Agent-Oriented Software Engineering

Transcription

1 Agent-riented Software Engineering Nick Jennings Dept of Electronics and Computer Science University of Southampton, UK.

2 Software Development is Difficult ne of most complex construction task humans undertake Computer science is the first engineering discipline ever in which the complexity of the objects created is limited by the skill of the creator and not limited by the strength of the raw materials. If steel beams were infinitely strong and couldn t ever bend no matter what you did, then skyscrapers could be as complicated as computers. Brian K. Reid True whatever models and techniques are applied the essential complexity of software Fred Brooks Software engineering provides models & techniques that make it easier to handle this essential complexity 2

3 Software Development is Getting Harder Shorter development lifecycles More ambitious requirements Less certain requirements Greater scope for change More challenging environments Greater dynamism Greater openness 3

4 Software Engineering: Continually Playing Catch Up Better Models components design patterns software architectures Better Processes light methods heavier methods interacting agents ur ability to imagine complex applications will always exceed our ability to develop them Grady Booch 4

5 The Adequacy Hypothesis Agent-oriented approaches can enhance our ability to model, design and build complex distributed software systems. 5

6 Talk utline I. The Essence of Agent-Based Computing II. III. IV. The Case for Agent-riented Software Engineering Potential Drawbacks Conclusions 6

7 Talk utline I. The Essence of Agent-Based Computing II. III. IV. The Case for Agent-riented Software Engineering Potential Drawbacks Conclusions The intolerable wrestle with words and meanings. T. S. Eliot 7

8 Agent encapsulated computer system, situated in some environment, and capable of flexible autonomous action in that environment in order to meet its design objectives (Wooldridge) 8

9 Agent encapsulated computer system, situated in some environment, and capable of flexible autonomous action in that environment in order to meet its design objectives (Wooldridge) control over internal state and over own behaviour 9

10 Agent encapsulated computer system, situated in some environment, and capable of flexible autonomous action in that environment in order to meet its design objectives (Wooldridge) control over internal state and over own behaviour experiences environment through sensors and acts through effectors 10

11 Agent encapsulated computer system, situated in some environment, and capable of flexible autonomous action in that environment in order to meet its design objectives (Wooldridge) control over internal state and over own behaviour experiences environment through sensors and acts through effectors reactive: respond in timely fashion to environmental change proactive: act in anticipation of future goals 11

12 Definitional Malaise My guess is that object-oriented programming will be what structured programming was in the 1970s. Everybody will be in favour of it. Every manufacturer will promote his product as supporting it. Every manager will pay lip service to it. Every programmer will practice it (differently). And no one will know just what it is. (Rentsch, 82) My guess is that agent-based computing will be what objectoriented programming was in the 1980s. Everybody will be in favour of it. Every manufacturer will promote his product as supporting it. Every manager will pay lip service to it. Every programmer will practice it (differently). And no one will know just what it is. (Jennings, 00) 12

13 Multiple Agents In most cases, single agent is insufficient no such thing as a single agent system (!?) multiple agents are the norm, to represent: natural decentralisation multiple loci of control multiple perspectives competing interests 13

14 Agent Interactions Interaction between agents is inevitable to achieve individual objectives, to manage interdependencies Conceptualised as taking place at knowledge-level which goals, at what time, by whom, what for Flexible run-time initiation and responses cf. design-time, hard-wired nature of extant approaches paradigm shift from previous perceptions of computational interaction 14

15 Agent-Based Manufacturing Control Part B s Part C s 3 rders A: 2 ; B: 3 ; C: 8 1 3, 4 7, 8 2 Part A s 2 6 1, Redo Buffer 6 15

16 Agent-Based Manufacturing Control 4 Part B s 3 Part C s 3 rders A: 1 ; B: 4 ; C: 7 1 3, 4 7, Part A s 2 6 1, Redo Buffer 6 16

