Upload
rosham-rasli
View
220
Download
0
Embed Size (px)
Citation preview
8/3/2019 Journal of Knowledge Management Practic2
1/14
Journal of Knowledge Management Practice, September 2003
Use Of Ontologies To Enhance The Design Of A Framework
For Multi-Party Collaboration And Decision-Making:
The Case Of Situation Room Analysis
Bob Roberts, Kingston University and Adamantios Koumpis, ALTEC S.A.
ABSTRACT:
Work presented in this paper forms part of a wider research in defining a methodological
framework for Situation Room Analysis (SRA), and its deployment for complex (business
enterprise) systems study. In our approach, we propose the use of ontologies as a powerfulmeans to support the implementation of multi-party collaboration and decision-making activities
that build on the paradigm of a Situation Room (SR). The approach is complementary to others
in the area of business planning and is characterised as top-down in that the SR paradigm isconceptualised through three related models: the Situation Room Model (SRM), the Information
Management Model (IMM) and the Situation Analysis Model (SAM). The ontology-based
approach includes the semantic features of the exchanged decision-making information thusoffering the integration of the SRA framework with existing corporate decision-making grids.
1. Setting The Stage: Situation Room Essentials
Work presented in this paper forms part of a wider research in defining a methodological
framework for Situation Room Analysis (SRA), and its deployment for complex (business
enterprise) systems study. In our approach, we propose the use of ontologies as a powerfulmeans to support the implementation of multi-party collaboration and decision-making activities
that build on the paradigm of a Situation Room (SR). The latter term is broadly used in the
context of military operations and has specific semantical connotations. These connotations aredeliberately exploited in order to propose an analytical scheme based on it, which aims to assist
planning initiatives and decision making in a particular application domain.
We recognize a unique opportunity to consider the attempt to organise enterprise-wide SR indirect connection with KM efforts that a company or an organisation is planning to undergo; in
this case, SRA may provide the means to organise KM practices, while the SR may provide a
concept that will be subject to valuation with respect to performance indicators, both at the micro(i.e. for a given situation and how the company or the organisation has performed with respect to
it) and the macro levels (i.e. how for a longer term horizon the company has organised its
resources to respond to external stimuli and demands for action that are regarded as situations,and how this helped increase the knowledge capital of the company for the given matters).
8/3/2019 Journal of Knowledge Management Practic2
2/14
Like all systems of notation and semantics, Situation Room Analysis can never be anything but a
model; it is a map to a territory, whose validity may be evaluated by reference to that underlying
reality in the real world. Historically speaking, a Situation Room is considered as theintelligence analysis centre used to stay abreast of the latest intelligence reports and updates.
Such intelligence allow an armys or an army units senior officers to make informed command
decisions and / or stay current on news throughout the federation of other units and beyond.Within this aim, i.e. the multi-party collaboration and decision-making activities from within the
Situation Room Analysis framework, it is easy to see that the latter should be data-driven. In this
respect we can consider the case of, for example, a company planning to create a virtual marketresponse Situation Room to improve how it collects and assesses information regarding its own
and / or competitors products. Such an approach would enable stakeholders to obtain important
data in a more timely and effective manner.
The SRA framework as presented by Koumpis and Roberts in (Koumpis, 2003) emphasises the
idiosyncratic characteristics of the ICT sector such as innovation, technological change, transferof technology and technology diffusion. These require the development of a design space where
different scenarios will be subject to in vitro assessment and evaluation. The employment of theframework may take place during any phase of the life cycle of an ICT service or product, i.e.from the early design phases up to the phase of its launching into the market.
We regard our approach as complementary to others in the areas of business planning; Mandal et
al (Mandal, 2003), for example, suggest that managers should conduct a pre-alliance planning
exercise to assess the compatibility of business goals of partners, determine a method for
implementation, and indicate the key informational as well as cultural challenges that may arise
throughout the alliances duration". In their work, they report on a set of key concepts related tothe formation of an alliance which may prove critical for its successful execution which include
the "efficient and effective decision making". They argue that actions to build and sustain a
strategic alliance include, amongst others, performance indicators concerned with benchmarking,knowledge management in the team environment and decision making involved with technology
management.
