=Paper=
{{Paper
|id=None
|storemode=property
|title=Ontology Repositories with Only One Large Shared Cooperatively-built and Evaluated Ontology
|pdfUrl=https://ceur-ws.org/Vol-596/paper-14.pdf
|volume=Vol-596
}}
==Ontology Repositories with Only One Large Shared Cooperatively-built and Evaluated Ontology==
https://ceur-ws.org/Vol-596/paper-14.pdf
Vol-596
urn:nbn:de:0074-596-3
C opyright © 2010 for the individual
papers by the papers' authors.
C opying permitted only for private and
academic purposes. This volume is
published and copyrighted by its
editors.
ORES-2010
Ontology Repositories and Editors for
the Semantic Web
Proceedings of the 1st Workshop on Ontology Repositories and
Editors for the Semantic Web
Hersonissos, Crete, Greece, May 31st, 2010.
Edited by
Mathieu d'Aquin, The Open University, UK
Alexander García Castro, Universität Bremen, Germany
Christoph Lange, Jacobs University Bremen, Germany
Kim Viljanen, Aalto University, Helsinki, Finland
10-Jun-2010: submitted by C hristoph Lange
11-Jun-2010: published on C EUR-WS.org
Ontology Repositories with Only One Large Shared
Cooperatively-built and Evaluated Ontology
Philippe A. Martin
ESIROI STIM, University of La Réunion, France
and adjunct researcher of the School of I.C.T. at Griffith Uni., Australia
Abstract. This article first lists reasons why an ontology repository - or, more
generally, a knowledge base (KB) server - should permit the collaborative
building of one well organized KB rather than solely be a repository for
heterogeneous KBs. To that end, the article proposes a KB editing protocol that
keeps the KB free of automatically/manually detected inconsistencies - and
leads knowledge providers to semantically organize their terms and statements -
while not forcing them to discuss or agree on terminology and beliefs nor
requiring a selection committee. Then, the article gives ideas on how to extend
this support to allow a precision-oriented collaborative evaluation of each
information provider and piece of information.
Keywords: knowledge sharing/integration/retrieval/evaluation
1 Introduction
Ontology repositories are often only imagined as being collections of static formal files
(e.g., RDF documents) more or less independently developed, hence loosely inter
connected and mutually partially redundant or inconsistent. Section 2 shows that this
"static file based approach" as opposed to a "collaborativelybuilt wellorganized large
knowledge base (cbwoKB) server approach" makes knowledge sharing/reuse tasks
complex to support and do correctly or efficiently, especially in a collaborative way.
Most Semantic Web related research works are intended to support such tasks
(ontology creation, retrieval, comparison and merging) and hence are useful. However,
most often, they lead people to create new formal files thus contributing to the
problems of knowledge reuse instead of inserting their knowledge into a cbwoKB
server. Indeed, it seems that WebKB2 [10] (webkb.org) is the only ontology server that
has protocols supporting governance-free loss-less well-organized knowledge sharing.
(There are no such protocols in CYC, Ontolingua, OntoWeb, Ontosaurus, Freebase,
semantic wikis ...). WebKB-2 also has a large general ontology and hence has at least
two of the elements necessary to build a cbwoKB (Other “shared ontology”
servers/editors i) let any authorized users make any change in the shared ontology
(this discourages information entering or leads to edit wars), or ii) rely on each user
or some privileged users to accept or reject changes made in the shared ontology (this
is bothersome for the evaluators, sometimes forces them to make arbitrary selections,
and is a bottleneck in information sharing that can cause long delays or discourage
information providers). Section 3 gives protocols - with many yet unpublished ideas -
to avoid these governance problems and thus support scalable collaborative building
of a cbwoKB, i.e., a KB where detected partial redundancies or inconsistencies are
prevented or made explicit via relations of specialization, identity and/or correction;
thus, in a cbwoKB, each object has one "right place" in the specialization hierarchy
and is then easily retrievable and comparable to the other objects. Section 4 gives
ideas on how this support can be extended to allow collaborative knowledge
evaluation.
2 Approaches Based on Files Versus cwoKB Servers
With files, information retrieval (IR) often leads to a list of possibly relevant files or
pieces of information (objects, e.g., a formal term or a informal sentence) whereas it
leads to an exact answer in a cbwoKB or within the content of one formal file. Such
an answer may be a portion of the cbwoKB, e.g., a part/subtask/specialization
hierarchy (with associated argumentation structures) if the query is of the kind "what
are the resources/tools/methods to do ...". Such semantically structured answers allow
a user to find and compare all relevant objects instead of getting a long redundant list
of objects/files where original/precise ones are hidden among/behind objects that are
more general, mainstream or from big organizations. This is also why IR quality
decreases when the size and number of the files increases, but not when the number
of objects increases in a cbwoKB.
The more objects two files contain, the more difficult it is to link these files via
semantic relations and hence to semantically compare, organize and evaluate them.
Instead, similarity/distance (statistical) measures have to be used. In a cbwoKB, when
needed, semantic queries can be used to filter objects or generate files, according to
arbitrary complex combinations of criteria, e.g., about the creators of the objects.
