LET’S SEMANTICISE THE WORLD!! … OR NOT??

Michele Missikoff

LEKS - Lab for Enterprise Knowledge and Systems, IASI-CNR, Italy

Keywords: Knowledge Economy, Ontology, Semantic Annotation, Semantic Business Process, Semantic Business

Services, Mapping Discovery, Enterprise Model, Model Driven Architecture.

Abstract: Many say that Knowledge is the oil of the Third Millennium. The European Council, in the Year 2000,

launched the Lisbon Strategy

, aiming at building in Europe the major Knowledge Economy in the world.

Then we had the 9/11 and the “subprimes” (the future is rarely as we expect) and an epochal economic

slowdown, without an equivalent since 80 years. Nevertheless, the knowledge economy, pushed by the

consistent evolution of the information technologies, is steadily progressing and further expansion is

foreseeable in the near future. We believe that for a renewed sustainable growth, semantic technologies will

play an important role. In this paper we briefly draw the main lines of a possible future evolution of the

application of semantic technologies to the business world.

1 INTRODUCTION

The knowledge moves the World. The human beings

are not among the fastest or the strongest animals on

the Planet, nor they are protected by a special fur or

by a hard carapace. However, they acquired the

leadership on the Planet thanks to the knowledge,

and the capacity to use it for practical purposes.

With the advent of ICT there has been a

tremendous impulse on how humankind deals with

knowledge. Since the first wave of computers, in the

social and industrial realities of the 60s, the

knowledge managed by a computer can be seen

divided in two different sorts: knowledge for

computers, e.g., software programs, and knowledge

for humans, e.g., digital representation of paper

documents. In the former case, the computer is able

to execute the software, without “understanding”

(this term will be better clarified later) it, while in

the second case the computer is just a container (i.e.,

incapable of execution nor understanding). In

parallel, Artificial Intelligence, and in particular

Knowledge Representation methods and systems,

started to develop solutions, including languages and

reasoners, to provide the computer with some forms

“understanding” and execution of knowledge.

This represents a third sort of knowledge, somehow

positioned between the two above.

To better clarify the different perspectives,

human- and computer-oriented, with respect to

knowledge understanding and processing, at an

intuitive level we may draw a diagram where on the

x-axis we place the three knowledge environments

corresponding to the three different sorts of

knowledge: (i) document and content management

systems (CMS), (ii) semantics representation and

processing systems (SRS), (iii) programming and

execution systems (PES). Then we draw three lines.

One concerning the ease of access for humans to the

knowledge represented in the three environments,

the second represents how easy is for the computer

to execute (or interpret, if you prefer) the

knowledge, and the third how much the computer

“understands” the knowledge

therein.

Time passes, and the basic issues of

programming and content management did not

change much. Conversely, knowledge and semantics

representation methods and tools have evolved,

thanks to the energies spent in research and

development, with the objective, among others, to

achieve a unified knowledge space: same knowledge

(possibly, in different digital formats) for humans

and computers, and similar “understanding” and pro-

http://en.wikipedia.org/wiki/Lisbon_Strategy, visited on

April 2009

For “understanding" we intend here the capacity of an

active entity to acquire knowledge and be able to modify

its behavior on the bases of to the acquired knowledge

(we disregard here the pure speculative knowledge).

Matsumoto M.

SERVICE COMPUTING EIS, WORLD PANIC AND OUR ROLE CHANGE.

DOI: 10.5220/0006811800010001

In Proceedings of the 11th International Conference on Enterprise Information Systems (ICEIS 2009), pages 5-15

ISBN: 978-989-8111-84-5

Figure 1: The KM effectiveness.

cessing capability. Today, the “Mission: Impossible”

is the progressive convergence towards such a

unified knowledge space.

May be, the Impossible Mission becomes more

possible if we restrict our world to enterprises.

