CONCEPTUALIZATION AND VISUAL KNOWLEDGE ORGANIZATION: A SURVEY OF ONTOLOGY-BASED SOLUTIONS

Abstract

Conceptualization and Visual Knowledge Organization are overlapping research areas of Intellect Augmentation with hot topics at their intersection which are enhanced by proliferating issues of ontology integration. Knowledge organization is more closely related to the content of concepts and to the actual knowledge items than taxonomic structures and ‘ontologies’ understood as “explicit specifications of a conceptualization” (Gruber). Yet, the alignment of the structure and relationship of knowledge items of a domain of interest require similar operations. We reconsider ontology-based approaches to collaborative conceptualization from the point of view of user interaction in the context of augmented visual knowledge organization. After considering several formal ontology based approaches, it is argued that co-evolving visualization of concepts are needed both at the object and meta-level to handle the social aspects of learning and knowledge work in the framework of an Exploratory Epistemology. Current systems tend to separate the conceptual level from the content, and the context of logical understanding from the intent and use of concepts. Besides, they usually consider the unification of individual contributions in a ubiquitous global representation. As opposed to this God's eye view we are exploring new alternatives that aim to provide Morphic-like live, tinkerable, in situ approaches to conceptualization and its visualization and to direct manipulation based ontology authoring capabilities. They are uniformly applicable and user-extendable through meta-design both at the content, as well as all emergent conceptual meta-levels, within collaborative frameworks like Conceptipedia. In such a framework negotiation games played at the meta-level of knowledge items can support emergent conceptualization until viable alternative taxonomic structures emerge and workable situated ideas win out. Distributing the workload and responsibility through Crowd Authoring, concept alignment and challenges of social, collaborative concept matching and revision can be addressed in a knowledge management kernel which is also capable to supplement education technology.
Keywords: Conceptualization, Visual Knowledge Organization, Exploratory Epistemology, Ontology Engineering, Ontology Integration, Traversal Frequency, Knowledge Graph, Morphic, Augmentation

Full text

CONCEPTUALIZATION AND VISUAL KNOWLEDGE ORGANIZATION:
A SURVEY OF ONTOLOGY-BASED SOLUTIONS
A. Benedek1, G. Lajos2
1 Hungarian Academy of Sciences (HUNGARY)
2 WikiNizer (UNITED KINGDOM)
benedek.andras@btk.mta.hu; gyuri.lajos@wikinizer.com
Abstract
Conceptualization and Visual Knowledge Organization are overlapping research areas of Intellect
Augmentation with hot topics at their intersection which are enhanced by proliferating issues of ontology
integration. Knowledge organization is more closely related to the content of concepts and to the actual
knowledge items than taxonomic structures and ‘ontologies’ understood as “explicit specifications of a
conceptualization” (Gruber). Yet, the alignment of the structure and relationship of knowledge items of a
domain of interest require similar operations. We reconsider ontology-based approaches to collaborative
conceptualization from the point of view of user interaction in the context of augmented visual knowledge
organization. After considering several formal ontology based approaches, it is argued that co-evolving
visualization of concepts are needed both at the object and meta-level to handle the social aspects of
learning and knowledge work in the framework of an Exploratory Epistemology. Current systems tend to
separate the conceptual level from the content, and the context of logical understanding from the intent and
use of concepts. Besides, they usually consider the unification of individual contributions in a ubiquitous
global representation. As opposed to this God's eye view we are exploring new alternatives that aim to
provide Morphic-like live, tinkerable, in situ approaches to conceptualization and its visualization and to
direct manipulation based ontology authoring capabilities. They are uniformly applicable and user-extendable
through meta-design both at the content, as well as all emergent conceptual meta-levels, within collaborative
frameworks like Conceptipedia. In such a framework negotiation games played at the meta-level of
knowledge items can support emergent conceptualization until viable alternative taxonomic structures
emerge and workable situated ideas win out. Distributing the workload and responsibility through Crowd
Authoring, concept alignment and challenges of social, collaborative concept matching and revision can be
addressed in a knowledge management kernel which is also capable to supplement education technology.
Keywords: Conceptualization, Visual Knowledge Organization, Exploratory Epistemology, Ontology
Engineering, Ontology Integration, Traversal Frequency, Knowledge Graph, Morphic, Augmentation
1 ONTOLOGIES AND INFORMATION ARCHITECTURES
1.1 The Metaphysical Heritage
Ontological research has a long history, from early reflections by Aristotle in his Metaphysics, Christian
Wolff’s use of the term in the C18th for the “study of being in general”, to its more recent upsurgence
in computer science, along with linguistics and conceptualization issues within a variety of other
sciences such as physics, cognitive psychology and genetics. Ontology Engineering (OE) has been
closely bound up with theories of formal ontology, which describe the basic categories of being, the
general features of reality and their relations. This focus on reality can be contrasted with recent
developments in the field of information organization. The “metaphysical heritage” of OE supplies us
with domain-independent systems of “upper ontologies”, which is to say high level categorical systems
which support the conceptualization of domain-specific knowledge and special purpose “domain
ontologies” in various specialized fields. This line of OE can be called ‘metaphysical’ because
ontologies in this tradition, be they scientific, philosophical, or derived from our common sense,
assume some kind of adequacy or correspondence between the categorical representations of a given
domain, and “the real world itself”. [3, 5] They seek entity taxonomies at every level of complexity, and
apply “adequatism”, or “substantialism” or other ontical doctrines. [10, 11, 12, 7, cf. 5] These approaches
readily ally with some form of “Conceptual Atomism” in philosophy and the natural sciences, and they
also adopt Representational Theories of Mind from the cognitive sciences. [2, 6] The most basic
categories of “being in general” however are far from uniform: the “highest” level of ontologies differ in

their universals, both in their abstract and concrete modes of existence, and their endurants and
perdurants, with the consequence that there are alternative top level terms and descriptive systems of
foundational ontologies (e.g., BFO, DOLCE, GFO, PROTON, SUMO or Sowa’s ontology [9]).
