A Pragmatic Approach to Conceptual Negotiation Support

Cristóvão Sousa

1,2

, Carla Pereira

1,2

and António Lucas Soares

2,3

CIICESI - ESTGF, Polytechnic Institute of Porto, Rua do Curral, Casa do Curral-Margaride,

4610-156, Felgueiras, Portugal

INESC Porto, Rua Dr. Roberto Frias, s/n 4200, Porto, Portugal

Department of Informatics Engineering, Faculty of Engineering, University of Porto,

Rua Dr. Roberto Frias, s/n 4200, Porto, Portugal

Keywords: Conceptualisation Process, Ontology Engineering, Conceptual Negotiation, Knowledge Representation.

Abstract: Collaborative conceptualisation processes are pervasive to most technical and professional activities, but are

seldom addressed explicitly due to the lack of theoretical and practical methods and tools to support it.

However, it seems not to be a popular research topic in knowledge representation or its sub-areas such as

ontology engineering. Our view is that collaboration between stakeholders for specifying an ontology

should be addressed at the conceptual, semi-formal level, in order to foster a collective learning of the

domain and reaching agreements about its representation. We developed a method to support conceptual

integration based in the conceptual blending theory - ColBlend - and implemented it in a collaborative

modelling environment. This "conceptual modelling environment - conceptME" supports teams of specialist

and facilitators in eliciting conceptual structures with the help of collaborative model editing and

terminology services. Conceptual integration and agreements are achieved through the ColBlend method.

This paper overviews ColBlend and ConceptME and describes in detail a test case.

1 INTRODUCTION

Representing knowledge through ontologies usually

requires domain experts to commit to some

particular formalism, which could derail or at least

delay the overall process of achieving a shared

representation of concepts and relationships between

concepts. Unfortunately, is evident that "While

different degrees of formalization have been well

investigated and are now found in various ontology-

based technologies, the notion of a shared

conceptualization is neither well-explored, nor well-

understood, nor well-supported by most ontology

engineering tools” (Staab, 2008). We confirm that

current knowledge about the early phases of

ontology construction is insufficient to support

methods and techniques to support collaborative

conceptualisation processes. Trying to address the

above identified gap, our research focus on the study

and support to collaborative conceptualisation

processes. In relation to an individual, a CP is a set

of cognitive activities that has as inputs information

and knowledge internally or externally accessible,

and as the output an internal or external conceptual

representation. Furthermore, a “collaborative

conceptualisation process” (CCP) is a CP that

involves more than one individual producing an

agreed conceptual representation. It involves social

activities that include meaning negotiation and

practical management activities for the collaboration

(Pereira et al., 2012). CP should account for

mechanisms to deal with the inputs for the process

once they are crucial to support the

conceptualisation tasks. The research reported in this

paper addresses the support to the CCB. From a

methodological approach developed in previous

research (Pereira and Soares, 2008), the so-called

ColBlend method, we developed a modelling

environment - conceptME - supporting the

collaborative creation, editing, discussion and

negotiation of conceptual representations (e.g.,

concept maps, topic maps, UML diagrams). Besides

being based on informal knowledge representation

notations, more close to the users cognitive

representation of a conceptualisation, our approach

also provides support to the externalisation of

concepts and relations. For this, conceptME

provides a suite of terminological services allowing

the users to get help from the processing of a

previously setup textual corpus about the domain to

349

Sousa C., Pereira C. and Lucas Soares A..

A Pragmatic Approach to Conceptual Negotiation Support.

DOI: 10.5220/0004158603490354

In Proceedings of the 14th International Conference on Enterprise Information Systems (SCOE-2012), pages 349-354

ISBN: 978-989-8565-11-2

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

be worked out. More specifically, this paper

illustrates the use of the ColBlend method (see

section 2), contributing to its improvement and

refinement, in particular to what concerns to the

implementation of the blend space by means of

terminology based techniques assisting users in their

conceptualisations activities - more specifically

concept elicitation and concept discussion.

