Formalising Human Computer Interaction in DSS Domain

Size: px

Start display at page:

Download "Formalising Human Computer Interaction in DSS Domain"

Adela Bradley
6 years ago
Views:

University College Cork Cork, Ireland SUPPORTED BY: 1.

1 Formalising Human Computer Interaction in DSS Domain Ọdẹtúnjí Ajadi, Ọdẹjọbí Cork Constraint Computation Center (4C) Computer Science Department University College Cork Cork, Ireland SUPPORTED BY: 1. Science Foundation Ireland Grant 05/IN/I Marie Curie Grant MTKD-CT April 2009 HCI-in-DSS 1

2 This presentation Describes DSS research at 4C (CREEDS) Discusses my current work on HCI in the context of the CREEDS project Highlights future directions 21 April 2009 HCI-in-DSS 2

3 The 4C A research centre at the University College Cork Under Computer Science Department Headed by Prof. E. C. Freuder About 50 academic researchers from about 11 countries About 15 PhD students Research funders include: Science Fundation Ireland Microsoft Marie Curie Many landmark achievments in constraint based modelling 21 April 2009 HCI-in-DSS 3

4 The CREEDS project Constraint Reasoning Extended to Enhance Decision Support Headed by Richard J. Wallace Involves about 10 academic researchers all based at 4C Aims 1) extend the scope of constraint based reasoning research activities into the field of decision support system 2) devise new methods, guided by theoretical treatment of decision making, to support decision making 21 April 2009 HCI-in-DSS 4

5 My work in the CREEDS project Human Computer Intercation in DSS domain Suicide suceptibility prediction as case study Constraint reasoning, ontology and fuzzy logic as modelling tools Formal methods as design framework 21 April 2009 HCI-in-DSS 5

6 The case study: Given the clinical and behavioural profile of an individual, can we develop a system that can support human decision makers to accurately predict susceptibility to commiting suicide or other life-threatening acts? 21 April 2009 HCI-in-DSS 6

7 The environment Developers System Users Knowledge Engineers Experts {psychiatrists psychologists} Software Engineers DSS Medical Doctors Nurses Social workers Educators Assumptions Assumptions Domain knowledge is precise Domain concept and ideas are definite HCI should be considered after development System will assist in decision making Users have same or similar problem solving skills They know what we want They understand our concept of problem They understand how we will use the system Their product will help us get better diagnosis 21 April 2009 HCI-in-DSS 7

8 The problem The design process cannot be verified for correctness Development paradigm does not provide proof of correctness Concepts are ambigouse and depend on context Solution process are vague (rule of the thumb) The domain knowledge as encoded in the DSS and users action are not harmonise Rely on the ability of users to understand and interpret information provided via GUI The domain knowledge are often cyclic and fuzzy hence difficult to represent in the from of a tree (Park &Kim, 1995). 21 April 2009 HCI-in-DSS 8

9 Research efforts Four main areas: Science of Human Decision Making: philosophy of information human cognitive modelling (e.g. cognitive maps) Decision Process Engineering: Design, coding, evaluation, implementation Human Computer Interaction: User behaviour during interaction with computer Evolution of users' concept of solution Formal Methods: Precise mathematical specification Definition of system activities Proof of completeness and soundness 21 April 2009 HCI-in-DSS 9

10 Objectives 1)Identify and analyse the HCI requirements for DSS in a selected medical domain 2)Appraise constraint based and related approaches for realising a specification of 1) 3) Design the development framework for 2) 21 April 2009 HCI-in-DSS 10

11 Methodology Domain knowledge design and encoding: Fuzzy logic Constraint satisfaction Ontology Formal methods: CASL (Common Algebra Specification Language) Our extension of CASL with constraint modelling Users action modelling (HCI context): That facilitates the capturing of domain specific: Ideas, Metaphor, Models, Key concepts 21 April 2009 HCI-in-DSS 11

12 The domain knowledge encoding Cues: Factual: Facts about the individual Physical obervations, family history, community history, racial background, gendar, education Expressed:Individuals views of self in society Feeling expressesed by bevarioural activities: e.g. Verbally: veiws and opinion: experience: about recent happenings Derived: Infromation obtained via theoretical treathment of factural and expressesed cues. 21 April 2009 HCI-in-DSS 12

13 Information type Physical experience Pain/No-Pain Sadness, Fear, Happiness Joy, pride, bliss, jealousy Distress, Disgust, Resentment Physiological consequence Heart rate, blood rate, temperature Heart rate, blood rate, blood pressure EEG in left frontal cortex EEG in right frontal cortex These instrumental measurements must be accompanied by decision makers' judgement 21 April 2009 HCI-in-DSS 13