1.1.1 Structural Domain Models
Different epistemological positions are reflected in the definitional differences between ontologies
which treat conceptual systems as reflecting the structure of reality independently of their (vernacular
or formal) linguistic representation, and language dependent systems which determine what
descriptive categorizations of a domain are admissible. In the former approach modeling
methodologies and ontology representation languages are primarily evaluated in terms of their
completeness and expressibility in addressing a state of affairs in a specific domain at different levels
of granularity [20]. The latter however are treated in terms of computability issues, and standards of
usability comprehensibility, and domain appropriateness for the purpose of conceptualization. [15]
Domain models (which are also called semantic data models and structural conceptual models) are
generally taken to be conceptual representations of a state of affairs in the real world or a system of
categories which can be used to describe aspects of the structure of a knowledge domain. [8, 14, 32]
They consist of general notions about the types of entities in that world and their instances (e.g.,
classification of objects and their intrinsic properties represented in value spaces); sorts of types (e.g.,
kinds, roles, phases, mixed contingent collections) and the relations between these entities (e.g.,
identity and the properties of relations). Since various sorts of categories and their relational properties
(rules/axioms defining the properties of relations) determine the admissible relations and their
instantiations, conceptualization clusters together sets of terms which articulate the abstract structure
of reality, creating a formal model of its state of affairs. These models, or rather the model specifications,
are formal representations of the conceptualization. [Cf. Figure 1. below, adapted from 3, p. 50]
1.1.2 Foundational and Domain Modeling Languages
General entity–relationship models [28] and domain specific modeling languages which have faced
problems of expressiveness and relevance were early developed to represent model specifications (e.g.,
LINGO, GLEO and IDEF5 for graphical expression of ontologies) before the prevalence of OWL and the
use of UML for ontology modeling. [22, 29, 32] Since a modeling language determines the grammatically
correct specifications which can be constructed in that language (limiting the space of possible
conceptualizations) the conceptualization formulated in the admissible terminology of the given modeling
language fixes the set of all possible models of states of affairs of the world or a given problem domain.
[60] The Modeling Language icon depicted in Figure 1. represents the specification of the conceptual
model underlying the language, or what following Guizzardi can be called ‘metamodel specification’. [3]
Formal Ontological theories, just as information systems grammars that served the evaluation and
redesign of these languages focus on specific sets of concepts in order to equip them with a reference-,
or an adequate real-world semantics. [3, 29] The latter usually employed “… an ontology named BWW
(Bunge-Wand-Weber) which is based on the original metaphysics proposed in [7]…that is … committed
to capturing the intrinsic nature of the world in a way that is independent of the conceptualizing agent,
and, consequently, an approach in which cognition and human language play a minor or non-existent
role.” [3, p. 9, italics added] Information systems grammars, reference models and communication
purposes forced us to further integrate language design, for the sake of general and universal
descriptions of models and categorical systems. The foundational ontologies that Guizzardi, and Masolo
et al. describe [3, 44] give us reference models which prescribe the general categories of a specific
conceptualization within a system of the formal relations of linguistic terms which a modeling language
admits satisfying general formal requirements for modeling languages. The primary objective of
Guizzardi [3] is “to establish a systematic relation between a modeling language and a reference
ontology, and to propose a methodological approach to analyze and (re)design modeling languages to
reinforce representation adequacy exploiting this relation.” (p. 35) This modeling approach is based on
the classic ‘Bolzano-Ogden-Richards’ semantic triangle between a thing in reality, its conceptualization,
and a symbolic representation of the conceptualization. Hence, the shared taxonomies of categorical
types, roles, attributes, values, relationships, and relational properties usually accord with a metaphysics
which underlies the various ontological accounts, since they are all based on the semantic foundations of
knowledge representation. [cf. 27] Consequently, the ontology modeling language standards which are
applied in various fields of computational, scientific, and social domains, share the following principles:
(1) the search for a uniform domain independent or explicitly domain dependent formalism which can
capture ontologies (top down and bottom up, resp.); (2) the requirement of a logic and machinery which
can support inference, while maintaining some measure of (3) completeness (since an incomplete
language is bound to produce incomplete specifications) and (4) consistency (because modifications allow

only conservative extensions) (5) domain dependent expressibility. These goals and requirements
inevitably lead to an ontological variant of the "closed world assumption": in which that which cannot be
expressed in the terms of your ontology is deemed not to exist.
Approaches which address the issue of conceptualization in a space that includes concepts,
metadata, domain modeling, the context of conceptualization, and the cognitive subject raise the issue
of whether or not concepts and features of reality exist without the personal knowledge and agency of
a knower. [19, 52] This is the case even without considering the historical embeddedness of concept
development. [21] These approaches are frequently suppressed by the practical requirements and
success of applied ontologies which span a wide spectrum of relatively well defined uses and
domains. [42] In a metaphysically neutral methodological categorization, taxonomies, meta-data
structures, the composition of topics, and the understanding of interpretative and learning contexts, do
not exist without an interpreter of the meaning of expressions performing intentional epistemic actions.
Since epistemic actions not only include public announcements but also doxastic interactions and
categorical revisions, mutual understanding requires the conciliation of Human-Agent Information
Architectures for personal and social conceptualization with collaborative knowledge organization. [17]
1.2 Metaphysically Neutral Information Architectures
Knowledge Organization (KO) is closely related to, and in several key aspects overlaps with, ontology
research in Information Science (IS). In IS ontologies are required for the purpose of information
retrieval, and for semantic transparency in the organization or arrangement of stored items. This form of
KO concentrates on the gathering, manipulation, classification, storage, and retrieval of recorded
semantic information. Given the close relationship between IS and library science, the ontologies which
have been developed for various application areas (including taxonomic systems and metadata
standards) tend to be more or less static (though extensible). [23,15] Ontological commitments, that is,
agreements to consistently use a controlled vocabulary such as the Unified Medical Language System,
became as conventional as the Dewey Decimal Classification, in widely accepted forms of KO and
interchange, from the general supply of health care to more domain specific areas such as security
threats and virus classification. Computational frameworks like the ARPA Knowledge Sharing Effort [35]
facilitate the reuse of ontologically grounded knowledge, and supply us with mechanisms for translating
between knowledge bases. Comprehensive and versatile intelligent databases such as the massive
CYC project accumulate commonsense knowledge for natural language understanding or in applications
like the Terrorism Knowledge Base. Standard computer readable representations like the Knowledge
Interchange Format (KIF) give us declarative knowledge description languages which have proved to be
effective tools for ontology reconciliation and integration. Since the most common solution for handling
the alignment of static ontologies is ontology matching via “upper ontologies”, these formats have been
actively used in projects developing Standard Upper Ontologies, such as SUMO or OpenCyc.