2 SUPPORTING BLEND SPACE:

CONCEPTME APPROCAH

2.1 Revisiting ColBlend Method

ColBlend method (see fig. 2) was designed to

support a collaborative conceptualisation process,

based on conceptual blending theory (CBT)

(Fauconnier et al. 1998) (see fig. 1). In practical

terms ColBlend aims at supporting the co-

construction of an agreed set of conceptual models,

which could be translated into taxonomies,

glossaries or ontologies.

Figure 1: CBT.

In (Pereira et al., 2012) ColBlend is detailed. In

an overview, the process comprises a set of virtual

spaces: a) the input spaces - where each party build

models representing their conceptualisation of the

domain; b) the blend space - containing the results

from the analysis of the input spaces presented for

discussion. Moreover, it propose new concepts from

a global analysis of the current spaces content and;

c) the generic space - which contains the common

domain model composed by the all parts of the

proposals that were accepted by all and "published"

to the this shared space.

ColBlend led to the development of the

conceptME (www.concepme.pt). In this paper, we

Figure 2: ColBlend method.

present the results of the implementation of the

blend space (negotiation support and decision-

making space) using coordinated corpus-analysis

and knowledge representation tasks (the CP main

concern). The implementation of blend space is

achieved by means of techniques to retrieve

immediate contexts of terms, assisting users on their

conceptualisations activities - more specifically

concept elicitation and concept discussion

(designation according to the conceptualisation

framework described in (Sousa et al., 2012).

2.2 ConceptME Approach

The core of conceptME platform is on supporting

collaborative modelling, focusing on graphical

knowledge representations and terminological

methods, accommodated into a Library of services.

The platform is organised as follows (see fig. 3): a) a

set of functionalities to manage collaborative

modelling projects; b) a collaborative modelling

environment, allowing users to build their models

individually or editing them collaboratively (either

on their own or through available templates), while

discussing around concepts; c) a set of

terminological services, supported by a domain

specific textual corpus, allowing users to associate

relevant resources to their projects, performing

extraction operations to retrieve candidate terms that

can be used in their CP.

At this level, conceptME provides: i) means for

corpus organisation and classification; ii) real-time

term contexts to detail existing representations; d) a

model negotiation baseline to ensure simple

negotiation mechanisms, towards a common shared

model. This module provides the interface and

ICEIS 2012 - 14th International Conference on Enterprise Information Systems

Figure 3: Conceptme high-level architecture.

environment conditions, allowing to connect other

advanced negotiation mechanisms, despite of their

nature and/or domain.

