SMART KNOWLEDGE DISCOVERY FOR AGILE POLICY DELIVERY - Ubiquitous Computing and Communication Journal

Document Sample
SMART KNOWLEDGE DISCOVERY FOR AGILE POLICY DELIVERY - Ubiquitous Computing and Communication Journal Powered By Docstoc

                                 Margarete Donovang-Kuhlisch, Mike Small
                                   IBM Deutschland GmbH, IBM UK Ltd.

               Counter insurgency, peace keeping and peace support operations require
               flexibility, efficiency and effectiveness at all levels of command. As examples of
               the complex endeavours that today’s commanders are confronted with, these
               operations are characterized by the need for coalitions to act as a value network: a
               heterogeneous set of mutually supporting governmental and non-governmental
               organisations committed to a common purpose. Doctrine, policy, ways of working
               and the Rules of Engagement (RoE), form the “business rules” which characterise
               the environment of such a value network. The people, organizations, processes
               and systems which are ‘configured’ to deliver services according to these business
               rules must be continuously adaptable to ensure the most effective response to the
               threats, risks and opportunities that confront commanders.Discovery – definition
               – enforcement describes the business-rules lifecycle of any value network. This
               paper describes the business architecture and the existing and emerging
               technologies which enable effect-oriented adaptation, and describes the benefits to
               coalition operations of a structured approach to discovering, modelling,
               implementing and monitoring the business rules which drive coalition operations.

               Keywords: knowledge management, policy adaptation, smart work, business


     The effects-based approach to operations
(EBAO, illustrated in Figure 1) of complex coalition
engagements relies on the capability to manage
knowledge. Knowledge is an intangible asset which
involves delivering the right information to the right
individuals (information superiority), but goes fur-        . Figure 1: Operations Intent and Effects Model [1]
ther, recognising that an individuals competence, the
structure and composition of teams, and the enablers             As any other business, military coalitions are
of the trusting relationships that make the teams           moving towards an era of dynamic, rapidly adapting
effective, must all be managed to deliver mutually          value networks. Characterized as loosely coupled
beneficial collaboration and information sharing.           (formed units deployed for short periods of time),
Knowledge management in a coalition operation               organized in complex configurations of units (mutu-
truly encompasses all lines of development. It re-          ally supporting but with a wide range of individual
quires the synergetic alignment of information sys-         characteristics, and capabilities), businesses and
tems, business-process development, the skills and          organisations engage in mutually beneficial activities
competencies of individuals, the selection and train-       in order to achieve the commanders’ shared intent.
ing of teams, and management of the formal and
informal organisational structures which support               In such an environment, it is not only necessary to
groups and communities of interest as they continu-         share information based on common semantics to
ously adapt in a rapidly changing military-political        achieve shared situational awareness, but also to
scenario.                                                   exploit that awareness to adapt the rules that govern
                                                            the sharing of information. Data sharing policies in
                                                            the value network should be dependent on the cur-
                                                            rent situation and should adapt to events as they
                                                            emerge. Within boundaries, regulations need to be
configurable within the digital coalition platform                The digital coalition needs new capabilities in
(see Error: Reference source not found).                    four categories all of which depend on the
                                                            understanding of the meaning of the data and the
                                                            ability to communicate via a common language:

