Knowledge Management In the Insurance Industry Making Organisational Knowledge Active.

Slides:



Advertisements
Similar presentations
Ch:8 Design Concepts S.W Design should have following quality attribute: Functionality Usability Reliability Performance Supportability (extensibility,
Advertisements

Visual Scripting of XML
Schedule and Effort. Planning Big Project: Waterfall-ish Style 1.Figure out what the project entails Requirements, architecture, design 2.Figure out dependencies.
From Words to Knowledge ORION Active Structure. ORION Active Structure Two Approaches We could separate the process of turning words into knowledge into.
Knowledge Management Using Orion Making Knowledge Active.
Orion Philosophy and Rationale. If it really is structure, what sort of structure is it? We are asserting that it is active dynamic undirected structure.
UNIVERSITY OF JYVÄSKYLÄ Yevgeniy Ivanchenko Yevgeniy Ivanchenko University of Jyväskylä
MS DB Proposal Scott Canaan B. Thomas Golisano College of Computing & Information Sciences.
DIVISION OF QUANTITATIVE METHODS SAMPLE SCENARIO 1.
System Design and Analysis
Lecture 13 Revision IMS Systems Analysis and Design.
Geography 465 Overview Geoprocessing in ArcGIS. MODELING Geoprocessing as modeling.
Orion Overview. We build an internal model of the world, so we can predict future behaviour - we make the model out of active structure so it is interoperable.
Introduction to Systems Analysis and Design
©Ian Sommerville 2004Software Engineering, 7th edition. Chapter 8 Slide 1 Tools of Software Development l 2 types of tools used by software engineers:
XP New Perspectives on Microsoft Access 2002 Tutorial 71 Microsoft Access 2002 Tutorial 7 – Integrating Access With the Web and With Other Programs.
 A data processing system is a combination of machines and people that for a set of inputs produces a defined set of outputs. The inputs and outputs.
Software Design & Development Software Design & Development Programming Types Event Driven Programming Event driven programming Is a type of programming.
TIBCO Designer TIBCO BusinessWorks is a scalable, extensible, and easy to use integration platform that allows you to develop, deploy, and run integration.
CASE Tools And Their Effect On Software Quality Peter Geddis – pxg07u.
A year 1 computer userA year 2 computer userA year 3 computer user Algorithms and programming I can create a series of instructions. I can plan a journey.
Data Mining Techniques
What is R By: Wase Siddiqui. Introduction R is a programming language which is used for statistical computing and graphics. “R is a language and environment.
11 Games and Content Session 4.1. Session Overview  Show how games are made up of program code and content  Find out about the content management system.
Appendix 2 Automated Tools for Systems Development © 2006 ITT Educational Services Inc. SE350 System Analysis for Software Engineers: Unit 2 Slide 1.
Systems Analysis – Analyzing Requirements.  Analyzing requirement stage identifies user information needs and new systems requirements  IS dev team.
CSI315 Web Applications and Technology Overview of Systems Development (342)
Chapter 8 Architecture Analysis. 8 – Architecture Analysis 8.1 Analysis Techniques 8.2 Quantitative Analysis  Performance Views  Performance.
1.eCognition Overview. 1 eCognition eCognition is a knowledge utilisation platform based on Active Knowledge Network technology eCognition covers the.
Active Structure A + B = C. Logic and Numbers IF A + B = C THEN D + E = F.
Goals  Correlation/dependency: The storing of correlations and dependencies  Integration: Methodology for integrating correlated risk distributions into.