2.3 Discussing and Collaborating

through Concepts

In accordance with the principles of the CBT (see

fig. 1) and the ColBlend method (see fig. 2), the

blend space is implemented by means of three key

activities (Fauconnier et al., 1998): composition,

completion and elaboration. With the goal to

supporting the discussion and collaboration in a

collaborative conceptualisation process these

activities were designed as follows: The composition

activity comprises the identification of the

“counterpart” elements (concepts of the input spaces

subsumed by concepts of the generic space) of the

individual proposals. This calls for specific

algorithms to cope with the analysis of common

elements between the available models. Available

methods in ontology merging perform a one-to-one

approach (Doan et al., 2004). Within collaborative

settings, where multiple proposals could be

available, the question about which pair of models

should be analyzed arises. This issue could be

mitigated by following a clustering based approach

(Araujo et al., 2010). Right before individual

proposals appear in the blend space, an intermediate

action is triggered whose goal is to provide

assistance on defining the order in which

composition is performed over the input conceptual

structures (ICS) - conceptualisation proposals

according to figure 2. Having the elements of each

proposal indexed, clustering techniques categorize

the ICS into clusters, and then the composition

activity will start from the cluster with the highest

score. The user could either select the ICS to be

analyzed for counterpart elements identification or

could accept the platform suggestions. These

clustering techniques are available through Solr

(http://lucene.apache.org/solr/) by means of carrot

clustering engine (http://project.carrot2.org/). As

mentioned before, the first step towards blend space

is performed by a composition activity. Regarding

the conceptME approach, this is accomplished by

merging the individual proposals and placing, at the

blend space, a model containing the common

elements. The other elements are highlighted,

becoming potential negotiation targets. Hereafter,

comes completion activity comprising a cross-

checking corpus-based validation over the elements

contained in the ICS. The goal is to provide data to

support the decision about the inclusion of the non-

shared but non-conflicting elements of the ICS in the

blend space. The elaboration activity, by its turn,

aims at new concept discovery. This is accomplished

by a broader context retrieval, whose focus goes

beyond the scope of the elements enclosed in the

ICS. Term contexts could be immediately retrieved

in order to get clues on possible new concepts or

relations related to the current concepts and to the

corresponding conceptual representation. Moreover,

term contexts could additionally support discussion

around a specific concept, justifying its use by

showing evidence about term occurrence in corpus

or infer on concept semantic metadata (using an

RDF triple store) or even highlighting patterns

(<noun><verb><noun>) within text, indicating the

incompleteness of the overall conceptual structure

and suggesting new elements to negotiation. The

goal is to identify terms (nouns) that co-occur with

some other term in the available structures and

possible linking phases to connect them. The linking

phases are typically verbs or expressions that match

the following lexical patterns: i) verb preceded or

followed by a preposition or subordinating

conjunction; ii) a verb preceded or followed by a

coordinating conjunction; iii) a verb preceded or

followed by a “TO”; iv) a verb preceded or followed

by a determiner and; v) a verb preceded or followed

by another verb. These patterns are implemented as

xml queries (Xquery/Xpath) and/or regular

expressions. A practical experiment resulting of the

blend space execution is presented in the next sub-

section.

3 ILUSTRATION OF THE

APPROACH

Some experiments were performed exploiting the

creation of a corpus about the domain of urban

rehabilitation (http://www.h-know.eu/). Three

documents selected from the urban rehabilitation

corpus were posted (and indexed) on Solr. During

A Pragmatic Approach to Conceptual Negotiation Support

the conceptualisation process, users added concepts

and relationships, either on their own or calling upon

extraction services or even through available

templates. After the release of a term on canvas,

several tasks may follow: a) Provide a definition for

the term; b) complete the structure adding another

term and a linking phrase between two terms.

Context-analysis could help on both. Still in the

scope of urban rehabilitation domain, at a certain

stage of the CP, the definition of the conceptual

structure around the Moisture Control concept was

started and a set of input structures were achieved

(see fig. 4a – 4b – 4c).

Figure 4a: Input conceptual structure 1.

From the available proposals, Input conceptual

structure 1 (fig. 4a) and Input conceptual structure 2

(fig. 4b) are those containing the largest number of

counterpart elements, thus, composition will perform

over them in first place, unless user has decided

differently. In a first iteration, a set of elements was

found as being common (Moisture Load, Climate

Conditions, Construction Type, Moisture Control

and the linking phrase depends on) and was

proposed to be merged.

Figure 4b: Input conceptual structure 2.

Figure 4c: Input conceptual structure 3.

Figure 4d: Blended conceptual structure.

The elements Internal Source, External Source,

Construction Moisture and the linking phrases has

and is a, were proposed to be appended to the

resulting structure. The elements in conflict

(responds to and controls) were proposed for

discussion. Afterwards, completion will perform and

the blend space begins to emerge. Grounded on

corpus, completion activity checks the composition

result performing analysis over the common base

information such as: source documents, co-

occurrences of the terms in corpus through context

retrieving and available definitions, but focusing

only on the elements of the ICS. Regarding context

analysis information depicted in table 1, the

elements: Construction Moisture, Climate

Conditions, Internal and External Sources, remained

in the blend space. Moreover, Moisture Control

prevails over Moisture Control Strategies, since the

first term appears in corpus unlike the second one

ICEIS 2012 - 14th International Conference on Enterprise Information Systems

(see table 1). The same happened to responds to and

controls linking phrases. Once “responds to” was

found as contexts were retrieved, it prevails over

“controls” relation. Regarding the structure

“Moisture control acts on Building Constructions”, it

emerges along with the elaboration activity, whilst

contexts are processed to reveal specific patterns in

the form of <noun>[prefix]<verb>[suffix]<noun>

(prefix and suffix are optional and regards to the

lexical patterns mentioned earlier in section 3.2). In

order to avoid uninteresting results, pattern

discovery services considered that, at least a <noun>

was already in use in some input structure. Further

iterations could disregard such requirement.