17 Agent-Based Manufacturing Control 4 Part B s 3 4 Part C s , 4 7, Part A s 1, Redo Buffer 6 17

18 Agent-Based Manufacturing Control Part B s Part C s 4 1 3, 4 7, Part A s 1, Redo 2 Buffer 6 18

19 Agent-Based Manufacturing Control Part B s Part C s 1 3, 4 7, Part A s 4 5 1, Redo 2 Buffer 6 19

20 Agent-Based Manufacturing Control Part B s Part C s 1 3, 4 7, Part A s , Redo Buffer 6 20

21 Agent-Based Manufacturing Control Part B s Part C s 7 1 3, 4 7, Part A s , Redo Buffer 6 21

22 Agent-Based Manufacturing Control Part B s Part C s 7 1 3, 4 7, Part A s 5 5 1, Redo Buffer 6 22

23 Agent-Based Manufacturing Control Part B s Part C s 7 1 3, 4 7, 8 Part A s , Redo Buffer 6 23

24 Agent-Based Manufacturing Control Part B s Part C s 8 1 3, 4 7, Part A s , Redo Buffer 6 24

25 Agent-Based Manufacturing Control Part B s Part C s 8 1 3, 4 7, Part A s 1, Redo Buffer

26 Agent-Based Manufacturing Control Part B s Part C s 8 1 3, 4 7, Part A s 9 9 1, Redo Buffer

27 Agent-Based Manufacturing Control Part B s Part C s 1 3, 4 7, Part A s , Redo Buffer

28 Agent-Based Manufacturing Control Part B s Part C s 1 3, 4 7, Part A s 1, Redo Buffer

29 rganisations Agents act/interact to achieve objectives: on behalf of individuals/companies part of a wider problem solving initiative underlying organisational relationship between the agents 29

30 rganisations This organisational context: influences agents behaviour relationships need to be made explicit peers teams, coalitions authority relationships is subject to ongoing change provide computational apparatus for creating, maintaining and disbanding structures 30

31 A Canonical View Interactions Agent rganisational relationships Sphere of influence Environment (see also: Castelfranchi, Ferber, Gasser, Lesser,..) 31

32 Talk utline I. The Essence of Agent-Based Computing II. III. IV. The Case for Agent-riented Software Engineering Potential Drawbacks Conclusions 32

33 Making the Case: Quantitatively 50 Software Metrics Productivity Reliability Maintainability bjects Agents There are 3 kinds of lies: lies, damned lies and statistics Disraeli 33

34 Making the Case: Qualitatively 34

35 Making the Case: Qualitatively Software techniques for tackling complexity 35

36 Tackling Complexity Decomposition Abstraction rganisation 36

37 Making the Case: Qualitatively Software techniques for tackling complexity Nature of complex systems 37

38 Complex Systems Complexity takes form of hierarchy not a control hierarchy collection of related sub-systems at different levels of abstraction Can distinguish between interactions among subsystems and interactions within sub-systems latter more frequent & predictable: nearly decomposable systems Arbitrary choice about which components are primitive (Herb Simon) Systems that support evolutionary growth develop more quickly than those that do not: stable intermediate forms 38

39 Making the Case: Qualitatively Software techniques for tackling complexity Agent-based computing Nature of complex systems 39

40 Making the Case: Qualitatively Software techniques for tackling complexity Degree of Match Agent-based computing Nature of complex systems 40

41 The Match Process 1. Show agent-oriented decomposition is effective way of partitioning problem space of complex system 2. Show key abstractions of agent-oriented mindset are natural means of modelling complex systems 41

42 The Match Process 1. Show agent-oriented decomposition is effective way of partitioning problem space of complex system 2. Show key abstractions of agent-oriented mindset are natural means of modelling complex systems 42

43 Decomposition: Agents In terms of entities that have: own persistent thread of control (active: say go ) control over their own destiny (autonomous: say no ) Makes engineering of complex systems easier: natural representation of multiple loci of control real systems have no top (Meyer) allows competing objectives to be represented and reconciled in context sensitive fashion 43