With respect to IS Fonseca [24] differentiates ontologies of information systems from ontologies for
information systems. The first describes information systems and supports the creation of better
modeling tools. The second are computational forms which support the validation of ontology-driven
information systems, and ensure that our conceptual modeling schemas are correct. He contrasts
Gruber’s (generally accepted, 1992) definition that an ontology is a “specification of a
conceptualization” [22] with Guarino’s conception [23] which takes an ontology as a “logical theory”
which accounts for the “intended meaning of a formal vocabulary” (including Gruberian specifications).
If we interpret the latter conception as suitable for understanding “ontologies of information systems”
as ontologies applied for building “ontologies for information systems”, we can consider ontology
languages such as DAML+OIL, OWL, RDF as tools which are used to construct ontologies (in the
sense of Gruber’s definition) for information systems. If however we accept that ontologies behave as
theories, UML or OML should be considered as proper level modeling languages, equipped with a
grammar capable of describing the conceptual schemas of information systems if and only if we have
a conceptual modeling method and an associated intended model that together generate a theory
which corresponds to Guarino’s conception. Like Smith [5] Fonseca criticizes ‘instrumental ontologies’
and confronts Gruber’s formulation of the ‘closed world assumption’ of artificial intelligence (in which
“what ‘exists’ is that which can be represented” [22, p. 907]) with ‘Gruberian’ instrumentalist OE.
According to Guarino’s interpretation it is not ontologies we are constructing, but theories of
conceptualization which are being developed to construct info-bases on ontological grounds.
1.2.1 The problem of Non Categoricity
Guarino and his colleagues argue [27] that the intended meaning of a particular ontology (which
should be described in modal, intensional logical frameworks) is always underdetermined by our

theory, which always has unintended models (which can be considered as an ontological
generalization of the Duhem-Quine thesis for scientific theories. [46]). Consequently, a theory of
conceptualization necessarily provides only a partial “approximate characterization” of its intended
meaning. Thus the requirement that our interpretation must be isomorphically determined (i.e.
categoricity in the usual model-theoretic sense) turns out to be an unrealistic requirement for ontology
building. This does not seem to be a serious problem if we are working in the framework of a single
ontology with a relatively homogeneous scope. If however this theory encapsulating an intended
model encounters different conceptualizations, non-categoricity becomes a serious problem: merging
or integration oriented comparisons unavoidably focus on the differences, and it may turn out that
even the intended models are different as it is illustrated by the known problem of false agreement
(Figure 2.) adapted from [4, cf. 16]. The consolidation of such differences in case of overlapping
vocabularies and intents clearly points to the need for better techniques of reflection at the meta-
levels. The situation however is not less acute/severe if the intended models themselves overlap (e.g.
in a context of discovery), though the purposes of application, the “intents”, are different.
Figure 1. Relations between conceptualization, Model,
Modeling Language and Model Specification [3, p. 50]
Figure 2-3. The relationship of disjunctive intended and
overlapping all models in a domain D (adapted from [4]).
1.3 Change Management based on Upper Ontologies
After the initial period of large scale intelligent data-base development, upper ontology research soon
faced the problem of the situated and intent dependent nature of ontology construction. [29] The goal
that any “local” ontology be defined in terms of a single shared “upper ontology” so that every term in one
ontology and every term in another ontology can be boiled down to categories in an upper ontology
turned out to be tenable only for relatively homogeneous domains. Merging ontologies in heterogeneous
contexts [38] was relatively successful only within similar reality perspectives and areas of application,
e.g. in health science, enterprise modeling or genetics. [+59] In order to accommodate change as new
results are discovered, and old hypotheses refuted in science oriented ontologies for example,
frameworks like the OBO Foundry [30] were designed with the additional goal “on the meta-level […] to
establish ontology development itself as being, like statistics, a recognized part of the scientific
enterprise” but even in case of the development of successful existing ontologies, like the Ontology for
Ontology for Biomedical Investigations, maintaining principles like orthogonality for the sake of
interoperability had its pros and cons. [31, pp.23-26] Tracking changes in Change History Logs,
providing modules for change recovery, and on the fly visualization of change effects, e.g. in Protégé,
are relatively new results, [37] though early experiences already showed that versioning is not sufficient
for the maintanance of multi-agent ontology revision. Extensions and revisions run into obstacles in
heterogenous domains especially in case of distinct scope. [38, 65] The parallel dynamics of re-
constructing the upper and the local domain ontologies led to several logical and computability problems,
since changes in the top level ontology affect all local ontologies, even if their context did not change.