                                                                -    new intelligence for insight as data is ex-
                                                                     ploding and residing in silos
                                                                -    smart working techniques to address new
                                                                     business and process demands
                                                                -    dynamic infrastructure for flexibility,
                                                                     effiency and effectiveness
                                                                -    environmental awareness to make the most
                                                                     of existing resources.
Figure 2: Coalition as a digital Ecosystem [2]              3   ONTOLOGY-BASED TEXT ANALYTICS
2   THE INTELLIGENCE CYCLE                                       Managing and preparing the gathered data for
                                                            usage and exploitation is a necessary first step in
     Informing the decision cycle (classically de-          data-driven decision making. Not only is the volume
scribed in Figure 3), and, in particular, the ability to    increasing exponentially in the interconnected world,
pull value from massive amounts of data and respond         but convergence between the digital and physical
appropriately to real-time information has become a         domains is resulting in vastly greater proportions of
competitive differentiator. In all value-networks,          unstructured data with a broad variety of data and
value and advantage are derived from smart data-            metadata quality.
driven decisions enabled by advanced data analytics.
                                                                The Resource Description Framework (RDF, [6])
                                                            is a family of specifications by the World Wide
                                                            Web Consortium (W3C) originally designed as a
                                                            metadata data model. It has evolved to be a general
                                                            method for conceptual description or modelling of
     Figure 3: Intelligence in the Decision Loop            information that is implemented in web resources
                                                            using a variety of syntax formats. The number of
     Data arises from many sources. Instrumentation         semantically tagged documents and datasets in inter-
describes the gathering of data through all kinds of        net/intranet/extranets is growing exponentially.
sensors – ranging from complex ISR systems to               Semantic annotation of data is improving the data
video surveillance to subscription to RSS feed.             collection and analysis capabilities as well as putting
Automation can be understood as the generation of           the involved entities at risk if the annotation is pub-
data on a periodic basis e.g. the writing of daily per-
formance reports. Finally, both friendly and poten-        Regulation
                                                            lished improperly.
                                                                                                                       Po
tially adversary, online communities create data rele-            Fact extraction and semantic linking of facts
vant to any given problem or task – by communicat-
ing in natural language.
                                                            based on an ontology, (taxonomies and rules which
                                                            relate entities), enables both knowledge management       M
                                                            and inference. This approach has the potential to
    Interconnection describes the integration of sen-       enable technology to provide significantly greater
sors, processing systems and decision makers                assistance in answering questions from the EBAO
through pervasive communications already, and be-           operation planning process such as:
coming more readily available. The result is increas-
ing convergence of physical and digital infrastruc-             °    what could happen if …?
tures with more Intelligence embedded in systems,               °    what could happen next?
processes and the infrastructures which support the             °    how can we best achieve an outcome?
businesses they underpin.

    Improving the flow of information between the
                                                                                                    Ensu
people, systems and processes and communicating in
                 Standardization                                                                    Defin
                           T h e D e c is io n
a purpose defined language is a mandatory require-
ment for such a convergence. Battle Management

               Bodies (OASIS,OMG,4: Data Analysis Lifecycle
Language and the underpinning formal lexical gram-
mar ([3], [4], [5]) can be the foundation for such a
                                                                                             L o o p Form

                  NATO, EU ,..) intelligence from raw data drawn from anG a th
language.                        Figure 4 illustrates the IT processes applied to

                                              D e fin e P ro b le m
instrumented environment.


    Analytics is that part of business intelligence (BI)
that focuses on forward-looking analysis. Analytics
can provide more insight to decision makers than
reporting, but is more complex to develop, deploy,
use, and understand. It encompasses simulation, ma-
chine learning, optimization and stochastic analysis.
                                                              Figure 6: Linguistic Analysis based on UIMA
    Naturally, the answer to the above questions can-
                                                               Based on the Object Management Group (OMG)
not be deterministic, but with have probabilities at-
                                                           Business Motivation Model (BMM, [8]), business
tached, making the business rules conditional, e.g.
                                                           rules and policies govern coalition value networks by

                                                           specifying design aspects other than information and
    If A and (B or C) then Action 1.
                                                           operational models.
    By making better use of the information available,
                                                                 The application of policies in a situation and
but because of its volume, not able to be analyzed
                                                           event context leads to their description with highly
and assimilated, these tools could be used to detect
                                                           structured, discrete and atomic statements – ex-
relationships and dependencies between the entities
                                                           pressed in terms of the task-specific vocabulary. This
in the value network. In the example visualization in

                                                                                         Original Data
                                                           concept of using structured English to express poli-
Figure 5, colour codes are used to display capability
                                                           cies as rules in a meaningful and enforceable manner
and role in the value network. Using analytics and
                                                           has been standardized by the BMM and the OMG
powerful visualization tools, such relationship mo-
                                                           Semantics of Business Vocabulary and Rules
dels can lead to the definition of new policies to be
incorporated in military doctrine and reflected in the
RoE for a particular mission.                                                     S truc ture d Data
                                                           (SBVR, [9]). Rules within the business architecture
                                                           are about strategic requirements, not tactical
                                                           execution. They specify modalities about what is
                                                                                   Re q u e s te d By: Co l
                                                                           possible,   impossible,    obligatory,
                                                           permitted or prohibited. Table 1lists the key

                                                                                   Re q u e s t: Ad v a n c e
                                                           modalities defined in SBVR and provides an
                                                           example of each.