44 Decomposition: Interactions Agents make decisions about nature & scope of interactions at run time Makes engineering of complex systems easier: unexpected interaction is expected not all interactions need be set at design time simplified management of control relationships between components coordination occurs on as-needed basis between continuously active entities 44

45 The Match Process 1. Show agent-oriented decomposition is effective way of partitioning problem space of complex system 2. Show key abstractions of agent-oriented mindset are natural means of modelling complex systems 45

46 Suitability of Abstractions Design is about having right models In software, minimise gap between units of analysis and constructs of solution paradigm techniques natural way of modelling world 46

47 Complex System Agent-Based System Sub-systems Sub-system components Interactions between sub-systems and sub-system components Relationships between sub-systems and sub-system components 47

48 Complex System Agent-Based System Sub-systems Agent organisations Sub-system components Interactions between sub-systems and sub-system components Relationships between sub-systems and sub-system components 48

49 Complex System Agent-Based System Sub-systems Agent organisations Sub-system components Agents Interactions between sub-systems and sub-system components Relationships between sub-systems and sub-system components 49

50 Complex System Agent-Based System Sub-systems Agent organisations Sub-system components Agents Interactions between sub-systems and sub-system components: at any given level of abstraction, find meaningful collections of entities that collaborate to achieve some higher level view (Booch) Relationships between sub-systems and sub-system components cooperating to achieve common objectives coordinating their actions negotiating to resolve conflicts 50

51 Complex System Agent-Based System Sub-systems Agent organisations Sub-system components Agents Interactions between sub-systems and sub-system components Relationships between sub-systems and sub-system components - change over time - treat collections as single coherent unit cooperating to achieve common objectives coordinating their actions negotiating to resolve conflicts Explicit mechanisms for representing & managing organisational relationships Structures for modelling collectives 51

52 The Adequacy Hypothesis Agent-oriented approaches can enhance our ability to model, design and build complex distributed software systems. 52

53 The Establishment Hypothesis As well as being suitable for designing and building complex systems, agents will succeed as a software engineering paradigm [NB: will be complementary to existing software models like, patterns, components, ] 53

54 Agents Consistent with Trends in Software Engineering Conceptual basis rooted in problem domain world contains autonomous entities that interact to get things done 54

55 Agents Consistent with Trends in Software Engineering Conceptual basis rooted in problem domain Increasing localisation and encapsulation apply to control, as well as state and behaviour 55

56 Agents Consistent with Trends in Software Engineering Conceptual basis rooted in problem domain Increasing localisation and encapsulation Greater support for re-use of designs and programs whole sub-system components (cf. components, patterns) e.g. agent architectures, system structures flexible interactions (cf. patterns, architectures) e.g. contract net protocol, auction protocols 56

57 Agents Support System Development by Synthesis An agent is a stable intermediate form able to operate to achieve its objectives and interact with others in flexible ways construct system by bringing agents together and watching overall functionality emerge from their interplay well suited to developments in: open systems (e.g. Internet) e-commerce 57

58 II. Talk utline I. The Essence of Agent-Based Computing The Case for Agent-riented Software Engineering III. IV. Potential Drawbacks Conclusions The moment we want to believe something, we suddenly see all the arguments for it, and become blind to the arguments against it George Bernard Shaw 58

59 Isn t autonomous software and making decisions about interactions at run-time a recipe for disaster? 59

60 Yes If expect to just throw agents together and have a coherent and efficient system. Sometimes this is exactly what is desired Simulation software Also the default for synthesised systems No If engineer system appropriately Interaction Engineering rganisation Engineering 60

61 BT s Provide Customer Quote Process (Jennings et al.) Provide Quote Customer Service Agents Producer ServiceName Consumer 61