Collaborative business decision-making software, and all forms of business software, may be
viewed as a representation or a set of symbols, for some underlying reality. An analogy may be
drawn with classic double entry accounting systems that may be seen as providing both amethodology and metadata framework capable of representing different kind of business
transaction and financial state change that can exist in the business problem space (See for
instance the work of Baruch Lev as reported in his bookIntangibles: Management, Measurement
and Reporting, Brookings Institution, June 2001.). Similarly, there is a need for a systems todescribe corporate decision making activities that builds on (elementary) information
management transactions. Howeer, such a system also needs to represent subtle relationships and
hierarchies and it is at that point that the need to exploit the expressive power of ontologies
comes to the foreground.
2. Use Of Ontologies
8/3/2019 Journal of Knowledge Management Practic2
3/14
According to Hahn (Hahn, 2003), semantic web technologies and management tools can be used
to link the model entities of a distributed model. More specifically, two integration concepts havebeen used:
file-based where the information is stored in a XML representation enriched with meta-
information expressed in RDF (Resource Description Framework), and
online, where tools can provide the information online with an interface implemented as aWeb Service.
A promising approach, which we propose exploits semantic indexing techniques. The latter,though not new, is a sophisticated indexing schema, which allows us to support the kinds of
operations necessary in an efficient way.
The employed indexing scheme is based on ontologies: taxonomic information with additional
links that represent associated properties. While some ontologies represent very general
knowledge, others specifically target a particular domain (Ankolekar, 2001).
In order to use ontologies for indexing we have to establish links between the data in the SR data
warehouse and the concepts in the ontology. All pairs of attributes and values in the database aremapped to concepts in the ontology. An example is given in the Figure below. The solid links are
ontological links and the dashed links are indexing links.
8/3/2019 Journal of Knowledge Management Practic2
4/14
Figure 1: Use Of Ontologies For Indexing Data / Information Entities Used For SRA
The aim is to provide support for queries such as: Which actions are effective against 80% of
situations sharing commonality element A.1 with the situation we are facing today?
The integration of ontologies into corporate legacy information and database systems is ofgrowing interest as they can increase the efficiencies of the way a company uses existing
information (re)sources. For instance, in the case of a company that uses a traditional division of
their activities into different cost or profit or value centres, each centre is related to different
tasks and the aims it is expected to fulfil relate to different elements of the corporate objectives(see Figure below). To address this, each department or business unit uses a particular ontology,
which provides the communication means for assisting coordination of its core business.
8/3/2019 Journal of Knowledge Management Practic2
5/14
Figure 2: Organisation And Mapping Of Semantics For The Different Cost/Profit/Value
Centres With Ontologies
Such a local ontology may heavily vary from centre to centre. For example, the introduction of a
new product relates to profitability aspects for the Finance Dept and invokes the possibility of
stopping any further production and selling of older products that have reached their maturity inthe market. It also raises questions related to campaigning, competitors attitude, e.t.c. for the
Marketing Dept, as well as minimisation of production costs and waste for the Production Dept.
From the above, it is clear that there is need for a treatment of the semantics for each different
notion as this appears at the local ontology level; this is the role that can be assigned to the
Global Ontology, which affects the entire corporate process grid.
Before going into a deeper level of analysis for the use of ontologies in our framework, we
provide some background information on modelling aspects of the Situation Room Analysisframeworks as well as some basic services it is expected to provide to SR participants, and for
which the use of ontologies is considered as a contributing factor to overall efficiency.
3. Modelling Aspects Of The Situation Room
8/3/2019 Journal of Knowledge Management Practic2
6/14
This section includes some additional information on the models pertaining to the SR concept.
Our goal is to support high-level corporate operations, such as planning and projectprogramming, by means of defining the Situation Room Analysis as an expressively powerful
vehicle to support this need. The main entities for defining the basics of Situation Room Analysis
are related with:
the concept of the Situation Room per se,
the managed information within the SR, and
the main items of the conducted analysis which in our case focus on products and services inthe IT market.
In regard to all three of them three corresponding models are defined, namely:
The Situation Room Model (SRM),
The Information Management Model (IMM), and
The Situation Analysis Model (SAM)
They all concern descriptive conceptualisations of entities and activities, annotated with the
interactions and possible relationships amongst them, which results in a super-model namely theSituation Room Analysis. Furthermore, we elaborate on this by providing the specifications for
setting up the implementation of this to an IT framework using emerging technologies (XML,
software Agent, Semantic Web and ontology technologies) and established system design
approaches (UML). In both modeling and interpreting the impact of information in the context of
the virtual network the research is also informed by game theory and transaction cost theory(Friedman, 1991) and (Casson, 1991).