However, intent and context dependence proved to be the main reasons, in addition to the needs of
knowledge discovery and collaborative content creation, why ontologies have to be evolvable and
changeable via revision and refactoring. [1, 39, 43] These requirements imply the integration of KO
capabilities not just in “design time” but at run time and, in the collaborative case, “on the fly” throughout
the whole ‘life cycle’ of the ontology. [63] This way, dynamic ontology building and maturing [42] arrived
at technologies for permanent refactoring with open ended lifeline including the possibility of
bootstrapping the ontology management framework. [65] Upper ontologies and rules of extension help
us to keep the ontology in context, but they ultimately contribute to closed world assumptions. It is not
only that the knowledge we wish to capture expands relentlessly, but also our meta knowledge about
what we have learned about how best to organize and use that knowledge. When building a system what
seemed a good idea at the “design” time is bound to become a limiting factor, unless we can rebuild the
ship while still at sea. The primary motivation of upper ontologies can be seen in the desire to find uniformly
applicable means for managing ontologies. Uniformity is required for satisfying users expectations to learn

only once how to “drive” the system, but this aim should not lead to the homogenization of the content, as
it usually the case with upper ontologies. Since the turn of century, when refactoring had come to the
fore, the aim has been to develop computational means for managing change keeping the system fully
working. It also applies to upper ontology construction that the growth of alternative conceptualizations
requires the application of uniform refactoring methodologies and technologies at all levels.
2 PROBLEMS OF CONCEPTUALIZATION IN THE CONTEXT OF
KNOWLEDGE ORGANIZATION
Knowledge Organization (KO) is more closely related to the content of concepts and to the actual
knowledge items than taxonomic structures and ‘ontologies’ understood as “explicit specifications of a
conceptualization” (Gruber [22]). The further component of Gruber’s definition (which has settled debates
about earlier discussions on what ontologies are), namely, that they are “shared” specifications, bounds
conceptualization to the goal of communication. “Communication between people with different needs
and viewpoints arising from their differing contexts” [59] positioned ontologies as ‘Inter-Lingua’ providing
descriptive and normative models of shared understanding originally for multi-agent systems, e.g.,
Ontolingua, KQML, and FIPA-ACL] This goal which in the heydays of artificial intelligence research
tended to refer to the internal and external communication of multi-agent systems [18] has changed into
an explicit recognition of the importance and priority of human communication, especially in result of the
semantic needs of the web. [42] Although Gruber’s definition was criticized in certain aspects [23, 24] the
relatively early point of view became widely accepted that “the adequacy of a conceptual modeling
notation rests on its contribution to the construction of models of reality that promote a common
understanding of that reality among their human users.” [14, p. 52. Italics added.]. This approach which
applies both to automated ontology construction and human centered collaborative and communicative
conceptualizations suppressed the personal aspects of knowledge building and the context of discovery
gained attention only recently [43, 45] when the exploration of the web for the personal purposes of KO
became standard practice nearly in all fields. [52] The requirements of pragmatic efficiency, conceptual
clarity and ability to support means of communication were consequently complemented with
comprehensibility and domain appropriateness [15] and further formal requirements (e.g., soundness,
lucidity) discussed by Guizzardi [60]. It is increasingly suggested that these requirements need to be a
subject of re-evaluation in the context of discovery [45, 49, 50, 56]. Gricean conversational maxims may
turn out to be untenable for dynamic knowledge building in light of Brewter’s and Wilks’ conception of
ontologies for example, which are expected to function as a “Medium of Human Expression”. [45]
2.1 Dynamic Knowledge Architectures
2.1.1 Referential meta-models versus meta-circular development
In the context of discovery the main problem is not representational adequacy but representational
flexibility emphasizing the process of elaboration instead of insisting on “frozen” results of precision.
[1] The exploration of new knowledge and its organization forced recent efforts by Semantic Web
practitioners to address the problem of conceptualization from the point of view of linked information
as emergent knowledge structures that do not assume a pre-existent reality to which they have to
correspond. [42, 51] The semantic information, the content at given nodes of a Knowledge Graph for
example, may refer to such reality and the emergent architecture may correspond to abstractions of
certain state of affairs, but the conceptualization and the representational structure is not measured in
terms of adequacy or correspondence to a domain of the real world, rather it develops together with
the user’s understanding of the information, her personal knowledge organization process that is
dependent upon its purpose. For this reason recent efforts in visual KO point to a more
flexibleapproach to conceptualization. [1, 54, 55, 56] These works are seeking situated, goal directed,
cooperative, approaches to knowledge exploration and a dialogical, interactive emergent semantics
that assumes, from the point of view of web based KO, a “self-supporting” [55] bootstrappable
ontology authoring environment. [48, 51?] Emergent “bottom up”, evolutionary semantics, “refers to a
set of principles and techniques analyzing the evolution of decentralized semantic structures in large
scale distributed information systems” in which representation of semantics and the discovery of the
proper interpretation go hand in hand. [53] In the case of augmented environments of
conceptualization such as Codex [56], WikiNizer (regarding its collaborative reference model:
Conceptipedia [54, 64]), or OntoWiki [48], it is also a goal to built upon universal (in the sense of
universal function) and “meta-designable” means of combinations with closure operations to create
objects that are within the original domain of investigation, allowing recursive compositions, and
“meta-reflective” “means of abstractions” for the generated contexts of application [36, 58]. If we add

the goal of self sufficiency, in the sense of providing meta-circular self-supporting executable
definitions of the systems entire repertoire of capabilities, to the requirement of flexibility, the
knowledge and experience gained in organizing knowledge can be articulated within the system.
Thus, self sufficiency yields ever more powerful tools of knowledge organization that are
bootstrappable in an Engelbartian sense. [57] This allows not only for rebuilding the ship, while at sea,
but could be building a better ship. Systems capable of this kind of meta-reflective meta-design can be
called in the spirit of his NLS, an Augmentation Engine [58]. The implications of these solutions can be
drawn and contrasted with the principles formulated above (1-5 resp.) in section 1.1.2 in the following
terms: (1) Elaborate particular domains of interest in a dynamic form that is open, and extendable. (2)
Each domain and meta-domain can have their own domain specific inference scheme, and can co-
evolve with the growth of knowledge. (3) Scalability to increasing levels of completeness, (4) optionally
lifting the requirements of consistency for the sake of admitting non conservative operations. (5)
Increasing expressibility via bootstrappig without changing the computational metaphor at the meta-levels.