© 2001 Business & Information Systems 2/e1 Chapter 8 Personal Productivity and Problem Solving.
Lead Black Slide Powered by DeSiaMore1. 2 Chapter 8 Personal Productivity and Problem Solving.
Copyright 2002 Prentice-Hall, Inc. Chapter 2 Object-Oriented Analysis and Design Modern Systems Analysis and Design Third Edition Jeffrey A. Hoffer Joey.
Presented at: 1998 DFA Seminar July 13-14, 1998 Presented at: 1998 DFA Seminar July 13-14, 1998 lmn Dynamic Financial Analysis: Objectives & Design Gerald.
Database Design and Management CPTG /23/2015Chapter 12 of 38 Functions of a Database Store data Store data School: student records, class schedules,
DEBUGGING. BUG A software bug is an error, flaw, failure, or fault in a computer program or system that causes it to produce an incorrect or unexpected.
1 Introduction to Software Engineering Lecture 1.
CS Data Structures I Chapter 2 Principles of Programming & Software Engineering.
Computational Autonomy. Broadening the Focus Computational Autonomy is seen as a way of enlarging the narrow focus of a program, which carries out one.
SOFTWARE DESIGN. INTRODUCTION There are 3 distinct types of activities in design 1.External design 2.Architectural design 3.Detailed design Architectural.
The Software Development Process
1 CSCD 326 Data Structures I Software Design. 2 The Software Life Cycle 1. Specification 2. Design 3. Risk Analysis 4. Verification 5. Coding 6. Testing.
Connecting with Computer Science2 Objectives Learn how software engineering is used to create applications Learn some of the different software engineering.
CASE (Computer-Aided Software Engineering) Tools Software that is used to support software process activities. Provides software process support by:- –
Software Engineering1  Verification: The software should conform to its specification  Validation: The software should do what the user really requires.
Chapter 4 Automated Tools for Systems Development Modern Systems Analysis and Design Third Edition 4.1.
Review of Parnas’ Criteria for Decomposing Systems into Modules Zheng Wang, Yuan Zhang Michigan State University 04/19/2002.
Knowledge Networks or “The Rise of the Machines”.
Copyright (c) 2014 Pearson Education, Inc. Introduction to DBMS.
V7 Foundation Series Vignette Education Services.
ECognition Model Components. 2. eCognition Model Components1 Model Components Variables and Constants Operators and Functions Links Logical States Messages.
Tupai Business Systems Predictive Models using eCognition TM.
CASE (Computer-Aided Software Engineering) Tools
Some of the utilities associated with the development of programs. These program development tools allow users to write and construct programs that the.
1 INTRODUCTION TO COMPUTER GRAPHICS. Computer Graphics The computer is an information processing machine. It is a tool for storing, manipulating and correlating.
Chapter 7: Modifiability
Modern Systems Analysis and Design Third Edition
Modern Systems Analysis and Design Third Edition
Unified Modeling Language
System Design.
UML dynamic Modeling (Behavior Diagram)
Tools of Software Development
Verification and Validation Unit Testing
Modern Systems Analysis and Design Third Edition
Computer Science Testing.
Modern Systems Analysis and Design Third Edition
Test Case Test case Describes an input Description and an expected output Description. Test case ID Section 1: Before execution Section 2: After execution.
Tutorial 7 – Integrating Access With the Web and With Other Programs
©Ian Sommerville 2004Software Engineering, 7th edition. Chapter 8 Slide 1 Tools of Software Development l 2 types of tools used by software engineers:
Presentation transcript:

Knowledge Management In the Insurance Industry Making Organisational Knowledge Active

Knowledge Is Structure What does that mean? What sort of structure does it imply? A “live” structure can have the flow in it changed, or its elements rearranged to produce a new structure, for a new purpose. The structure contains everything needed to represent the knowledge - nothing is “outside” it. The topology of the structure is modifiable by the structure - it is self-adapting.

ORION Orion is a knowledge utilisation system based on using an active, undirected structure to represent knowledge. Orion covers the full knowledge cycle – creation or capture, marshalling and deployment. It is particularly suited to dynamically changing problems, where its ability to modify its own structure comes to the fore.

Instead of writing programs, turn the knowledge itself into a computing machine Programmatic: if input(a) and input(b) then c=a+b else if input(a) and input(c) then b=c-a else… Example: a + b = c A new approach to knowledge Structure NYK - Not Yet Known

Values flow through the structure in any direction. The EQUALS operator allows for logical control. Example: a + b = c Use the structure itself

The user can turn what they know into a shareable and reusable piece of knowledge by describing relations among objects - the system will use the relations any way it needs to, when it needs to. Users can refine their thinking by observing a model of it in operation. The System as a Thinking Tool Orion

Combining Knowledge Domains Knowledge models from different domains can be combined - each expert operates in their own domain, thinking in ways they are comfortable with. Economy Funding Credit Risk FX Integrated Model

Techniques Knowledge about the problem area is turned into a structure, made up of variables, operators and links. The structure is undirected and extensible, and supports the following problem-solving techniques: Logical and Numerical Analysis Simulation Structure Self-modification Ranges of Values Backtracking Constraint Reasoning

Logical & Numerical Analysis Here are logic and numeric ranges interacting together in a plan.

Using the System

Logic Editor - Used to enter declarative knowledge in textual form. The text is immediately transformed into active model structure.

Network Display The user can: trace the structure of the network observe the values in particular components set and unset values trace the source of inconsistencies debug the network by halting propagation and observing states

Stochastic Editor The user can visualize and manipulate distributions and N- dimensional relations. apply constraints to variables and observe the impact on other variables construct ad hoc distributions and relations

Data Miner One of the shortcomings of existing data mining technologies is that in order to use the findings, the user needs to understand them. The reason is that the technologies used for mining are different to those used in operational systems. With Orion, the same technology is used for both tasks. The Miner actually morphs newly found correlations in the data into an active component of the operational system. The system can start with mined data, then “learn on the run” from new transactions.

Additional tools and facilities Structural debugger Control Panel Simulator Graphing XML Link Time Series Analysis module

Insurance Applications Knowledge Management Dynamic Financial Analysis Simulation of Reinsurance environment Risk analysis Knowledge Representation

Capturing the “Big Picture” Analytic, quantitative This morning’s new developments Databases Experiential “Soft”, qualitative Integrates all forms of knowledge and information.

DFA Model the asset portfolio Model the insurance risks Combine these models easily Simulate strategies in diverse scenarios Refine an effective strategy using machine learning DFA is often set up with very primitive strategy - trouble is, no- one believes its results because they know it is primitive The ability to combine analysis and experience makes the simulation an accurate description of what the organisation would do under adverse circumstances

Network Structure – Bond Valuation The network structure created by the statements entered through the textual logic editor.

Keeping Close to the Meaning Minimising the transformation of knowledge maximises the understanding of what’s happening GetV(Tv, Tmat, N, C) = N * ( GetP( Tv, Tmat) + SIGMA( Tv + 1, Tmat - 1, currT, GetP( CurrT, Tmat)) * C) GetP(t, Tmat) = AtTs%[ Tmat - t ] * exp( -1 * BtTs%[ Tmat - t ] * Rates%[ t ] )

History Based Rates Forecasting The distribution of RateNow The relation between RateNow and NextRate The distribution of NextRate Knowing the rate now, we can use history to forecast the rate in the next time period. The history is mined from the database and stored in operators that are used to propagate distributions. Setting RateNow to a value changes the distribution of NextRate, used by a RANDOM operator to generate a singular number. Full population RateNow set to 3.14 RateNow set to 7 RateNow DISTRIB RELATION DISTRIB NextRate RANDOM

Catastrophe Modelling Using Available Knowledge The distributions of storm severity and of number of storms per year are used to generate values. Once generated, a histogram connecting those with geographic location is used to generate the anticipated damages. The network structure below shows the flow of information. Some of the distributions and histograms used. All can be “massaged” graphically by on-screen dragging.

Creation and Destruction of Assets When the simulation starts, the portfolio has 10 bond invocations maturing on 2001 to After a 20 years simulation, where maturing bonds are destroyed and new bonds are purchased if the cash position allows, the portfolio is left with 3 bonds purchased in 2018, 2019 and Different strategies for asset management can be studied under different rate scenarios.

Reinsurance Simulation Knowledge Model based on Modelling the Impact of Reinsurance on Financial Strength by S. Coutts & T. Thomas Source

Model the Problem Area Programmatic approach Orion Business Domain Household Commercial Property Quota Share treaty Surplus treaty Risk Access treaty Catastrophe treaty The model takes the form of the business structure it represents. It is “alive” all the time, every piece of it is visible and accessible.