In the table below we provide a description of the supported actions on a given information
entity as this is realised within the Information Management Model of the SRA framework:
Nr Identifier Action type Description1 RM Remove It is destroyed as if it never came under consideration within a set structure
under use in the Situation Room. This is not a usual or recommended
practice, but may simplify procedures in several situations. A more
recommended practice is to justify reasons for its irrelevance and ignore it
(see below). However, and as long as logging of events is taking place,
tracing back to this state is possible.2 IGN Ignore It exists but is not used for any current inferences made within a set structure
under use in the Situation Room. This is the case of trying to simplify a
problem by ignoring (temporarily or permanently) a set of information
regarding specific aspects of the subject under consideration.
3 LN Link With some other piece of information within a set structure under use in theSituation Room. How? By means of choosing one of the enabled link types:
8/3/2019 Journal of Knowledge Management Practic2
7/14
3a LN_TO As above Link as related to with a unidirectional linkto the other information entity
3b LN_FROM As above Link as related to with a unidirectional linkfrom the other information entity
3c LN_BOTH As above Link as related to with a unidirectional linkfor both information entities
3d LN_ONL As above Link "only" to the other information entity without any further pre-definedrelationship between them
3e CUST_LN As above This type enables user defined link types to be created by means of enabling
users of the system to develop their own link categories, which may bedomain- or user-specific and which may vary amongst each of the users or
usage types.3f LN_LN As above This forms an important type of linkage as it provides the means to link one
link with another link.4 ADD Add it It concerns the insertion of a particular information entity to a set structure
under use in the Situation Room.5 CRT Create Creation of a new placeholder
Table 1: Potential Supported Actions On An Information Entity
Note that in regard to the above Table, one possible difficulty in the implementation, which may
result in consistency and constraint satisfaction problems is this of the space ofLink type
relationships: in our description we define this to be between two entities. It is easy to see for
instance that especially for 3d, 3f and 3e it is essential to support linkage with more than one
information entity. For implementation reasons, we propose that the system design might
proceed in the definition of a Group action which enables groupings. However, such an action is
included in the design but aims to facilitate aggregation operations for information entities.
Thus, the approach we would propose is one of implementing a generic type of link that support
tuples of 2, 3 or N information entities. Bearing in mind existing development environments and
programming languages, this is trivial to support. However, 10 years ago it would have
necessitated the development of a mechanism to handle this as a separate activity
As seen from the above, the central notion for an information entity within the InformationManagement Model is this of linking it to other entities. Furthermore, also important are the
"placeholders" in which a specific entity will be input. These may either be predefined if we
expect specific entities to populate them, or realised ad hoc. Such ad hoc creation of aplaceholder often takes place under time and resource pressure and therefore its results are
usually suboptimal. For this reason it is essential that placeholders are reconsidered on a periodic
schedule and - if needed - adapted, renamed or consolidated with others.
In regard to the placeholders the same actions hold as for the information entity. There is,
however, an exception and this is for the creation of a new placeholder. The reason for this is thatwhile a piece of information has arrived and we recognise its existence, a placeholder is anartificial artifact for which we are solely responsible for its construction. It has been considered
as out of the scope of our research to further investigate this aspect. However, this is not always
the case: there is frequently the case of information creation, based on synthesis of other
(previously existing) information or even out of nothing (the latter is also the case of making
some hypothesis because we simply want to make it or need to make it).
8/3/2019 Journal of Knowledge Management Practic2
8/14
A further important aspect of the Information Management Model relates to the ability for
representing all actions performed or attributed to particular information entities. In this respect,
what is actually needed is a device that guards some conditions and performs some actionswhen the conditions are true. This idea is not new in the Computer Science theory and practice as
it is expressed by well-known metaphors like demons in AI and triggers in databases, and it has
been widely used in several modelling languages and development environments (see forinstance Widom & Ceri (Widom, 1996)). Our notion of a linkerelement(in brief: linker) realises
this idea in a slightly different fashion. While normally, the condition is defined by a universal
predicate, which means that the guard needs to observe the whole, or a large part of a database tofind any place where the condition is true, our linker works locally, as it guards only its own
operands.