2.1.2 Static versus Dynamic Presentation of Semantic Information
Although there are process ontologies, and ontologies of temporal (social and historical) events, i.e.
descriptions of dynamic aspects of the world, these are not designed for the description of the
personal or collective development of conceptualizations. [30, 37] The creation of the “great chain of
bootsrappable meaning” requires autopoetic expressive power to elaborate dynamic knowledge
architectures. Instead of seeking to formulate uniform metaphysical upper ontologies these endeavors
seek to develop a uniform meta formalism, “the simplest” self-applicable and extensible thing “that
could possibly work”. Live systems of this kind are hard to find in the literature, the closest are [55, 56,
58,]. Cunningham developed the Wiki as the "simplest database that could possibly work". OntoWiki
aimed at creating a semantic wiki as the "simplest knowledgebase that could possibly work". WikiNizer
is aiming to take this line of thinking to its conceptual conclusion by bootstrapping the "simplest
possible personal augmentation engine that could possibly work". WikiNizer aims to take us "Beyond
Ontologies" in the spirit of Codex. [56] In addition, by taking advantage of emerging technologies like
“mobile first” HTML5 as a platform and web scale NoSQL databases, it is now possible to develop
computer support for personal as well as collaborative knowledge work (as described in the reference
model of Conceptipedia [54, 64]) that could possibly bring about the “cultural shift” [56, p. 672] needed
to bring the benefits of open collaborative computer support of knowledge work to the way knowledge
is produced and disseminated. The development of a dynamic, visual collaborative technology has to
satisfy the evolving needs of bootstrapping in the process of knowledge building. The meta framework
supported by augmented technologies can be used to capture, identify and define all possible objects
and their relations of interest at the necessary emergent meta-levels. Taken together these
methodologies constitute the organization and conceptualization of a knowledge domain together with
all the content organizing meta-concepts that will emerge. Discursive elaborations of domain
knowledge thus get fully integrated into a workflow that allows the ad-hoc but purpose/intent-ful
introduction of meta-level concepts without breaking the original flow of expositions. As in a Wiki,
when we get to the edge of our knowledge, be it at the domain or any meta level, we simply create a
new “page” (as a node in a meta Knowledge Graph) to be worked on as the needs and the available
knowledge dictates. Those existing approaches to ontology change that are not static and “thing”-
based, are aware of the importance of traceability [37, 65] and that “history matters” in incremental
ontology reasoning [61] but they are not reflective. They do not satisfy the dynamic requirements of
reorganization, intellectual manageability, and cognitive flexibility. Restructuring knowledge and contexts of
discovery require more dynamic forms of knowledge representation. Augmentation engines [58] and visual KO
tools can help us build conceptual meta-structures through design processes. Interactive visualizations of
problem spaces in the form of conceptual graphs serve problem framers by playing an active role in
defining the problems to be solved, and “Going Meta” in Simonyi’s sense, can make cross-interdependence
between conceptual domains comprehensible by changing the level of abstraction and exploring the class of
problems of which the current one is an instance, in a bottom up way.
2.1.3 Bottom Up Live Micro Ontologies
The structures that emerge in the course of knowledge building that includes discovery and
conceptualization typically have all the characteristics we have described in the previous section. To
distinguish the emerging Knowledge Architectures from alternative approaches we propose to refer to
them as “Bottom Up Live Micro Ontologies”. Bottom up’, in compliance with the literature, [62], because
they are created in the course of elaborating a concrete domain and ‘micro’, because the meta terms
introduced can affect a single node or any that are linked to a node in a piecemeal agile way and in close
contact to the context from which they emerge. These micro ontologies are usually smaller in size than
the so called “local ontologies” in domain modelling [16] because they are amenable to reuse from any

context, way beyond the one that gave rise to them. Using Micro Ontologies it becomes possible to
define and manipulate domain knowledge with the aid of meta-level structures introduced on the fly, and
these meta-structures can also be treated later as domains of their own right. Elaborating meta level
structures as domains of their own right, leads to additional meta-meta level structures, and the same
process can be repeated as far as needed. So knowledge architecture constructions are “turtles all the
way up”. In a bottom up approach domain specific, as well as meta-level concepts and methods can be
developed in a form of “instance first”. “In instance-first development, one implements functionality for a
single instance, and then refactors the instance into a class that supports multiple instances” [13], which is
to say we are “going meta”. Only through live exploration and elaboration of descriptions of exemplars,
specific instances of objects of interests, it is possible to develop suitable situated elaborations and
conceptualizations that can capture ontologically what “there is” across many instances. This can be stated
as the methodological requirement of the “primacy of bottom up live development”: the characteristics of
instance descriptions and the relationship with other instances should not be lost as we construct
conceptualizations that are applicable to the class of things that are being described. Hence, instead of
“conceptual atomism” [2] and correspondences between descriptions and some aspects of reality, KO
seeks to establish correspondence between the structure including the relationships between instances
and their class models in a more abstract sense of ‘images’, or using a current term ‘visual models of
reality’, in the spirit of Hertz’s Principles of Mechanics
1
. In the process of KO the formation of these ‘images’
is however, much closer both historically and methodologically, to Whewell’s “consilience of inductions”
trough the “colligation of facts”. [41, p.74]. To paraphrase Ward Cunningham’s quoted dictum: the
emerging live, visual knowledge architectures should be “the simplest thing that could possibly work” that
enable us to achieve our knowledge goals and intentions in a given situation. With respect to ontology
evolution timelines, it is not only the results of conceptualization that matter but the creation of “knowledge
model[s] that preserves audit trails of resource manipulation” as the records of “concept growth can
increase the transparency of a research enterprise”. [56, p. 672] The vision that takes us “beyond
ontologies” had largely been explored with the Augmentation System that Engelbart created in his NLS
half a century ago on a ‘milli iphone’. With the millionfold increase of computing resource available
even to individuals today, we can embark on developing the means to promote the “culture shift” iIbid.]
that could lead to collaborative creation of the ‘great chain of emergent meanings’. In this quest we
need dynamic mechanisms for recognizing and merging alternative conceptualizations.