Table 1: Excerpt of Moisture control contexts.

Textual corpus Context for “Moisture Control” concept

Effective moisture control has to respond to the exterior as

well as the interior moisture

loads acting on building constructions.

However, good moisture control design depends on a

variety of parameters such

as climate conditions and construction type which changes

from region to region.

Rehabilitation guidelines and moisture

control Some heritage constructions are more vulnerable

to moisture loads than modern buildings.

4 RELATED WORK

The work on ontology engineering methodologies

has been extensively discussed and compared in

(Corcho, 2003), (Fernández-López, 2002), (López et

al., 1999) and (Gasevic et al., 2006). We can

conclude that the ontology-engineering field has laid

a lot of emphasis on the “specification of the

conceptualization” as an engineering task.

Nevertheless, the early phases of the ontology

development life-cycle have been poorly addressed.

In particular, the importance of the social processes

involved in the formation of a collective

conceptualization (e.g., of a domain) has not been

recognized. Dealing more directly with collaboration

in ontology development (see (Simperl et al., 2006),

(Kotis et al., 2006), (Aschoff, 2004), (Zhao, 2005),

(de Moor et al., 2006), (Staab et al., 2001), (Sure,

2002), (Gómez-Gauchía et al., 2004), and (Pinto et

al., 2004) for a complete account of those

approaches), lead to the conclusionthat few research

works recognise the importance of supporting the

collective construction of a conceptualization. Some

particular questions come out from this review: (i)

the importance of representational tools and user

interfaces for interacting with knowledge

representations are generally underestimated; (ii)

negotiation and consensus building regarding the

conceptualization content has not been a priority

either; there are a few proposals that claim to

support the process of reaching consensus or

agreements, but only one addresses the issue of what

conceptual content should be included in the shared

conceptualization; (iii) the reutilization of existing

ontologies is an obvious requirement; nevertheless,

there is not any approach that integrates reutilization

with the conceptualization building in a systematic

way.

5 CONCLUSIONS

This paper addresses the support to collaborative

conceptualisation processes, an overlooked research

topic in ontology engineering and knowledge

representation in general. It showed that conceptual

agreements within a group developing initial models

for the creation of an ontology, can be improved

through the semantic processing of input models

based on the Conceptual Blending Theory.

Furthermore, the combination of this conceptual

integration with basic terminological approaches,

providing assistance in concept elicitation and

validation, makes this approach quite innovative.

The conceptME platform is already available, is free

to use and we are planning to realise it as an open

source project. Future work will be focused on the

running of empirical experiments with two aims: (i)

to deepen the knowledge about collaborative

conceptualisation processes and (ii) to continue the

development of conceptME, specifically in the

terminological support and conceptual negotiation

techniques. As someone said (Floridi, 2008),

"humans are the only semantic engines available",

thus we are strongly embracing a socio-semantic

perspective in the development of tools for

collaborative knowledge representation.

ACKNOWLEDGEMENTS

This work is funded by the ERDF through the

Programme COMPETE and by the Portuguese

Government through FCT - Foundation for Science

and Technology, project PTDC/EIA-EIA/103779/

2008 "CogniNET".

REFERENCES

de Moor, A., Leenheer, P. D., Meersman, R. (2006).

A Pragmatic Approach to Conceptual Negotiation Support

Dogma-mess: A meaning evolution support system for

interorganizational ontology engineering. Proc. of the

14th Inter. Conf. on Conceptual Structures (ICCS

2006), Aalborg, Denmark, July 17-21.

Kotis, K., Vouros, G. (2006). Human-centered ontology

engineering: The HCOME methodology. Knowledge

and Information Systems, vol. 10 (1), pp. 109 – 131.

Staab, S., Studer, R., Schnurr, H., Sure, Y. (2001).

Knowledge processes and ontologies. IEEE Intelligent

Systems, vol. 16 (1), pp. 26 – 34.

Sure, Y. (2002). A tool-supported methodology for

ontology-based knowledge management. In ISMIS

2002, Methodologies for Intelligent Systems.

Gómez-Gauchía, H., Díaz-Agudo, B., González-Calero, P.

