Visual Specification and Design of Component-based Slow Intelligence Systems

1
Visual Specification and Design of
Component-basedSlow Intelligence Systems
2
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Applications

3
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Applications

4
Why Slow Intelligence Systems

• The slow intelligence system (SIS) technology
  is a novel technology for the design and/or
  improvement of complex information systems.

5
Characteristics ofComplex Information Systems

• Connected
• Multiple sourced
• Knowledge-based
• Personalized
• Hybrid
• Prodigious

6
Smarter Planet

• We are all now connected - economically,
  technically and socially. Our planet is becoming
  smarter via integration of information scattered
  in many different data sources from the
  sensors, on the web, in our personal devices, in
  documents and in databases, or hidden within
  application programs. Often we need to get
  information from several of these sources to
  complete a task. Examples include healthcare,
  science, the business world and our personal
  lives. (Quoted from Josephine M. Cheng, IBM
  Fellow and Vice President of IBM Research)

7
(courtesy of IBM)
8
Hybrid Intelligence

• While processor speed and storage capacity
  have grown remarkably, the geometric growth in
  user communities, online computer usage, and the
  availability of data is in some ways even more
  remarkable. Hybrid Intelligence offers great
  opportunities. We have to harness this data
  availability to build systems of immense
  potential. While today s large scale systems are
  evolutionarily based on the distributed computing
  technologies envisioned in the 70 s and 80 s,
  sheer scaling has led to many unanticipated
  challenges. (quoted from Alfred Z. Spector, Vice
  President, Research and Special Initiatives,
  Google, USA)

9
Prodigious Hybrid Intelligence Systems

• Users and computers doing more than either could
  individually (quoted from Alfred Z. Spector,
  Google).

10
Characteristics ofComplex Information Systems

• Connected
• Multiple sourced
• Knowledge-based
• Personalized
• Hybrid
• Prodigious
• gt CONSTANTLY CHANGING

11
Challenges in the Design of Complex Information
Systems

• The operating environment, individual/collective
  user behavior and underlying technology base of
  such complex information systems are constantly
  changing.
• There is never a stable and static solution for
  an optimal complex information system.
• There are no general techniques for the design of
  a complex information system that can gradually
  improve and/or optimize its performance over time
  in a changing environment.

12
What is a Slow Intelligence System

• A Slow Intelligence System (SIS) is a
  general-purpose system characterized by being
  able to improve performance over time through a
  process involving enumeration, propagation,
  adaptation, elimination and concentration. A SIS
  is characterized by employing super components,
  i.e., multiple components that can be activated
  either sequentially or in parallel to search for
  better solutions. A SIS continuously learns,
  searches for new solutions and propagates and
  shares its experience with peers.

12
13
The SIS Technology

• This SIS technology consists of the visual
  specification of SIS as a system of super
  components, design principles of the timing
  controller, techniques for incremental
  application system design, SIS development
  framework and the SIS experimental test bed.

14
How can SIS technology help?

• The SIS technology can be applied to design
  and/or modify a complex information system
  capable of improving its performance over time in
  a changing environment. In this presentation we
  will concentrate on visual specification,
  incremental design, development framework, user
  interface, and application to social influence
  analysis.

15
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Application to Social Influence Analysis

16
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time.
• A slow intelligence system is
• a system that (i) solves
• problems by trying different
• solutions, (ii) is context-
• aware to adapt to different
• situations and to propagate
• knowledge, and (iii) may
• not perform well in the
• short run but continuously
• learns to improve its
• performance over time.

17
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration

18
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration
• Propagation

19
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration
• Propagation
• Adaptation

20
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration
• Propagation
• Adaptation
• Elimination

21
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration
• Propagation
• Adaptation
• Elimination
• Concentration

22
Slow Intelligence Systems

• Slow Intelligence Systems are general-purpose
  systems characterized by being able to improve
  performance over time through a process involving
  
• Enumeration
• Propagation
• Adaptation
• Elimination
• Concentration
• Slow Decision Cycle
• to complement Fast
• Decision Cycle

23
Slow Intelligence Systems

• A SIS continuously learns, searches for new
  solutions and propagates and shares its
  experience with other peers.
• From the structural point of view, a SIS is a
  system with multiple decision cycles such that
  actions of slow decision cycle(s) may override
  actions of quick decision cycle(s), resulting in
  poorer performance in the short run but better
  performance in the long-run.
• Timing Controller What decision cycle to take is
  determined by the timing controller.

24
Slow Intelligence Systems

• A SIS continuously learns, searches for new
  solutions and propagates and shares its
  experience with other peers.
• From the structural point of view, a SIS is a
  system with multiple decision cycles such that
  actions of slow decision cycle(s) may override
  actions of quick decision cycle(s), resulting in
  poorer performance in the short run but better
  performance in the long-run.
• Timing Controller What decision cycle to take is
  determined by the timing controller.

25
Slow Intelligence Systems

• A SIS continuously learns, searches for new
  solutions and propagates and shares its
  experience with other peers.
• From the structural point of view, a SIS is a
  system with multiple decision cycles such that
  actions of slow decision cycle(s) may override
  actions of quick decision cycle(s), resulting in
  poorer performance in the short run but better
  performance in the long-run.
• Timing Controller What decision cycle to take is
  determined by the timing controller.

26
Slow Intelligence Systems

• . A Slow Intelligence System is constructed from
  super components, which are the building blocks
  of SIS. Therefore SIS is an advancement over
  component based software systems while a
  component-based software system is constructed
  from software components, in SIS some or all of
  its components are replaced by super components
  capable of improving their performances over time
  in a changing environment.
• There are two types of super components the
  basic building block and advanced building block.