                                                                                        m : English SBVR
                                                                                   FroStructuredCo lo g n e
                                                                                       Example     in

       Figure 5: Typical Relationship Network
                                                                                         Bo n n
                                                                                   To :exactly one commander
                                                                           Necessity   An order always has

                                                                           Possibility       It is possible that a
                                                                                             COA has more than one

     The language analysis process also can be inte-
                                                                                    Q: Co lo n e l Ex o rd e r
                                                                              Impossibility An order never includes
                                                                                             another order
grated in the UIMA (the open-source and open-stan-
dard unstructured information management architec-
                                                            Behavioural Obligation  th e a d v a n c e o f tro o p
                                                                                             Each order must be
                                                                                             carried out within the
ture, [7]) framework, as depicted in Figure 6.
The RDF facts extracted by the linguistic analysis                                  fro m b a s e in Co lo g n
                                                                                             set execution time
                                                                                             A frigate may pursue a
                                                                                             pirate vessel
are used to augment the ontology and continue build-
ing the operation process management language.
Stored in a data warehouse (possibly virtual and
                                                                                    to c e n te r o f Bo n n .
                                                                              Prohibition    A commander must
                                                                                             operate within the rules
                                                                                             of engagement (RoE)
distributed), the facts are represented in a RDF triple
store where they can be searched and analysed to
detect further business process rules that are operat-
                                                                                  Uns truc ture d Da
                                                            *) SBVR keywords are given in boldface, terms (names
                                                            for objects) are underlined and verbs (names for relation-
                                                            ships) are in italic.
ing in the complex endeavour. For example, an
analysis of the huge quantities of intercepted voice       Table 1: Types of Business Rules expressed in
and message traffic in a counter insurgency opera-         SBVR Modalities

                                                                                  1-AAcquire Data
                                                                                     Acquire Data
tions would provide much greater understanding of              These atomic statements can be transformed into
the allegiances, communications flows and relation-        complex rules by the use of logical connectors that
ships between the actors in complex ‘human terrain’.       link the clauses built from relationships in the triple
                                                                  Ingests heterogeneous co
                                                                   Ingests heterogeneous co
                                                                  in formats such as docum
                                                                   in formats such as docum
                                                                  emails, voice recordings,
                                                                   emails, voice recordings,
                                                                  warehouses, reference da
                                                                   warehouses, reference da
                                                                 Figure 8: Trust in Business Operation

                                                            The text analytics technology could also be used
    Figure 7: UIMA compliant Analytics Module           to analyse business process rules and detect rules
                                                        which include uncertainty and therefore can support
      Figure 7 conveys the components behind the        inference not only in operational actions (e.g. the

                           Report Generator
geospatially referenced graphical user interface
(GUI). The semantic query builder is used to execute
semantic searches in the fact extraction process de-
                                                        recommendation of specialized suppliers based on
                                                        their current value proposition), but also for process
                                                        optimization and coarse-of-action (COA) adaptation

scribed above (from basic intelligence in the virtual   (e.g. suggesting change in measurements).
knowledge base) to shape operational decision mak-
ing and planing.                                        7     BUSINESS RULES ENFORCEMENT

6   BUSINESS PROCESS OPTIMIZATION                            Mutually beneficial collaboration undertaken to
                                                        meet a coalition’s mission can be considered service
    The Business Process Maturity Model (BPMM),         consumption, with each unit in the coalition consum-
as presented to OMG members in 2006 ([10]) de-
fines the characteristics of business process manage-
                                                        ing the services of each other. This insight defines a
                                                        coalition as both a value network and as a service-
                                                        oriented enterprise (SOE). While the concept of ser-
ment on five levels:
                                                        vice offerings and the consumption of business ser-
    -   level 1: initial – inconsistent management      vices is not new, new challenges resulting from
    -   level 2: managed (via repeatable practices)     globalization and global sourcing of business func-
        – work unit management                          tions has led to the need to orchestrate them in a
                                                        flexible, agile manner.