Today, an enterprise produces every day an

incredible amount of documents, all of them in

digital form (from technical reports to meeting

minutes, from emails to field studies, to market

analysis). Therefore, we have a sort of Enterprise

Image in Digital Form (EIDiF) that reflects all the

possible knowledge, produced and/or acquired, that

traverses across the enterprise organization, the

production, the marketing, and all the other

departments of the company. The EIDiF primarily

consists in the whole set of human-oriented

knowledge (i.e., digital documents, but comprises

also the vast amount of information typically

maintained in the enterprise databases). The content

of the documents, as they are produced and

exchanged by humans, cannot be directly understood

by computers. To allow computers to access such a

content, there is the need of a significant pre-

processing: the semantic content of the documents

need to be extracted and represented in a formal

way. This is one of the primary goals of semantic

technologies.

In recent years, semantic technologies are

proving to play a central role. New (and renewed)

solutions are spreading, by using formal knowledge

coding, ontologies, reasoners, and semantic

annotations, addressing both the static (e.g., business

objects) and the dynamic part (e.g., business

processes) of the enterprise. We are witnessing the

progressive evolution of such technologies towards

the achievement of the aforementioned Unified

Semantic Space. Indeed, we see that computers and

humans are increasingly cooperating, helping each

other in more sophisticated, knowledge-intensive

activities (naturally, having humans superordinated,

to avoid, e.g., an Asimov scenario).

As anticipated, a key issue in this scenario is the

possibility of (semi)automatically extracting the

semantic content of enterprise documents, encoding

it in a machine “understandable” form. In essence, it

means to develop a formal semantic theory of the

addressed matter, of the enterprise and the context it

operates in. This is today a clear trend, but how

much will it be possible to capture and model in

formal terms? There are various signs showing that

we are indeed proceeding along this line. From

enterprise ontologies, to formal business process

modelling, from automatic knowledge extraction to

business rules, the research is eagerly progressing in

the direction of developing new formal methods to

model and manage the business reality. Furthermore,

there are important theories that can be applied in

specific business sectors, such as chaos theory and

system dynamics (not to mention applied

mathematical theories that go from finances to

logistics, from market analysis to accounting).

SRS PES CMS

Computer understanding

Computer executing

human understanding

Effectiveness level

Figure 2: The Enterprise Digital Image.

The target is a business world where a rich

collection of formal theories and precise models

allow the computer to meaningfully manage an

increasingly large fraction of the business

knowledge, covering the large majority of activities,

behaviours, objects, and actors that are involved in

business transactions.

In the current scenario the research and

innovation is proceeding in a very fragmented way.

New theories produced by different disciplines, may

be effective on specific problems, but are hard to

relate each other to achieve a coherent,

comprehensive solution. We still have a huge level

of fragmentation that, optimistically, yields to a sort

of federation of different theories, produced by

different communities, with very limited capacity of

integration.

But reality is one, so the goal is the construction

of a

holistic formal enterprise theory. There have

been scientists who anticipated a vision of this sort.

For instance, the work of Max Tegmark (Tegmark,

2008) whose visionary approach can be condensed

in the idea of a mathematical TOE (Theory Of

Everything) based on the notion of a Mathematical

Universe (MUH: Mathematical Universe

Hypothesis). While Tegmark is trying to build such

a vision in a top-down approach, there are several

other promising research lines that proceed bottom

up. It is worth citing the emerging field of

Econophysics (Rosser, 2006). Here the researchers

acquire all possible quantitative data representing

any possible business phenomena. Then, advanced

rigorous methods, mainly drawn from theoretical

physics, are applied to discover regularities. Very

promising results are obtained, for instance, by

applying statistical quantum mechanics.

Any mathematical theory, to be fully effective

and in order to relate a mathematical theory to the

real world (the enterprise world, in our case), needs

to be associated with a clear semantics. The latter

has the essential goal of creating a bridge between

symbols and expressions in a mathematical theory

and objects and events in the real world. In essence,

a mathematical theory always needs a semantic

apparatus connected to it, no matter how advanced,

sophisticated, accurate it is, to explains and predict

relevant phenomena in the modelled (business)

domain. Such a semantic apparatus can assume

different forms. It can be mainly intuitive (for

instance, simply using natural language, with words

connected to mathematical expressions, variable

symbols, quantities, and other elements of the

theory) or it can be a formal theory on its own part.