No Programming Business Domain Analysts Programmers Programmatic Approach Software Domain Business Domain As is With ORION Software Domain Distortion

Reinsurance Simulation Generated Commercial Property Claims A Surplus treaty Claims paid after the Surplus recoveries Household Risk Excess treaty Claims paid after the Surplus and RiskXs Combine with Household Claims after Quota Share (into Catastrophe treaty) The structure generated for Commercial Property (each column represents a policy) The structure of the reinsurance program is in the form that insurance people understand it – claims flow into a reinsurance treaty, which recovers some of the payments. What remains can be combined and flow into another treaty.

Dramatically Shorter Implementation and Modification C++ Implementation Modification 18 Months / 45 Man-months 3 weeks / 1.5 Man-months 6-8 Weeks 1-3 days Test case Test case results, May/June 2001

Stability of Model C++ Model “The model took 18 months to build....the authors found they wanted to enhance the structure almost on a daily basis....regular enhancements would bring about a collapse of the project...” ORION Model The structure of the model can be quickly and easily changed, because it is a knowledge model, with the structure being part of the knowledge.

Risk Analysis Using Current Knowledge The area of risk being analysed in this example is Reinsurance of Earthquakes This is an area of physical science with relatively little analytical support. Part of the model is to update its knowledge from free text.

Earthquake Knowledge Attenuation Greece Info (GIS) Intensity/ Damage Acceleration attenuation based on magnitude, distance and local site conditions Find distance between site and epicentre, local conditions, etc. Relations between acceleration, intensity and damage ratio Earthquake Event Frequency/ Amplification Relations between magnitude and frequency, building type, number of floors and natural frequency The model is built out of the same variables, operators, links as the knowledge structures being used to analyse the text Lloyds List

Analytic Component - Wave Attenuation A formula for attenuation (note left hand side)

Propagation An example of one of many value propagation paths, from event magnitude to damage ratio. Magnitude – Event submodel Attenuation submodel  SAh Experiential submodel  intensity Experiential submodel  Damage Ratio

Experiential Structure The Sauter curves, linking Intensity, Building Type and Damage Ratio, are loaded into distributions and relations. These curves represent distributions that respond to other factors - earthquake codes, locales. Wood Reinforced Concrete

Accept Anything Here is KM at work. Need to accept: Different forms of knowledge - analytic, experiential Different timescales for validity - yesterday, last month, always Different sources - journals, consultants, suppliers, customers Integrate it, iron out the inconsistencies, if necessary change it on the run - manage it like any other resource.

Orion can handle a wide range of problem areas. Its power comes from its simple form - variables, operators, links - and the fact that it is easy to combine structures which do not have a beginning or an end.

What Can You Put in a Model Numbers Strings Logicals Lists Objects For loops RANDOM Generators Distributions and Relations Bayesian logic New operators

How Is It Different? It is based on: Connection Influences can only take place through connections. Activity Active structures carry out processing of the information that flows through them directly, and can adapt themselves to each other. Visibility All the states are visible all the time - no structure is hidden or built on a stack.

It Sounds Too Complex Complexity appears quite quickly. A + B = C is static. X = SUM(List) is already dynamic. “The earthquake struck Sao Paulo last Thursday.” is just more dynamic. Knowledge in an organisation is dynamic - it is just that humans handle it without a second thought. For a system to handle it well, the system needs some similar properties - activity, connectivity, visibility.

Is It Really So Different We are asserting that knowledge can only be captured in active structure - structure that is capable of adapting itself to its environment. Efforts at capturing knowledge in static structure founder on two reefs - the pieces of structure will not fit together statically, and an algorithm that could manage their combination would be more complex than the combination of the pieces, and is thus both unmanageable and unreachable from text. Active structure avoids both problems - the pieces adapt to each other, and the behaviour of the combination is managed by the interaction of the pieces.