According to our approach the linker is the only way:
To express relationships amongst information entities, be they passive or active relationships.Thus, we use the same notion for describing both static information entities and actions to them.
The uniformity allows treating actions in the same way as static links, i.e. we can add and deleteactions in the same way as we add and delete static relationships during a database transaction.
To define actions on information entities. In this respect, any action that takes place within the
SR to enrich or explain an information entity is simply linked to the previous state of the entity,providing also the last inherently proprietary characteristic of that last action.
4. Implemention Of Ontologies Within The SRA Framework
Our approach builds on the adoption of a widely adopted service oriented architecture thatencompasses the simplicity and scalability of the Web services model (see Figure 3 below).
Besides the simplicity of implementation, the advantage of modularity also enables the"repackaging" of any existing SRA services into new, composite services. This increases the
added value of the framework and should encourage companies to invest in its usage andpopulation with new situation data. Preliminary developments that we have been experimenting
with took place across a network of workstations emulating the conditions of a realistic Situation
Room.
8/3/2019 Journal of Knowledge Management Practic2
9/14
Figure 3: Adoption Of The Web Services Model For The Provision Of Basic
SRA Services
As a result of this, we identify the need for introducing semantics in our approach as:
synonyms of situations in various corporate contexts,
equivalent situation types in various corporate contexts,
but also for:
Synonyms of situations in the same corporate context,
Equivalent situation types in the same corporate context.
It is in this respect that there is need to support semantic level processing for the collaborative
SRA services to be delivered through the underlying application. For this, the technical goal is toprovide a transparent system architecture acting as a broker between cross-Cost / Profit / ValueCentres that will automatically handle the inconsistencies amongst the different situations and
coordinate inter Cost / Profit / Value Centre process management.
Based on the above, it is obvious that we need declarative forms of scripting complex web
services, which would also enable composition of scenarios that represent real-world
8/3/2019 Journal of Knowledge Management Practic2
10/14
coordination amongst the different members of the SR, also taking into account temporal and
synchronization aspects.
Ankolekar et al. (Ankolekar, 2001) describe DAML-S, a DAML+OIL ontology, designed by the
DARPA Agent Markup Language (DAML) Services Coalition, to specify the capabilities of
Web Services. DAML+OIL provide a semantic and further expressive power to the ExtensibleMarkup Language (XML) and the Resource Description Framework (RDF). Furthermore,
DAML-S provides service descriptions in three conceptual areas:
the profile, which describes what the service does,
the process model, which depicts how the service works and
the grounding which states how the particular service is used.
4.2. Representative SRA Services
In this section we describe three basic services that are provided by the framework and which
have been implemented using the aforementioned approach.
We used our services by means of building and adapting three of the essential products of theC4ISR Architecture Framework which are indicative of the expressive power of the SRA model,
thus providing the means of application domain- and context-specific customisation of it.
4.2.1. Situation Synopsis
The Situation Synopsis facility addresses essential aspects of a situation considered by means of
providing answers to questions related to Who? What? When? Where? How? of a particularsituation under consideration. In this respect, it may facilitate the initial phases of planning. It iseasy to recognize the need for it to be provided in a consistent form that will allow quick
reference and comparison amongst other situations, thus disabling error proneness with respect
to linkings with the wrong situations. It is upon the Situation Synopsis that indexing andretrieval operations will be based. It is time-dependent, i.e. as time goes by it may changeafter
the completion of a situation, it is still important that this has been appropriately documented in
the Situation Synopsis.
The following apply when providing the Situation Synopsis:
Identification. Provide a unique descriptive name for the situation, identify the personresponsible for its handling (: the "Account" manager for that particular situation),
identify involved units, e.t.c.
Purpose. Explain why the SRA is needed, what it is intended to achieve, the types of
analysis expected to be applied to it, who is expected to perform the analysis, whatdecisions are expected to be made on the basis of that analysis, who is expected to make
those decisions, and what actions are expected to result from the architecture.
8/3/2019 Journal of Knowledge Management Practic2
11/14
It is highly possible that the answers to these questions cannot be given from the
beginningit is however imperative that the person responsible for the situation will tryto provide answers.