3 THE DYNAMICS OF LIVE CONCEPTUALIZATION AND ITS VISUALIZATION
The simplest visual knowledge architectures capable to manage the content correlated dynamics of
concept development are the graph structures of visual semantic wikis like ThinkBase. [47] They need
proper mechanisms to cope with the various aspects of ontology change management [65]: mapping,
matching, alignment, revision, versioning, merging, evolution and integration. Flouris et al. [33] give a
classification of terms, methods and operations that can correlate concept development with the
enrichment of knowledge domains. AGM theory and its recent extensions in the general framework of
Dynamic Epistemic Logic provide formal belief revision operators for model fitting applicable to iterated
stepwise update and expansion of a knowledge base. Formerly, truth maintenance systems applied non
conservative entrenchment to implement nonmonotonic reasoning for the purposes of knowledge
discovery. Until recently, however, ontology engineering worked either by expert-based or
automated/semi-automated methods of centralized Knowledge-based Systems for solving problems of
conceptualization. Human-Agent Negotiation techniques and protocols required by the Semantic Web
came to the fore partially as a result of the poor record of automatic ontology generation and partly as
a reaction to the rise of new issues of collective conceptualization in emergent semantics. [26, 42, 51,
52, 53] The requirements of live visualization and collaboration for a wiki-like concept organizer of a
global giant graph knowledge base [1], emerging concepts and their intellectual manageability become
central issues. As it is described in the reference model of “Conceptipedia” [54, 64], conceptualization
can be improved by social interaction, cooperative learning and through collaborative sense-making.
In collaborative knowledge work and problem solving the discovery of informative connection
subgraphs can be based not only on data mining techniques and ranking-based discovery of semantic
associations or similarities between linked data, (e.g. in SWETO Schema – Visualization) but also on
patterns of user interaction with her resources and peers. Current technology makes it possible to
trace and generate trails from a user’s movements trough links between entities in a knowledge graph.
1
“We form for ourselves i mages (‘Bilder’ in Hertz's original German) or symbols of external objects; and the form which we give them is such that the necessary
consequents of the images in thought are always the images of the necessary consequents of the things pictured.” [40, Introduction, p.1.]

The analysis of the traversal frequency of the graph provides meaningful relationships and patterns of
user behavior which can be utilized for modeling the dynamics of the personal use of the concept
graph. Such an analysis could provide user generated emergent semantics for categorical and aspect
dependent conceptualizations at the organizational meta level of our knowledge resources. These
conceptualizations usually take place during knowledge discovery and organization determined by the
intent of the user. Since our knowledge grows not in isolation but in the form of collective in situ efforts
to share our explored, re-usable paths of conceptualization, co-evolving direct manipulable
visualization of concepts and learning objects are needed both at the object and meta-level to handle
the social aspects of learning and knowledge work in the framework of an Exploratory Epistemology.
[54, p. 117] Current systems tend to separate the conceptual level from the content, and the context of
logical understanding from the intent laden use of concepts. Besides, they usually consider the
unification of individual contributions in a ubiquitous global representation. As opposed to this God's
eye view current research explores new alternatives that aim to provide live , tinkerable, in situ
Morphic visualization of concepts and direct manipulation based authoring capabilities. [55, 58]
Liveness at the visualization level means that screen objects are interactive and active, they react to
user input and can have custom behavior. [55, p. 3] Direct manipulation means that the manipulative
affordances of all objects are made clear to the user. ‘Worlds’ in Lively Kernel are the units of
persistence. In contrast to Lively Kernel, in WikiNizer the basic units of persistent elements that are
modified through user interactions, are not Worlds, but individual nodes and edges in the underlying
graph. In WikiNizer Worlds (arbitrary collections of nodes, edges or graph fragments) are more like
conceptual units like named tags in a Concurrent Versioning System. The implication of this solution is
that collaboration can take place at a much finer grained level. For this reason, the authoring
capabilities are uniformly applicable and user-extendable through meta-design both at the content, as
well as all emergent conceptual meta-levels, within a collaborative frameworks like Conceptipedia.
Morphic integrates verbal and pictorial models of conceptualization in its graphics system of intentful
workflows as flexible, reusable building blocks for collaborative work. Distributing the workload and
responsibility through Crowd Authoring, concept alignment and contributions from an online
community filters out errors and misconceptions. Negotiation games played at the meta-level of
knowledge items supported by a reputation system mechanism can support emergent
conceptualization until workable ideas win out and taxonomic structures are settled. This way,
challenges of social and collaborative concept matching can be addressed in a augmented knowledge
management kernel that is also capable to supplement education technology.
REFERENCES
[1] Benedek, A. & Lajos, G. (2012): Building Augmented Knowledge Architectures: Requirements for
Collaboration Platforms of NextGen Concept Organization Tools. In A. López Martínez, L. Gómez
Chova, I. Candel Torres (Eds.) Proceedings of ICERI2012 Conference 19th-21st November 2012,
Madrid, Spain, IATED, pp.1492-1506.
[2] Fodor, J., 1998. Concepts: Where Cognitive Science went Wrong. Oxford University Press, New York.
[3] Guizzardi, G. (2005). Ontological foundations for structural conceptual models. CTIT, Centre for
Telematics and Information Technology.
[4] Guarino, N. (2004). Toward a formal evaluation of ontology quality. IEEE intelligent Systems, 19(4), 78-79.
[5] Smith, B. (2003). Ontology. In L. Floridi (Ed.) The Blackwell Guide to the Philosophy of Computing
and Information, Oxford: Blackwell, 2003, 155–166.
[6] Margolis, E. and Laurence, S. (2007). The Ontology of Concepts—Abstract Objects or Mental
Representations? Nous, 41(4) pp. 561–593.
[7] Bunge M. (1977-79). Ontology I-II: The Furniture of the World, Treatise on Basic Philosophy, vol. 3-4,
New York, and Dordrecht, Holland: D. Reidel Publishing.
[8] Guizzardi, G. (2007).On Ontology, ontologies, Conceptualizations, Modeling Languages, and
(Meta)Models In: Proceedings of the 2007 conference on Databases and Information Systems IV:
Selected Papers from the Seventh International Baltic Conference DB&IS'2006, pp. 18-39.