(Some of these criteria may be used for the internal organization of the cbwoKB but
the resulting "views" or "contexts" are language/representation dependent choices and,
unlike (semi-)independently created static files, lead the users to strongly relate
objects of different views). Ontology libraries, from the first ones such as the
Ontolingua library to imagined ones such as "The Lattice of Theories" [15], are often
organized into "minimal and internally consistent theories" to maximize their re-use.
However, this also leads to few relations between objects of different ontologies, as
well as implicit redundancies or inconsistencies between them, and hence more
difficulties to compare, merge or relate them. On the other hand, as acknowledged by
the author of [15], if the objects are organized into a cbwoKB, such (lattices of)
theories can be generated via queries.
With files, change management requires version management (which leads to more
files and many information management complications); not within a cbwoKB, as
Section 3 shows.
With formal files as inputs and outputs, knowledge re-use or integration leads to
the creation of even more files and requires people to select, compare, relate, merge,
adapt and combine (parts of) files. Except for simple applications where fully
automatic tools can deliver good-enough results, these are complex tasks that have to
be done by trained people who know the domain. Most works in collaborative
knowledge sharing or "ontology evolution in collaborative environments" are about
(semi-)automatic procedures for integrating two ontologies [5] and for rejecting or
integrating changes made in other ontologies, e.g., [2][12][13]. In a cbwoKB, no
adaptation or integration has to be done for each re-use: the most important/defining
relations from an object to other ones have to be entered by its creators and then they
can be incrementally complemented or corrected by any user. Indeed, it is often the
case that only the object authors know what their objects really mean or have some
other kinds of information required for relating their objects to other ones.
A cbwoKB maximizes the use of principled multi-inheritance hierarchies (for
specialization/mereological/spatial/... relations) where each object has one "right
place" in the sense that different users would search or insert this object at the same
place. Only a KB server with a large cbwoKB can permit a knowledge provider to
simply/directly add one new object "at its right place" and guide her to provide precise
and re-usable objects that complement the already stored objects. The protocols of the
next two sections work only with a cbwoKB.
3 Collaborative Editing of a KB
The next points describe the principles behind the editing protocols implemented in
WebKB-2 to make it a cbwoKB server, and make some comparisons with features of
RDF (which only supports a personal-file based approach). WebKB-2 allows the use
of several knowledge representation languages (KRLs): RDF/XML (an XML format
for knowledge using the RDF model), KIF and other ones which are here collectively
called KRLX and that were specially designed to ease knowledge sharing: they are
expressive, intuitive and normalizing (i.e., they guide users to represent things in ways
that are automatically comparable). One of them is named Formalized English (FE). It
will be used for the examples.
1. In WebKB-2, every object is a term or a statement (generally, a relation between
two quantified terms or some relations within the same context, i.e., meta-
statement). A term refers to a concept/relation type or an individual (an instance of
a first-order type). A statement is an individual and is either informal, formal or
semi-formal (when it uses a formal syntax and some terms/objects that are
informal or referring to informal/semi-formal objects). A (semi-)formal term is a
unique identifier for a (semi-)formal object. An informal term is a name for an
object. Different objects may have common names, not common identifiers.
Every (semi-)formal object has an associated source: creator or source file. The
(unique) meaning of a (semi-)formal object may be left implicit and hence might
be known only by its creator. Informal objects may also have an associated creator:
their meanings are those that their source has implicitly given them. These
distinctions permit the differentiation of (in-)formal objects and create one
specialization/generalization hierarchy categorizing all objects. More precisely,
this is an "extended specialization/generalization" hierarchy since in WebKB-2 the
classic "generalization" relation between formal objects (logical implication) has
been extended to apply to informal objects too.
In KRLX, informal objects are double quoted, and object identifiers are either
URIs or include their source identifiers as prefixes or suffixes. This is a common
solution to avoid lexical conflicts. KRLX allows the use of shortcuts for a source
may be used, e.g., wn#bird refers to one of the WordNet categories for the
English word "bird". The informal statement "birds fly"_[u1] was
created by the user u1. A difference with XML name-space prefixes in RDF/XML
is that the lexical declaration of a shortcut is also a semantic declaration of a term
for the source, thus encouraging the creator of the declaration to specify what the
source is (a person, a file, etc.); this is also possible in RDF but is not mandatory.
More importantly, RDF has no notion of "belief" whereas in WebKB-2 each object
is, in a sense, contextualized by its source. For example, if a statement S created by
a user U is not a definition, it is a belief of U. Similarly, a statement by a user U on
another user's statement S2 is actually U's belief on his interpretation of S.
In KRLX, a user also has an easy way to i) represent his belief that certain
statements belong to a certain source, or ii) associate a private key with its user
identifier to prevent another user to impersonate him, and iii) use an encrypted
form of this key (i.e., the related public key) for identifying himself.
A KRL that is meant to support knowledge sharing should offer normalized ways
to allow this so that knowledge sharing tools can support reasoning or
collaboration based on the knowledge sources. RDF and RDF/XML do not yet
offer a standard way to allow this. This will come: SPARQL and N3 already offer
a way to specify that a statement belongs to a source.