Of course, a situation that was initially identified to be related with a cause X might be
attributed to cause Y; and the interactions with some other entities may have beeninitially erroneously attributed to some reasons Z. For this reason it is essential that the
Synopsis will enable enrichments and furthercontinuous - updates.
Scope. Identify the situation views and implications and its particular temporal nature,
such as the time frame covered, e.g. whether by specific years or by designations such as
as-is, to-be, transitional, e.t.c.
Context. Describe the interrelated conditions that compose the setting in which thesituation exists. Include such things as doctrine, relevant goals and vision statements,
concepts of operation, scenarios, and environmental conditions. Identify the tasking that
led to the architectures development, and known or anticipated linkages to othersituations.
Document-specific assumptions and constraints regarding the situation analysis effort,and identify authoritative sources for the rules, criteria, and conventions that were
followed in developing the particular syllogisms.
Findings. State the findings and recommendations that have been developed based on the
above. Examples of findings include recommended actions, identification of shortfalls or
successful implementations, and opportunities for reaction.
Tools and file formats. Identify the tool suites to be used to support the SRA exercise.Identify the system and file names, and location of the data and appropriate resourcesincluding also human actors.
4.2.2. Integrated Situation Dictionary (ISD)
There is considerable textual information in the form of definitions and metadata (i.e., data aboutan item) associated with the various situations encountered. The Integrated Situation Dictionary
provides a central source for all these definitions and metadata, including those that may be
provided for convenience within another architectural component as well. At a minimum, the
Integrated Situation Dictionary is a glossary with definitions of terms used in a given situation
description. The Integrated Situation Dictionary makes the set of components capable ofstanding alone and allows a set of situation related documents to be read and understood without
reference to other documents.
Each labeled item (e.g., terms, phrase, or acronym) in the situation literature should have a
corresponding entry in the Integrated Situation Dictionary. For instance, when we speak about a
Sales downsizing the ISD provides a unique explanation for this. The same also applies when
speaking about a Sales downscalingwhatever this may mean too. By using specific
8/3/2019 Journal of Knowledge Management Practic2
12/14
terminology, actions and reactions can be standardized and this saves time, decreases error rates
etc.
The type of metadata included in the Integrated Situation Dictionary for each type of item will
depend on the type of the component from which the particular service item is taken. For
example, the metadata about a labeled input/output connector from an activity model will includea textual description of the type of input/output information designated by the label.
The contents for the Integrated Situation Dictionary entries for each component type should be
regarded as evolving, as is the case for any dictionary of a natural language. SR participants
should use standard terms where possible (i.e., terms from existing, approved situationdictionaries). However, in some cases, new terms and/or modified definitions of existing terms
may be needed. This can happen when new concepts are devised. In those cases, the new terms
contained in a given architectures Integrated Situation Dictionary should be submitted to the
maintainers of the SR for approval. All definitions that originate in existing dictionaries shouldprovide a reference to show the source, which may be the first situation in which a particular
term was used. Furthermore, indicative references to a term may be used for helping thecomprehension of the particular term(s). In this respect, the terms Sales downsizing and Sales
downscaling might have been used for the first time in situations ABC and XYZ respectively,
while the best example for conceiving their notion may be situations abc and xyz respectively.
Indexes and thesauri that provide support for synonyms, or other type of processing of theparticular semantics of a term is not considered as part of the Integrated Situation Dictionary.
4.2.3. Situation Concept
The Situation Concept is the most general of the architecture-description service components and
the most flexible in format. Its main utility is as a facilitator of human communication, and it is
intended for presentation to SR participants and decision makers. This kind of service can also beused as a means of orienting and focusing detailed discussions. A possible template may show
generic icons that can be tailored as needed and used to represent various classes of actors in a
particular situation under consideration. The icons could also be used to represent missions ortasks. The lines connecting the icons can be used to show simple connectivity, or can be
annotated to show what information is exchanged.
How the template is tailored depends on the scope and intent of the implementation, but in
general a Situation Concept should be capable of communicating to interested parties some basicinformation regarding causality and time dependencies, as well as interactions amongst the
various actors involved.