[9] Sowa, J.F. (1984). Conceptual Structures. Information Processing in Mind and Machine. Addison-
Wesley, Reading, MA.
[10] Carnap, R. (1967). The Logical Structure of the World: Pseudoproblems in Philosophy. Univ. of California P.
[11] Ingarden, R. (1964/65). Der Streit um die Existenz der Welt. Bd. I/II, Tübingen:Niemeyer.
[12] Chisholm, R. M. (1996). A Realistic Theory of Categories: An Essays on Ontology. Cambridge: CUP.
[13] Instance First Development http://c2.com/cgi/wiki?InstanceFirstDevelopment;
http://osteele.com/archives/2004/03/classes-and-prototypes. (Accessed: 01.13, 2014.)

[14] Mylopoulos, J. (1992). Conceptual modeling and Telos., In P. Loucopoulos and R. Zicari, (eds.)
Conceptual modeling, databases, and CASE, Wiley, pp. 49—68.
[15] Guizzardi, G., Pires, L. F., & Van Sinderen, M. (2005). An ontology-based approach for evaluating
the domain appropriateness and comprehensibility appropriateness of modeling languages (pp.
691-705). Berlin Heidelberg: Springer.
[16] S. Borgo, S., Gangemi, A., Guarino, N., Masolo, C. and Oltramari, A. (2002). Ontology RoadMap.
WonderWeb Deliverable D15. http://wonderweb.man.ac.uk/deliverables/documents/D15.pdf
(Accessed: 01. 13. 2014.)
[17] Alvarez, H. A. (2012). Collaborative knowledge creation. Draft Working Paper. http://rcrc.brandeis.edu/
pdfs/Collaborative%20Knowledge%20Creation%20and%20Relational%20Coordination.pdf (Accessed: 01. 13. 2014.)
[18] Steels, L. (1998). The Origins of Ontologies and Communication Conventions in Multi-Agent
Systems. Autonomous Agents and Multi-Agent Systems, 1, 169-194.
[19] Polanyi, M. (1962). Personal knowledge: Towards a post-critical philosophy. Psychology Press.
[20] Bittner, T., & Smith, B. (2003, March). Granular spatio-temporal ontologies. In Proceedings of the
AAAI spring symposium on foundations and applications of spatio-temporal reasoning.
[21] Lakatos, I. (1971). History of science and its rational reconstructions. Springer Netherlands. pp. 91-136.
[22] Gruber, T. R. (1992-95). Toward Principles for the Design of Ontologies Used for Knowledge Sharing. Int.
J. of Human Computer Studies, 43(5/6), 907-928. cf.: http://www-ksl.stanford.edu/kst/what-is-an-ontology.html
[23] Guarino, N. (1998). Formal Ontology and Information Systems. In N. Guarino (Ed.), Formal
Ontology in Information Systems (pp. 3-15). Amsterdam, Netherlands: IOS Press.
[24] Fonseca, F. (2007). The Double Role of Ontologies in Information Science Research, Journal of
the American Society for Information Science and Technology, 58(6), pp. 786-793.
[25] Gruber, T. R. (1993). A translation approach to portable ontologies. Knowledge Acquisition, 5(2), 199–220.
[26] Aberer, K., Cudré-Mauroux, P., Ouksel, A. M., Catarci, T., Hacid, M. S., Illarramendi, A., ... &
Studer, R. (2004). Emergent semantics principles and issues. In Database Systems for Advanced
Applications, Berlin Heidelberg: Springer, pp. 25-38.
[27] Guarino, N. and Giaretta, P. (1995). Ontologies and knowledge bases: Towards a terminological
clarification. In N. Mars (Ed.): Towards Very Large Knowledge Bases: Knowledge Building and
Knowledge Sharing. Amsterdam: IOS Press, pp. 25–32.
[28] Chen. P. (1976).The entity-relationship model: Towards a unified view of data., ACM Transactions
on Database Systems 1(1).
[29] Chandrasekaran, B., Josephson, J. R., & Benjamins, V. R. (1999). What are ontologies, and why
do we need them? Intelligent Systems and Their Applications, IEEE, 14(1), 20-26.
[30] Smith, B. et al., (2007). The OBO Foundry: Coordinated Evolution of Ontologies to Support
Biomedical Data Integration. Nature Biotechnology 25, 1251–1255.
[31] Smith, B. (2008, July). Ontology (Science). In C. Eschenbach and M. Gruninger (eds.)
Proceedings of FOIS 2008, Amsterdam/New York: ISO Press, 21–35.
[32] Falbo, R. D. A., Guizzardi, G., & Duarte, K. C. (2002). An ontological approach to domain engineering. In
Proc. of the 14th Intern. Conf. on Software Engineering and Knowledge Engineering, 351-358. ACM.
[33] Flouris, G., Manakanatas, D., Kondylakis, H., Plexousakis, D., & Antoniou, G. (2008). Ontology change:
classification and survey. The Knowledge Engineering Review 23(02):117-152.
[34] Gómez Chova, L. , et al. (Eds.) INTED2013 Proceedings, Valencia, Spain: IATED.
[35] Neches, R., Fikes, R. E., Finin, T., Gruber, T. R., Patil, R., Senator, T., & Swartout, W. R. (1991).
Enabling technology for knowledge sharing. AI Magazine, 12(3), 16-36.
[36] Abelson, H., Sussman, G.J.,with Sussman, J. (1966). Structure and Interpretation of Computer
Programs. The MIT Press.
[37] Khattak, A. M., Latif, K., Han, M., Lee, S., Lee, Y. K., & Kim, H. I. (2009). Change tracer: Tracking
changes in web ontologies. In 21st Intern. Conf. on Tools with Artificial Intelligence, IEEE, pp. 449-456.
[38] Wache, H., Voegele, T., Visser, U., Stuckenschmidt, H., Schuster, G., Neumann, H., & Hübner, S.