The idea is that a comprehensive, encompassing,

accurate, effective formal semantic infrastructure,

conceived for a given business sector, will be

capable of correctly positioning and cross-relating

all the specific theories of different sorts that are

proposed for modelling and explaining different

sectors and viewpoints in an enterprise. For this

reason, we will proceed in analysing the possibility

of building such a semantic infrastructure of an

enterprise based on a formal setting, achieving what

we refer to as the Semantic Enterprise.

2 TOWARDS THE SEMANTIC

ENTERPRISE

In proceeding in the direction of the semantic

enterprise, it is useful to first sketch a layered

architecture where the various semantic services can

be placed in a rational way. We start to illustrate the

sketchy architecture reported in Figure 3 starting

from the bottom layer.

Figure 3: A semantics driven architecture.

2.1 Basic Semantic Services (BSS)

We start bottom up, with the aim of building a

supporting semantic infrastructure. Such an

infrastructure comprises three essential elements,

and the corresponding platforms and services.

Ontologies. When we talk about semantics, we

primarily intend, in accordance with (Ushold 1998),

a set of concepts with their relationships, internal

structure (e.g., represented by the associated

properties), and constraints (i.e., axioms to be

enforced). An ontology (a taxonomy of structured

concepts) can be formally represented by using some

sort of logic.

Given a complex reality, such as an enterprise or a

business domain, generally it is practical to develop

more than one ontology, each of which having a

clear focus. E.g., a marketing ontology, production

ontology, HR ontology. But the reality is one, and

humans tend to segment it to be able to cope with

the complexity, therefore the segmented ontologies

need to be related each other. This can be achieved

with an inter-ontology mapping infrastructure.

Semantic Annotation. The rich and articulated

collection of documents and (factual) data sets that

we can find in an enterprise always refer to real

business entities, representing therefore the EIDiF

(i.e., the digital image of the enterprise): the

products, the people, the markets, the competitors,

etc., where each element class has its own form of

digital representation, typically, to be presented to

and managed by humans.

A key step in the “semanticisation” of the reality is

the systematic creation of mappings between entities

and phenomena of the business reality and concepts

of one (or more) ontology. We refer to such

mappings as semantic annotations. In its simplest

form, a semantic annotation is a direct link between

a concept in an ontology and an element of the

enterprise. E.g.,: (Mario ÅÆ Employee). In general,

to provide an articulated account of a fragment of

the business reality, it is necessary to create an

ontology expression.

Reasoner & Truth Maintenance. The two above

structures, ontologies and semantic annotations, are

essentially repositories of knowledge. Then we need

to complete this layer with an engine capable of

processing the above structures to perform two basic

computations: (i) derivation of new knowledge and

(ii) verification that the existing structures do not

contain contradictions.

These functionalities are necessary for the very same

management and maintenance of the semantic

repositories. In fact, the reality continuously evolves

and the enterprise documents will reflect every

significant evolution, therefore the semantic

repositories, that reflect the state of the affaires,

must evolve accordingly. If we add a new piece of

semantic knowledge in an ontology that we know is

free of contradiction, a TMS (truth maintenance

system) is responsible to guarantee that also the new

version of the ontology will be released in a

consistent state, i.e., we did not introduce any

elements that invalidates what we already know

2.2 Semantic Service Utilities (SSU)

On top of the base semantic services we can build a

number of semantic service utilities. A service utility

is a software facility available to everybody,

accessible with a predefined universal protocol at

predefined published conditions (cost, SLA, etc.).

The proposed SSUs provide the key functions on top

of which any other possible semantically-enhanced

application can be built. In turn, to exist the SSUs

need the BSSs seen in the previous paragraph. The

key SSUs are listed below.

Here we touch a first limitation of the current semantic

technology: the limited capacity of representing

contradictions. While we know that the reality is full of

contradictions

Base Semantic Services

Semantic Service Utilities

Semantic VAS

Semantically Enhanced

Application Services

Semantic Knowledge Mining. This service utility is

capable of accepting a human-oriented document

(from a plain text document to an XML file, from a

business process diagram to an email) and an

ontology, returning a set of semantic annotations. In

essence, this service utility is capable of extracting

the semantic content of a document, building an

explicit, formal ontology-based structure (therefore,

by using different ontologies it is possible to have

semantic annotations emerging from different

disciplines, different perspectives).