(2004). Towards a pragmatic methodology to build

lightweight ontologies: a case study. In Procs. of the

IADIS - 2004, Lisboa, Portugal.

Pinto, S., Staab, S., Sure, Y., Tempich, C. (2004).

Ontoedit empowering swap: a case study in supporting

distributed, loosely-controlled and evolving

engineering of ontologies (diligent). In 1st ESWS

2004, Springer.

Aschoff, F. (2004). Knowledge mediation: A procedure

for the cooperative construction of domain ontologies.

Master's thesis, University of Heidelberg.

Zhao, G. (2005). AKEM: an ontology engineering

methodology in FF POIROT. Deliverable 6.8 of FF

POIROT project.

Corcho, O., Fernández-López, M., Gómez-Pérez, A.

(2003). Methodologies, tools and languages for

building ontologies: where is their meeting point?,

Data Knowl. Eng., vol. 46 (1), pp. 41 -- 64.

Fernández-López, M., Gómez-Pérez A. (2002) Overview

and analysis of methodologies for building ontologies.

The Knowledge Engineering Review, vol. 12 (2), pp.

129 – 156.

López, M., Gómez-Pérez, A., Sierra, J., Sierra, A. (1999).

Building a chemical ontology using methontology and

the ontology design environment. IEEE Intelligent

Systems, vol. 14 (1), pp. 37 -- 46.

Gasevic, D., Djuric, D., Devedzic, V. (2006). Model

Driven Architecture and Ontology Development.

Springer-Verlag, Berlin.

Simperl, E., Tempich, C. (2006). Ontology Engineering: A

Reality Check. In R. Meersman and Z. Tari and others,

The 5th Intenational Conference on Ontologies,

DataBases, and Applications of Semantics

(ODBASE2006), volume 4275 of Lecture Notes in

Computer Science (LNCS), Springer, Montpellier,

France.

Staab, S. (2008) On understanding the collaborative

construction of conceptualisations. International and

Interdisciplinary Conference "Processing Text-

Technological Resources" at the Center for

Interdisciplinary Research (ZiF - Zentrum für

interdisziplinäre Forschung), Bielefeld University, 13-

15 March.

Fauconnier, G., Turner, M. (1998). Conceptual Integration

Networks. Published in Cognitive Science, vol. 22 (2),

pp. 133 – 187.

De Araujo, F. F., Lopes, F. L. R. & Loscio, B. F., 2010.

MeMO: A Clustering-based Approach for Merging

Multiple Ontologies. In IEEE, pp. 176–180.

Doan, A., Madhavan, J., and Halevy, A., 2004. “Ontology

Matching: A Machine Learning Approach”. In S.

Staab and R. Studer (eds.), Handbook on Ontologies in

Information Systems (pp. 385-403). New York:

Springer.

Sá, C., Pereira, C., Soares, A. (2010. Supporting

collaborative conceptualization tasks through a

semantic wiki based platform. OTM 2010 Workshops

Lecture Notes in Computer Science, 2010, Volume

6428/2010, 394-403.

Pereira, C., Sousa, C., Soares, A. (2012). (Interoperable)

Supporting conceptualisation processes in

collaborative networks: a case study in an R&D

project. International Journal of Computer Integrated

Manufacturing. Special Issue Paper.

Sousa, C., Soares, A., Pereira, C., Costa, R. (2012).

Integrating terminological methods in a framework for

collaborative development of semi-formal ontologies”.

In Proceedings of the colabTKR - collaboration in

Terminology and Knowledge, part of the LREC 2012.

Floridi, L., The Semantic Web vs. Web 2.0: a

Philosophical Assessment, Episteme, 2009, 6, 25-37.

Pereira, C., & Soares, A. L. (2008). Ontology

Development in Collaborative Networks as a Process

of Social Construction of Meaning. In R. Meersman,

Z. Tari, & P. Herrero (Eds.), On the Move to

Meaningful Internet Systems: OTM 2008 Workshops,

Vol. 5333. Springer Berlin / Heidelberg, pp. 605-614.

ICEIS 2012 - 14th International Conference on Enterprise Information Systems