(2001, August). Ontology-based integration of information-a survey of existing approaches. In
IJCAI-01 workshop: Ontologies and Information Sharing, Vol. 2001, pp. 108-117.
[39] Moen, A., Mørch, A.I. & Paavola, S. (Eds.) (2012). Collaborative Knowledge Creation. Rotterdam:
Sense Publishers.
[40] Hertz, H. (2004). The principles of mechanics: presented in a new form. Dover publications.
[41] Forster, M. R. (1988). Unification, explanation, and the composition of causes in Newtonian
mechanics. Studies In History and Philosophy of Science Part A, 19(1), 55-101.

[42] Braun, S., Schmidt, A. P., Walter, A., Nagypal, G., & Zacharias, V. (2007). Ontology Maturing: a
Collaborative Web 2.0 Approach to Ontology Engineering. In: WWW-07 CKC2007 Workshop.
[43] Bundy, A. (2008). Why ontology evolution is essential in modeling scientific discovery. In
Proceedings of the 2008 AAAI Fall Symposium on Automated Scientific Discovery.
[44] Masolo, C., Guizzardi, G., Vieu, L., Bottazzi, E., & Ferrario, R. (2005, November). Relational roles and
qua-individuals. In AAAI Fall Symposium on Roles, an interdisciplinary perspective, pp. 103-112.
[45] Brewster, C. A., Iria, J., Ciravegna, F., & Wilks, Y. (2005 February). The Ontology: Chimaera or Pegasus.
In (ed. by the same authors) Proc. of the Dagstuhl Seminar Machine Learning for the Semantic Web.
[46] Laudan, L. (1990). Demystifying underdetermination. Minnesota Studies in the Philosophy of
Science, 14, 267-297.
[47] Dori, D. (2004). ViSWeb—the Visual Semantic Web: unifying human and machine knowledge
representatations with Object-Process Methodology. The VLDB Journal — The International Journal
on Very Large Data Bases, 13(2), 120 – 147. New York: Springer-Verlag.
[48] Frischmuth, P., Martin, M., Tramp, S., Riechert, T., & Auer, S. (2012). OntoWiki–An Authoring,
Publication and Visualization Interface for the Data Web. Semantic Web 1 (1–5). IOS Press.
[49] Mohagheghi, P., Dehlen, V., Neple, T. (2009). Definitions and approaches to model quality in model-
based software development–A review of literature. Information and Software Technology, 51(12), 1646-69.
[50] Stock, K., Karasova, V., Robertson, A., Roger, G., Small, M., Bishr, M., ... & Gardner, Z. (2013).
Finding Science with Science: Evaluating a Domain and Scientific Ontology User Interface for the
Discovery of Scientific Resources. Transactions in GIS. 17(4): 612-639.
[51] Cudré-Mauroux, P. (2006). Emergent semantics: rethinking interoperability for large scale
decentralized information systems (Doctoral dissertation, Google Inc). 53-al össze
[52] Mika, P. (2007). Ontologies are us: A unified model of social networks and semantics. Web
Semantics: Science, Services and Agents on the World Wide Web, 5(1), 5-15.
[53] Cudré-Mauroux, P. (2010). Emergent Semantics. CRC Press.
[54] Benedek, A., Goodman, C., P., and Lajos, G. (2013). The ‘Conceptipedia’ of Visual Semantic
Wikinizers: a Reference Model for Collaborative Conceptualization. In: Garrigós Simón, et
al. (Eds.) INNODOCT 2013 Proc. of the International Conference: “New Changes in Technology
and Innovation” May 6-7, Valencia Spain: UPV, pp. 109-119.
[55] Krahn, R., Ingalls, D., Hirschfeld, R., Lincke, J., & Palacz, K. (2009). Lively Wiki a development
environment for creating and sharing active web content. In Dirk Riehle & Amy Bruckman (eds.),
Proc. of the 5th International Symposium on Wikis and Open Collaboration: ACM
[56] Pike, W. and Gahegan (2007). Beyond ontologies: Toward situated representations of scientific
knowledge, Int. J. Human-Computer Studies 65. 659–673.
[57] Engelbart (1962). Augmenting Human Intellect: A conceptual Framework. SRI Summary Report
AFOSR-3223. http://www.dougengelbart.org/pubs/augment-3906.html (Accessed 09.06, 2013), cf.
http://www.dougengelbart.org/about/bootstrapping-strategy.html (Accessed 01.02, 2014.)
[58] Lajos, G. and Benedek, A. (2013). “Building WikiNizer, a Personal Knowledge Augmentation
Engine for Next-Gen Semantic Collaboration Platforms.” In [34], pp. 1333-1343.
[59] Uschold, M., & Gruninger, M. (1996). Ontologies: Principles, methods and applications.
Knowledge Engineering Review, 11(2), 93-136.
[60] Guizzardi, G., Pires, L. F., & van Sinderen, M. (2006). Ontology-based evaluation and design of
domain-specific visual modeling languages. In Advances in Information Systems Development,
New York: Springer, pp. 217-228.
[61] Grau, B. C., Halaschek-Wiener, C., & Kazakov, Y. (2007). History matters: Incremental ontology
reasoning using modules. In The Semantic Web, Berlin Heidelberg: Springer, pp. 183-196.
[62] Maynard, D. et. al (2010) D1. 5.4 Bottom-up evolution of networked ontologies from metadata.
http://www.neon-project.org/deliverables/WP1/NeOn_2010_D154.pdf (Accessed 01.02, 2014.)
[63] Togia, T., & McNeill, F. (2012). On-the-fly repair of structural and lexical ontology mismatches in
multi-agent systems. Artificial Intelligence Review, 1-17.
[64] Benedek, A. & Lajos, G. (2013). From Personal to Collaborative Concept Organization:
Concepripedia as a Visual Tool for Educational Initiatives In [34], pp. 1303-1316.
[65] Khattak, A. M., Latif, K., & Lee, S. (2013). Change management in evolving web ontologies.
Knowledge-Based Systems, 37, 1-18.