14 Theory G. Kaufmann(1985) Locus of decision making High Low Low Linguistic Visual images Problem novelty (decision makers skill) Overt exploratory activities High Consequence for HCI design is that The higher the skill of the user the better the use of overt exploratory activities The lower the skill of the user the better the use of linguistic information 21 April 2009 HCI-in-DSS 14

15 Preference specification si sj Fuzzy relations Decision variables C = 1.0 if si is definitely preferred to sj 0.5<c<1.0 if si is slightly preferred to sj D(si,sj)=c= C = 0.5 C < 0.5 C = 0.0 if si is not preferred to sj if sj is slightly preferred to si if sj is definitely preferred to si 21 April 2009 HCI-in-DSS 15

16 Constraint reasoning Si Si Si Si Si Si Si Si Si Si Si Si Si Sj Sk Sl Decision variables Si Si Si Sm Constraint satisfaction model Ui UiUi Ui UiUj Ui Ui Uk Ui UiUl Ui Ui Um HCI actions Linear Constraint: within same context Non-Linear Constraint: Over more than one contexts 21 April 2009 HCI-in-DSS 16

17 HCI modelling Allow users to manipulate decision information using actions provided by the system Adjust to users' ability to understand and interpret decision actions in different contexts Users activities Scanning: Browsing the interface Identification: Focusing on part of the interface that is of interest Classification:Assessing of areas of interest Decision: Setting preferences and initials processing Interaction: Iterating actions to refine decision outcome 21 April 2009 HCI-in-DSS 17

18 Hypothesis It is possible to describe the important activities users in DSS domain It is possible to use use formal methods to model a description of users activities Why formal methods? Produces a precise and systematic modelling of users' actions Facilitates a means of proofing that the model meets its requirement Supports the process of mapping model into an appropriate implementation 21 April 2009 HCI-in-DSS 18

19 CASL CASL: Common Algebraic Specification Language Formal method like Z, B or π-calculus Extension of CSP (Communicating Sequential Processes) Combines process algebra with concept of loose specification (uses 2-step semantics) Used mainly in formal system specification and testing Robust and easy to use and intend Include software for modelling and proof checking 21 April 2009 HCI-in-DSS 19

20 Extensions of CASL Higher order algebraical extension (e.g. HASCASL, CoCASL) Reactive Extensions (CASL-LTL, SB-CASL, CSP-CASL) Structural extension (HET-CASL) All these deal with the modification of CASL implementation algebra OR the structure of its process 21 April 2009 HCI-in-DSS 20

21 Our CASL extension We replaced: Data part of CASL with fuzzy constraint terms AND Process part with constraint satisfaction modelling Our extension deals with the modification of CASL implementation algebra AND the structure of its process This allows us to: Implement recursive processes where fuzzy terms are used to describe communication 21 April 2009 HCI-in-DSS 21

22 Modelling issues We expect the semantic model in our CASL to: First-step-semantic: to generate a class of model for the data part Second-step-semantic: refines the denotations and assigns meanings to the process part 21 April 2009 HCI-in-DSS 22

23 Formalisation Fuzzy constraint term Fc= X,C,O X => Finite set of variables C => Finite set of Constraint Over variables O => Domain of X where C is valid (Contextualise X) Fuzzy regular grammar Fc= N,T, S,P N => Finite set of non-terminal (partial decisions) T => Finite set of terminals (fuzzy constraint terms) S => S є N starting symbol P => Finite set of fuzzy production (transition based on degree of truth) 21 April 2009 HCI-in-DSS 23

24 Sort types UserSorts: source :t r satisfied: Χ(t,r) input :r repr:ω The transformation t to r, requested by the user; satisfies constraint Χ; and implies repr, in context Ω Issues of ontology ActionsSorts engage :a1 a2 buffer :D(engage) trans : T(D()) InterfaceSorts bind :s1 s2 buffer :R(bind) trans :H(buffer) Issue of GUI ProcessSorts engage :p1 p2 satisfied :P(p1,p2) trans :G(P()) Issues of Constraint grammar Issues of Fuzzy constraint reasoning 21 April 2009 HCI-in-DSS 24

25 Looking forward Completion: Complete the CASL formalisation and proof system Publication: hope to publish the work in conference, (next year), Journal (when idea become matured) Implementation: Extend CASL software to accommodate our ideas 21 April 2009 HCI-in-DSS 25

26 HELP!! Suggestions? Education? Questions? 21 April 2009 HCI-in-DSS 26

A FORMAL METHOD FOR MAPPING SOFTWARE ENGINEERING PRACTICES TO ESSENCE

A FORMAL METHOD FOR MAPPING SOFTWARE ENGINEERING PRACTICES TO ESSENCE Murat Pasa Uysal Department of Management Information Systems, Başkent University, Ankara, Turkey ABSTRACT Essence Framework (EF) aims