Semantic Matchmaking (SMM). This service utility

is based on some mapping discovery methods.

Mapping discovery is a very vast research area

aimed at solutions that, given two structures, are

able to derive a formal relationship that characterises

the correlation between the two structures as a whole

and among the individual elements of the two

structures (in case, considering also relevant sub-

structures). There is a wealth of available methods in

the literature, with a great variety of structures

considered (from graph stractures to logical theories,

from geometrical figures to natural language

sentences and paragraphs). Here we restrict our

focus to ontologies (or fragment of them). The

output of a SMM service utility can be of various

sorts from an algebraic condition, such as

equivalence, containment, of disjunction, to a

quantitative measure (e.g., a value between 0, in

case of disjunction, and 1, in case of identity).

Semantic Interoperability Service Utility (SISU).

This service utility is necessary when we have two

structures and, once we identified with the previous

SMM that there are divergencies, we intend to

reconcile such differences. This service is similar to

the previous SMM, but while in the previous case

we intend to identify the similarity degree, as a static

declarative parameter, here we intend to identify an

active mapping, that is the operations necessary to

transform one structure into the other. The SISU is

capable of transforming the messages exchanged

between different software applications that

interoperate by exchanging information, therefore

understanding each other messages despite the

difference in their respective data organizations.

2.3 Semantic Value Added Services

(SVAS)

This third layer of semantic services is still of a

general nature, but the offered facilities are able to

concretely contribute in producing value for the

enterprises. We briefly list some of them.

- Semantic Search and Retrieval. This will be the

new frontier of search engines. The injection of

semantics in search engines will significantly

improve the performances, effectiveness, and

the user satisfaction. Especially when

developed jointly with user profiles (see the

next point).

- Semantic User Profiling and User-centred HCI

(Human-Computer Interaction). User profiling

is a very promising area. But the developed

solutions will be even more effective, if the

user profile will be enriched with semantic

annotations. In particular, when a semantic user

profile will be used for search purposes, the

semantic search engine will be able to consider

it, jointly with, for instance, contextual

information, to rewrite the user query and to

optimize its execution.

- Staffing and Experts Team Building. The

semantic profiles (including competences and

skills) can be used for the optimal composition

of working groups, where the gathered

competencies and skills are suitably blended

with respect to the activities to be performed.

- Enterprise Consortium Building. Here we

change scale, moving up to the level of a

consortium of enterprises. When a consortium

is built to respond to a business opportunity

(e.g., a public call for tender) it is necessary

that the gathered enterprises show a good

coverage of the capabilities required in the call.

A joint semantic analysis of the Call and the

enterprise profiles will provide important

elements to proceed in the formation of the

consortium.

2.4 Semantically Enhanced

Applications Services (SEAS)

Here we address specific (vertical) enterprise

applications, such as Accounting, HR Management,

Production Planning, Sales and Distribution.

Semantic technologies can have a wide potential

impact, empowering all possible enterprise

applications. Some enterprise applications will

deeply change with the injection of semantic

capabilities, but some other will simply disappear to

be substituted by new integrated Business-IT

solutions, not conceivable without the use of the

semantics. The innovative solutions will emerge

from the joint use of semantics and Web 2.0, user-

centred social software, significantly impacting

different industrial sectors. Among the most

innovative vertical applications, we may cite:

- Intelligent autonomic logistics systems

- Disaster and emergency prevention and

management

- Advanced cross sectorial health care

- Cognitive economics

3 THE KNOWLEDGE

ENTERPRISE NETWORK

The full achievement of the knowledge enterprise,

based on the semantic technologies, cannot be

reached with the enterprises we know today.

3.1 In Search for New Enterprise

Models

In parallel to the technological innovation, it is

necessary that the organization and operational

models of the enterprise undergo a deep change as a

precondition for the full deployment of the potential

of the knowledge infrastructures. In this perspective,

we may recall the vision of Stafford Beer, rephrasing

it in the perspective of the semantic enterprise:

asking how to use semantic technologies in the

enterprise is a wrong question, the right question is

what will be the transformations that semantic

technologies will induce on the enterprises. But,

even more correct is to ask what will be the

enterprise of the future once that semantic

technologies will be fully deployed. Then, let’s try

to depict the main lines of such a possible future.