5. Conclusions
The fast growth of innovations in the last 20 years (coming mainly from the service and
engineering disciplines) exposes companies and their shareholders to varied risks and differenttypes of risk that may be difficult to quantify. Though extended report-centric infrastructures
have been set (companies investing several thousands of Euros on them on an annual basis) these
8/3/2019 Journal of Knowledge Management Practic2
13/14
often result in extensive yet largely meaningless statements enumerating every possible risk yet
still exhibit insufficient specific risk disclosures.
The problem with business process modeling the way it is often currently done by companies is
that it is not an integral part of the development process. It is rather more of a pre-development
process where a model of how the business works is produced independently of the developers.From the developers perspective such a model has no relevance as a development artifact. The
challenge is to produce modelling artifacts which are an integral part of the development processand which automatically generate the supporting applications, as well as the respective
application protocols. The concept of Situation Room Analysis (SRA) is proposed as a means of
achieving this level of integration.
In the paper we have proposed the usage of an ontology-based approach that offers inclusion of
the semantic features of the exchanged decision-making information thus improving the quality
of integration of the SRA framework with existing corporate decision-making grids. Ourapproach was verified building representative applications by means of Web services and the
DAML-S language. Furthermore, in our SR family-specific modelling approach, the particularmodels as well as their instantiations are made up of elements representing notions that are partof the addressed corporate decision making domain world, not its implementation in the software
code world.
References
Koumpis A. (1997), Situation Room Analysis in the Information Technologies Market, in
Communications of the ACM, Vol. 40, No. 3, March
A. Ankolekar, M. Burstein, J.R. Hobbs, O. Lassila, D.L. Martin, S.A. McIlraith, S. Narayanan,
M. Paolucci, T. Payne, K. Sycara, H. Zeng (2001), DAML-S Semantic Markup for WeebServices, Proc. Of the First Semantic Web Working Symposium (SWWS 01), Stanford, The
SAML Services Coalition
Friedman J. W. (1991), Game Theory with Applications to Economics, Oxford University Press,London
Casson M. (1991), The Economics of Business Culture: Game Theory, Transaction Costs andEconomic Performance, Clarendon Press, Oxford
Widom, J., Ceri, S. (1996), Active Database Systems: Triggers and Rules for Advanced database
processing, Morgan Kaufmann, New York
Mandal, P., Love, p.E.D., Irani, Z. (2003), Pre-alliance planning: development of an informationsystem infrastructure to support strategic alliance activities, Management Decision, Vol. 41, No.
2.
Koumpis, A., Roberts, B. (2003), A framework for Situation Room Analysis and exploration of
its application potential in the IT sector, in First International Conference on Performance
8/3/2019 Journal of Knowledge Management Practic2
14/14
Measures, Benchmarking and Best Practices in New EconomyBusiness Excellence '03,
University of Minho, Guimaraes, Portugal, June 10-13
Hahn, A. (2003), Integration and Knowledge Management Platform for concurrent engineering,
in 9th International Conference of Concurrent Enterprising (ICE2003), Espoo, Finland, June 16-
18
About The Authors:
Dr Bob Roberts BSc (Hons), Msc, PhD is leader of the e-commerce group in CARIS and coursedirector of the MSc in E-Commerce at Kingston University. After graduating from the London
School of Economics, he worked in IS/IT consultancy for BT and then General Electric
Information Services supporting their EDI services in Europe. Since joining Kingston University
research interests have focused on the use of electronic commerce to support information sharing
between trading partners and the design implications of integrating electronic commercecapabilities into existing information systems. Recent research and consultancy activities cover
a range of e-commerce projects in the telecoms, health, construction and electronic sectors. This
includes DTI funded projects for researching the use of electronic commerce by SMEs. Otherexternal e-commerce related activities include being a member of TradeUK (DTI) Best Practice
Working Party on Electronic Commerce and Internet Marketing and also member of the InternetWorking Group of the Centre For Study of Financial Innovation. He may be contacted at:[email protected]
Adamantios Koumpis heads the Research Programmes Division of ALTEC S.A., which hefounded at 1996 (then as independent division of Unisoft S.A.). His research interests include
quantitative decision making techniques and Info Society economics. He successfully lead manycommercial and research projects in Greece in the areas of E-Commerce, public sector and
business enterprise re-organisation and information logistics, concerning linking ofdata/information repositories with knowledge management and business engineering models.
Mailto: [email protected]
mailto:[email protected]:[email protected]:[email protected]:[email protected]