    -   level 3: standardized (via standardized best
        practices) – business line mgmt
    -   level 4: predictable (via quantitative prac-
        tice mgmt) – capability mgmt
    -   level 5: optimized (via continuous improve-
        ment) – change management.
    The paper mentions in particular a pilot for the
application of BPMM conducted for healthcare ser-
vices with the objective to reduce billing errors and
                                                                            Work unit
to provide a framework for Six Sigma and business                 Tokenizer                     Rule
process reengineering.
     These healthcare services are a good and typical       Figure 9: Rules Detection – Definition – Enforce-
example of a smart eGovernment service based on                                  ment
common intent, shared experience and insight in the
field of interaction and purposeful collaboration be-
                                                            Once detected through semantic text analysis and
tween governmental and non-governmental organisa-       agreed upon, the business rules need to be enforced
tions.                                                  in the SOA runtime. In addition, in order to achieve
                                                        that flexibility and agility, the business rules
     The Business Process Maturity Model enables        optimization lifecycle, described in Figure 9, has to
greater fidelity between the actual performance of      be supported. The rich information in both know-
business processes and their model-based representa-    ledge base and the RDF triple store then enables hy-
tions as illustrated in Figure 8.                               Web
                                                        pothesis testing and inference.                            File
                                                        8      Crawler
                                                              THE FOUNDATION                                     Crawler
                                                               The foundation of solutions for such coalition-          A
                                                                            Individual                                in
supporting service-oriented ecosystems must have          [5] U. Schade, M.R. Hieb: Formalizing Battle
the characteristics of the „Future Internet“([11]):           Management Language: A Grammar for
                                                              Specifying Orders, 2006 Spring Simulation
• openness to ensure all the organisations and indi-          Interoperability Workshop, Huntsville, AL,
  viduals have access to any relevant service                 2006.
• trustworthiness to ensure confident adoption and        [6]
  usage                                                       _framework , last access: April 2009.
• scalability to ensure all sizes of organisation and     [7]         ,   last
  business to be served according to the service lev-         access: January 2009.
  els previously negotiated and agreed upon               [8] ,
• dynamicity and proactiveness as services should             last access: May 2009.
  smoothly adapt to different scenarios when              [9]
  needed and provide convenient support before                05.pdf , 2005.
  they are actually invoked                               [10]J. Alden, B. Curtis: The Business Process
• distributed control and coordinated governance as           Maturity Model, Capability Management, 2006.
  the complexity and the size of the system in-           [11]NESSI SRA Research Priorities for FP 7, NESSI
  creases over time                                           Roadmap Series Document, May 2009.
• ability to deal with uncertainty as the ecosystem       [12] M. Donovang-Kuhlisch, M. Small: Agile Policy
  does not have a fully predictable global beha-              Delivery through Smart Knowledge Discovery
  viour; although behaviour at the interaction level          and Decision Support, MCC 2009, Prague, 2009.
  is both predictable and should be guaranteed.


    In this paper, we have addressed the issue of con-
text- and situation-aware, agile adaptation of opera-
tional conduct in a complex endeavour of a coalition
of governmental and non-governmental organisa-

      We have also shown that whereas a variety of
tools and rules engines exist on the IT market, the
full value proposition can only be materialized, when
emerging text analytics capabilities are leveraged to
make sense of the growing volumes of unstructured
information which characterize complex coalition

    Business architecture principles for model-driven
capability development and integration, as described
in [12], plus the exploitation of linguistic principles
in the process management language have a great
potential and value proposition for future operations’


[1] P. M. Gustavsson, M. Hieb: Machine
    Interpretable Respresentation of Commander’s
    Intent, Proceedings of the 13th ICCRTS, Seattle,
[2] IBM Global Technology Outlook, GTO 2009.
[3] M. Hieb, U. Schade: Formalizing Command
    Intent through Development of a Command and
    Control Grammar, Proceedings of the 12th
    ICCRTS, Newport, 2007.
[4] U. Schade, M.R. Hieb: Battle Management
    Language: A Grammar for Specifying Reports,
    2007 Spring Simulation Interoperability
    Workshop, Norfolk, VA, 2007.

Description: UBICC, the Ubiquitous Computing and Communication Journal [ISSN 1992-8424], is an international scientific and educational organization dedicated to advancing the arts, sciences, and applications of information technology. With a world-wide membership, UBICC is a leading resource for computing professionals and students working in the various fields of Information Technology, and for interpreting the impact of information technology on society.