27
Basic Building Block (BBB)
28
Advanced Building Block (ABB)
29
SIS is a component-based systembuilt from BBBs
and ABBs
30
Comparison with Other Approaches
30
30
31
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Applications

32
Visual Specification of SIS
33
Dual visual representations by components and
Petri net
34
Dual visual representations by class diagrams and
sequence diagram
35
Super Components
36
Expanded Petri net based upon super components
37
A component generator for super components
38
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Applications

39
Partial dual visual representation (I-card1,
C-card1) for Product Service Customization
(PSC) system
40
Partial dual visual representation (I-card2,
C-card2) and (I-card3, C-card3)
41
Synthesis of partial visual representations
(I-card1, C-card1), (I-card2, C-card2) and
(I-card3, C-card3) into (I-card1-2-3,C-card1-2-3)
42
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• Timing Controller
• User Interface
• Application to Topic/Trend Detection
• Application to High Dimensional Feature Selection

42
43
Timing Controller

• Three Types of Timing Controllers
• Basic Timing Controller (using switching circuit)
• Advanced Timing Controller (using associative
  memory)
• Recursive Timing Controller (for super
  components)
• Fast Cycle and Slow Cycle
• Any cycle without super component is a fast cycle
• Any cycle with super component is a slow cycle

43
43
44
Basic Timing Controller

• The basic timing controller consists of a latch
  register and a switching circuit
• The latch register has as many cells as there are
  software components. Each cell stores a binary
  number indicating whether to invoke the
  corresponding software component (1) or not (0)
• The latch register outputs this binary vector to
  the switching circuit, which computes another
  binary vector as new input to the latch register,
  to control the next round of software components
  invocation
• This computation cycle repeats itself, until the
  binary vector stored in the latch register
  becomes (0,,0)

44
44
45
Advanced Timing Controller

• The associative memory also receives the binary
  vector from the latch register and uses it to
  search and access an associated binary vector (or
  vectors) as output

• The timing controller with non-deterministic
  associative memory determines the invocation of
  software components by performing additional
  computations on Petri net structure, attributes,
  probability, degree of certainty, fuzzy measures
  or some other means

• In SIS some or all of the software components may
  be super-components. In which case, certain
  cells in the latch register may be associated
  with super-components.

45
45
46
Recursive Timing Controller

• Each cell associated with a super-component can
  be expanded into another latch register. The
  secondary latch register is part of another
  timing controller. In other words, timing
  controllers can be recursively defined when super
  components are present in a system

• The T icon adjacent to a cell associated with a
  super component indicates it can be expanded into
  another timing controller. An equivalent, but
  simpler, notation is to write a T inside this
  cell, which can then be expanded into another
  latch register and its associative memory

46
46
47
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Application to Social Influence Analysis

48
SIS Test Bed

• Initially, to experiment with an application
  system under development, the SIS test bed can be
  employed.

49
SIS Test Bed for Healthcare Systems
50
SIS Framework

• The SIS framework is to test and develop the
  optimized application system based upon the SIS
  technology.

51
SIS Framework
52
SIS Framework

• The Enumerator reads in the specification of
  functional blocks and creates multiple candidate
  components for each functional block. The Tester
  tests all the functional blocks and records their
  performance in the DB. The Eliminator selects the
  best candidate components based upon their
  performance. The Concentrator packs the selected
  candidate components based on dependency
  specifications and generates a generic software
  package. The Transformer is used to transform
  the Concentrator-generated software package to
  target software package that serves specific
  purpose. The Timing Controller is the system
  manager that controls all the above mentioned
  components, telling them when to perform what
  actions.

53
Requirements for SIS Framework
ID Description
R1 Java support
R2 Dynamic lifecycle management of building blocks during runtime
R3 XML message based communications
R4 Code frame generation out of meta-models in a form of class diagram
R5 Scalability of components upto 1,000 10,000 components
R6 Integrated Development Environment (IDE) support
54
Related technologies for SIS Framework

• Eclipse has a big ecosystem for Java based
  software development both the IDE and runtime
  framework
• OSGi is a universal middleware which abstract
  heterogeneous communication protocols, support
  life-cycle management of software component
  during runtime which can be a basis for the SIS
  framework.
• Eclipse EMF provides code generation from the
  meta-model
• Eclipse ECF provides communication for
  point-to-point and publish-and-subscribe for the
  distributed systems
• Semantic self-organization, self-similarity and
  autonomic component model concepts from CASCADAS
  project can be useful to augment the SIS
  approaches.

55
Relationships of related technologies to SIS
56
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Application to Social Influence Analysis

57
SIS User Interface

• The user interface was developed in part based
  upon PIPE tool
• In C-card, places specify messages and
  transitions specify components
• In I-card, components can be simple components or
  super components
• T-card is used to specify test data
• Spec can be saved as PNML (Petri Net Markup
  Language) and XML documents

58
Screen Shot for C-card Design
59
Screen Shot for place editor and message editor
design
60
Screen Shot for I-card Design
61
Screen Shot for T-card Design
62
Outline

• Why Slow Intelligence Systems
• Introduction to SIS
• Visual Specification of SIS
• Incremental Design
• SIS Framework and Test Bed
• User Interface
• Applications

63
Applications

• Social Influence Analysis
• Product and service customization
• Topic/Trend Detection
• High Dimensional Feature Selection
• Personal healthcare
• Emergency Management
• Legacy Systems

64
Discussion

• A framework for knowledge-based software
  engineering.
• Since time is relative, slow intelligence
  systems for some can also be fast for others.
• A slow intelligence system can evolve into a
  fast intelligence system.

65
Further Work

• How to check and maintain consistency in
  incremental application system design
• How to find the critical components in a legacy
  system for replacement by super components

66
QA