62 BT s Provide Customer Quote Process (Jennings et al.) Provide Quote Customer Service Agents VetCustomer Producer ServiceName Consumer A B C D... n Customer Vetting Agents 62

63 BT s Provide Customer Quote Process (Jennings et al.) Provide Quote Cost&DesignNetwork Customer Service Agents Design Agents ProvideLegalAdvice Legal Agents VetCustomer Producer ServiceName Consumer A B C D... n Customer Vetting Agents 63

64 BT s Provide Customer Quote Process (Jennings et al.) Provide Quote Cost&DesignNetwork Customer Service Agents Design Agents Survey Agents SurveyPremises ProvideLegalAdvice Legal Agents VetCustomer Producer ServiceName Consumer A B C D... n Customer Vetting Agents 64

65 Interaction Engineering: Service Provisioning = Negotiation Two agents must agree about conditions under which service will be executed price quality start and end times Need to design appropriate negotiation protocol 65

66 The Vet Customer Negotiation Is a 1:many negotiation ne buyer, many sellers Structured as a reverse auction Efficient means for allowing agent to quickly find partner with highest valuations Multi-dimensional English (open cry) auction (Vulkan and Jennings, 00) 66

67 Multi-Dimensional English Auction Initiation Buyer announces declared utility function (U) maximum price will pay (P) minimum acceptable price time will wait between offers (T) minimum percentage increase that next offer must exceed (X) can lie about U and P Auction Sellers submit offers (on all dimensions) can lie and speculate ffer accepted if: protocol has not terminated offer exceeds last accepted one by X% Buyer makes acceptable bid public Terminates T seconds after last acceptable offer 67

68 Analysis of the Protocol Can prove that: negotiation will terminate no point in buyer lying about U and P truth telling is dominant strategy no point in seller speculating about competitors bid X% more than current price, up to reservation level is the dominant strategy in buyer s best interest to use suggested protocol no other protocol (either auction or direct negotiation) can yield a better result for it 68

69 rganisation Engineering Survey Dept is part of Design Division Have common goals and objectives Design agents have degree of power over survey agents Relationship explicitly represented in both agents Survey agents still autonomous though! Impact of organisational relationship Negotiation is cooperative in nature Don t reject requests unless cannot meet them Quickly move to region of agreement Agreement produced if one exists Search for win-win negotiation solutions Using concept of fuzzy similarity (Faratin et al., 2000) 69

70 II. III. Talk utline I. The Essence of Agent-Based Computing IV. Promise of Agent-riented Software Engineering Potential Drawbacks Conclusions We seldom attribute good sense, except with those who agree with us. Duc de la Rouchefoucauld 70

71 Promise Agent-based computing can: provide powerful metaphors, concepts and techniques for conceptualising, designing and implementing complex distributed systems 71

72 Realising the Promise Practical methodologies For analysing and designing agent-based systems Should be useable by practitioners Industrial strength toolkits So that don t have to start from scratch Re-useable know-how and technologies Libraries of interactions protocols rganisational patterns 72

73 Acknowledgements Cristiano Castelfranchi Ed Durfee Les Gasser Carl Hewitt Mike Huhns Vic Lesser Joerg MÜller Simon Parsons Van Parunak Carles Sierra Mike Wooldridge Franco Zambonelli 73

74 References P. Faratin, C. Sierra and N. R. Jennings (2000) Using similarity criteria to make negotiation trade-offs Proc. 4th Int. Conf on Multi-Agent Systems, Boston, N.R. Jennings (2000) n Agent-riented Software Engineering Artificial Intelligence 117 (2) N. R. Jennings, P. Faratin, T. J. Norman, P. Brien and B. dgers (2000) Autonomous agents for business process management Int. J. of Applied Artificial Intelligence 14 (2) N. Vulkan and N. R. Jennings (2000) Efficient mechanisms for the supply of services in multi-agent environments Int J. of Decision Support Systems 28 (1-2)