Skip to content

Latest commit

 

History

History
69 lines (52 loc) · 2.5 KB

README.md

File metadata and controls

69 lines (52 loc) · 2.5 KB
Raw

detparaminf - Parametrised Inference

detparaminf is a prototype of a learning algorithm that allows the inference of a regular languages with "parametrised symbols". It is based on the lessons learned with James tool and is designed as an extension on the blue-fringe algorithm.

The key differences with James tool are that detparaminf:

  • Is sound (it fits all input data).
  • It works on grammar level (it can be used for test generation but it does not currently support automatic instrumentation and, thus, it does not assume that the system being modelled is a Web Service).

Usage example

We can use detparaminf to infer a diagram from the following traces:

  • Positive traces:
  • Trace 1:
    • B = createBottle()
    • G = createGlass()
    • openBottle(B)
    • fillGlassUsingBottle(G, B)
  • Trace 2:
    • B = createBottle()
    • G = createGlass()
    • openBottle(B)
    • closeBottle(B)
    • openBottle(B)
    • fillGlassUsingBottle(G, B)
  • Negative traces:
  • Trace 3:
    • B1 = createBottle()
    • B2 = createBottle()
    • G = createGlass()
    • openBottle(B1)
    • fillGlassUsingBottle(G, B2)
  • Trace 4:
    • B = createBottle()
    • G = createGlass()
    • openBottle(B)
    • closeBottle(B)
    • fillGlassUsingBottle(G, B)

For that we first input the traces in the internal format. We can do it from ghci as follows:

import DetParamInf.ParamLang
let traces = ParamSample { posTraces = [[PCall "createBottle" [], PCall "createGlass" [], PCall "openBottle" [0], PCall "fillGlassUsingBottle" [1, 0]], [PCall "createBottle" [], PCall "createGlass" [], PCall "openBottle" [0], PCall "closeBottle" [0], PCall "openBottle" [0], PCall "fillGlassUsingBottle" [1, 0]]], negTraces = [[PCall "createBottle" [], PCall "createBottle" [], PCall "createGlass" [], PCall "openBottle" [0], PCall "fillGlassUsingBottle" [2, 1]], [PCall "createBottle" [], PCall "createGlass" [], PCall "openBottle" [0], PCall "closeBottle" [0], PCall "fillGlassUsingBottle" [1, 0]]] }

The inference algorithm is triggered by the function detParamInf in the DetParamInf module:

import DetParamInf.DetParamInf
let dfsm = detParamInf traces

We can now visualise the resulting machine as follows:

import DetParamInf.DFSMVisualise 
visualise dfsm

This will produce a diagram like the following:

Diagram produced by detparaminf

The module Inference contains some functions to use oracles and some toy examples.