In the previous section we presented a framework

for the development of enterprise wide software

applications based on the extensive use of semantic

services. In fact, Figure 2 represents the layers of a

Semantic SOA that will be realised in the next

decade or so. In this perspective, the semantics is

mainly involved in the achievement of the advanced

software services, but not in the production of the

available services: there is still a significant amount

of software to be developed, debugged, tested, and

maintained.

3.2 From Software Programming to

Knowledge Representation

Pushing further our vision, we can envisage the

future knowledge enterprise network, where new

paradigms for service development will be totally

based on the enterprise semantics. The resulting

services will not be coded with the traditional

software techniques. Enterprise IT applications will

be characterised by a separation between business

logic and business operations, pushing the MDA

(Model Driven Architecture) to its full

accomplishment. Essentially, the business logic,

including strategies, rules, and high level best

practices, will be represented by Semantic Business

Processes (SemBP), while activities and operations

will be represented by Semantic Business Services

(SemBS). The latter, if necessary, will be recursively

expanded, showing the internal structure in the form

of more detailed SemBPs, while the components

will be lower level SemBSs, until atomic SemBSs

will be reached. The latter are SemBSs that can be

fully (and operationally) specified in one

unambiguous step (human or automatic), to be

executed either directly in the enterprise or by an

external entity.

SemBP and SemBS will be (semi)automatically

derived from the extensive repositories of enterprise

documents. They will be represented in a declarative

form, e.g., rule-based, with the support of reference

ontologies.

The supporting networked infrastructure will be the

evolution towards a semantic version of the FInES

(Future Internet Enterprise Systems) that we know

today

. A possible global architecture has been

suggested by W3C, with the Semantic Web “cake”.

http://cordis.europa.eu/fp7/ict/enet/ei_en.html

3.3 The Unified Enterprise Knowledge

Space

The key idea of the scenario reported above is a tight

integration of human and computer knowledge, with

synergic capabilities of understanding and

proactively using the available shared knowledge: a

symbiosis between natural and artificial intelligence.

We assume that the principles, rules, and operations,

according to which an enterprise functions and

produces value, are all represented in a form or

another in human-oriented documents. From

business processes to roles and positions of the

personnel, from marketing strategies to assets

management, more and more there is a consistent (or

supposed so) production of strategic, tactical, or

operational documents. We can imagine that the

semantic technologies will be able to extract the

knowledge therein reported and codify it in the

forms that can be interpreted by a computer system.

In essence, when the management of an enterprise

decides to introduce a change in the operations of

the organization, this intention is reported in one or

more documents. These documents will be

transmitted to the interested sectors and people

therein, who will modify organizations, operations,

and employees behaviour accordingly. But in

parallel, there is a need of changing the enterprise

application systems for the parts affected by the

above mentioned decisions. In the knowledge

enterprise, the management documents will be

semantically analysed, the knowledge mining

services will extract the new instructions to be

propagated to the operational knowledge repositories

(i.e., those containing the SemBP and the SemBS).

Semantic Matchmaking services will discover where

(i.e., on which processes and services) the new

directions will impact. Then, the updated operational

knowledge will be activated, to guarantee that the IT

enterprise applications will behave according to the

new directions (in this way, the well known

Business/IT alignment will be largely solved).

4 THE OBSTACLES TO THE

ADVENT OF KEN

The progress towards the new Knowledge-centric

Enterprise Network (KEN) models, made possible

by the extensive use of knowledge and semantic

technologies, has been sought since long time. More

than a decade ago there have been the first important

results in this direction (see for instance (Fox et al.,

1998) and (Uschold et al., 1998). Similarly, the

Semantic Web has been proposed at the beginning

of this decade. The decade of the 90s has been

characterised by the explosion of the Web, the

current decade is characterised by the Social Web.

Many say that the upcoming decade will be finally

that of the Semantic Web and, as its natural

consequence on the business world, it will see the

advent of the Knowledge Enterprise Network. We

know that the technological innovation is

unstoppable, however there are several factors that

may delay the joint evolution of the technologies

and the enterprises along the lines described above.

