by

• i. LTS synthesis as a Grammar Induction
  problem using RPNI algorithm.
• ii. Interactive synthesis of the system LTS
  using the Merge function.
• iii. Projecting the global system LTS on
  agents.

12
Step3 a)before decoration
b)decorated with state variables
13
Step 3 is attained by

• Decorating states with Fluents.
• Forming the State Invariant.

14
Case studyMine Pump Control System

• This tool supported technique is implemented here
  to avoid the problem of explosion in the mine due
  to over level of methane gas.

15
Conclusion

• A new approach has been presented for
  synthesizing behavior models from simple forms of
  scenarios..
• A new fix point algorithm has been descried for
  generating state invariants on labeled transition
  systems so as to make such models easier to
  understand and validate-especially as they are
  automatically generated from scenarios.

16

• Thank you!