Here we can briefly summarise a few of the

hindering factors. Some of them are real, some other

depend on the wariness of the majority of the

businesses.

Knowledge Culture. We need that the enterprises

develop a diffused culture of semantics, a

better awareness of the advantages that a

diffusion of such technologies will bring to

enterprises, and to the society at large.

Semantic technologies are still considered as

research artefacts, far from the practical and

convenient level of maturity, in order to be

actually adopted by enterprises.

Semantic technologies are very demanding in terms

of processing power and storage systems. We

still need a great deal of research before their

extensive application in real industrial settings.

To adopt semantic solutions, and enterprise

needs a group of highly specialised experts,

such as ontology engineers or logicians.

Last but not least, costs and risks still appear high

with respect to the expected (but not concretely

proved) benefits.

5 CONCLUSIONS

In this paper we sketchily presented the main lines

of a possible future, where new technological

solutions will concur in the realization of a unified

semantic space. In this space humans and computers

will have access to the same knowledge, being

therefore able to tightly cooperate in different high

level activities. In such a reality, there will be a

parallel evolution of the business dimension, new

enterprise models will emerge and the notion of

value will be different from our understanding of

today.

We are at the verge of a new decade witnessing an

acceleration of the progress towards the Knowledge

Economy. But we know that the progress hardly

follows a linear trajectory, and there is no positive

determinism in the technological evolution.

Therefore, as usual, we are the creators of our own

fortune.

ACKNOWLEDGEMENTS

I wish to thank the colleagues Antonio De Nicola e

Francesco Taglino who helped me to better focus the

vision reported in this paper. Furthermore, I wish to

thank Sergio Gusmeroli for the long and open

discussions we had in digging the frame and mission

of the COIN European Project, that influenced the

layered service-oriented vision reported in Section 2.

REFERENCES

D. Bianchini, S. Castano, F. D'Antonio, V. De Antonellis,

M. Harzallah, M. Missikoff, S. Montanelli; Digital

resource discovery: semantic annotation and

matchmaking techniques, In proceedings of Interop-

ESA06 Conference, Bordeaux, March 2006.

M. Fox, M. Gruninger; Enterprise Modeling. AI

Magazine, V. 19 n. 3, AAAI Press, 1998.

Fabriani, M. Missikoff, P. Velardi: “Text Mining

Techniques to Automatically enrich a Domain

Ontology", Applied Artificial Intelligence Journal,

Vol. 18, No. 3, May-June 2003.

T. Berners-Lee, J, Hendler and O. Lassila; The Semantic

Web. Scientific American, May 2001.

J. Barkley Rosser; Econophysics. In The New Palgrave

Dictionary of Economics, 2nd edition, January 2006.

Nigel Shadbolt, Wendy Hall, Tim Berners-Lee; The

Semantic Web Revisited. IEEE Intelligent Systems,

May-June 2006.

M. Tegmark; The Mathematical Universe. Foundations of

Physics, V.38 n.2, February 2008.

M. Uschold, M. King, S. Moralee, Y. Zorgios; The

Enterprise Ontology. The knowledge engineering

review, Cambridge Univ Press, 1998

E. Von Schweber; Roadmap for Semantics in Netcentric

Enterprise Architecture. US General Services

Administration Report, February 2006.

BRIEF BIOGRAPHY

Michele Missikoff, Director of Research at IASI

(Istituto di Analisi dei Sistemi ed Informatica) of

National Research Council, and Coordinator of

LEKS: Lab for Enterprise Knowledge and Systems

since 1999. Scientific Director of the European

laboratory on Enterprise Interoperability, Interop-

VLab, has a long experience in European projects,

both in eBusiness and eGovernment. Elected

member of the National Steering Committee for

Information Science and Technology, from 1992 to

1997. He has been member of the editorial Board of

VLDB Journal and co-founder of EDBT Conference

series. Past-president of EDBT Endowment, he has

organised numerous conferences and workshops. He

has pubblished more than 150 technical and

scientific papers, the majority of which on

international journals and conference proceedings.