2. Any user can add any object and use it in any statement (as in RDF) but an object
may only be modified or removed by its creator. This last part has no equivalent in
RDF since it is a knowledge model, not a collaboration model.
3. Each statement has an associated source S, and hence, if it is not a definition of a
term created by S, is considered as a belief of S. When the creator of an object is
not explicitly specified, WebKB-2 exploits its "default creator" related rules and
variables to find this creator during the parsing. Similarly, unless already explicitly
specified by the creator, WebKB-2 uses the "parsing date" for the creation date of
a new object. Unless already specified, the creator of a belief is encouraged to add
restrictive contextualizing relations on it (at least temporal and spatial relations
must be specified).
A definition of a term T by the creator C of T may be said to be "neither true nor
false" or "always true by definition": a definition may be changed by its creator but
then the meaning of the defined term is changed rather than corrected. No one
(including C) is allowed to state something about T that is inconsistent with the
definition(s) of T. A user u1, is perfectly entitled to define u1#cat as a subtype
of wn#chair; there is no inconsistency as long as the ways u1#cat is further
defined or used respect the constraints associated with wn#chair. A definition
associated with T by a source S that is not C is actually a belief of S about the
meaning of T. At parsing time, WebKB-2 rejects such a belief if it is found
(logically) inconsistent with a definition of T by S.
Universally quantified statements are not definitions. Unlike KIF and N3, RDF
and OWL do not have a universal quantifier and hence force users not to make the
distinction. In WebKB-2, this distinction leads to very different conflict resolution
strategies (conflict between two statements of different sources).
• A conflict that involves two definitions by two sources S1 and S2 is a
misinterpretation by one of the sources, say S2, of the meaning of a term
S1#T created by the other source, and hence is solved by automatic term
cloning of S1#T, i.e., by creating S2#T with the same definitions except for
one and then replacing S1#T by S2#T in the statements of S2. The difficulty
is to automatically guess a relevant candidate for S1#T and a relevant
definition to remove for the overall change to be minimal. Annex 2 of [11]
provides some algorithms to do so in common cases.
• Otherwise, a loss-less correction is used (details in Point 6).
4. If adding, modifying or removing a statement introduces an implicit redundancy
(detected by the system) in the shared KB, or if this introduces an inconsistency
between statements believed by the user having done this action, this action is
rejected. Thus, in the case of an addition, the user must refine his statement before
trying to add it again or he must first modify at least one of his already entered
statements. An "implicit" redundancy is a redundancy between two statements
without a relation between them making the redundancy explicit, typically an
equivalence relation in the case of total redundancy and an extended specialization
relation (e.g., an "example" relation) in the case of partial redundancy.
In WebKB-2, a statement is seen as a graph with an interpretation in first-order
logic and graph matching is used for detecting if one graph (Y) is an extended
specialization of the other (X), i.e., if X structurally matches a part of Y and if
each of the terms in this part is identical or an extended specialization of its
counterpart term in X. For example, WebKB-2 can detect that the FE sentence
`Tweety can be agent of a flight with duration at least
2.5 hours'_[u2] (which means "u2 believes that Tweety can fly for at least
2.5 hours") is an extended specialization (and an "extended instantiation") of both
`every bird can be agent of a flight'_[u1] and `2 bird
can be agent of a flight'_[u1]. Furthermore, these last two
statements are respectively extended specializations of `75% of bird can
be agent of a flight'_[u2] and `at least 1 bird can be
agent of a flight'_[u2]. (Similarly, this last graph can be found to be
exclusive with `no bird can be agent of a flight'_[u3]).
Except for the fact that it takes into account numerical quantifiers and measures
instead of just the existential and universal quantifiers, the graph matching for
detecting an extended specialization is similar to the classic graph matching for a
specialization (or conversely, a generalization which is a logical deduction)
between positive conjunctive existential formulas (with or without an associated
positive context, i.e., a meta-statement that does not restrict its truth domain). This
last operation is sound and complete with respect to first-order logic and can be
computed with polynomial complexity if Y has no cycle [3]. Outside this restricted
case, graph matching for detecting an extended specialization is not always sound
and complete. However, this graph matching operation works with language of any
complexity (it is not restricted to OWL or FOL) and the results of searches for
extended specializations of a query graph are always "relevant".
The current reasoner used in WebKB-2 detects extended specializations as well as
the violation of relation signatures or exclusion relations. Since this reasoner
currently does not also use a rule based system or a theorem prover, it is not
complete with respect to first-order logic if rules are represented without using
specialization relations. However, this is irrelevant with respect to this article since
the presented protocols are not related to a particular inference method, they are
only triggered (and hence enforced) whenever an inconsistency or a redundancy is
detected or not when a new statement is entered.
However, it is important to note that i) the detection of implicit extended
specializations between two objects reveals an inconsistency or a total/partial
redundancy, and then ii) it is often not necessary to distinguish between these two
cases to reject the newly entered object. Extended instantiations are exceptions:
since adding an instantiation is giving an example for a more general statement, it
does not reveal a redundancy or inconsistency (here, an inconsistent belief or
incorrect interpretation of a term) that needs to be made explicit.
It is important to reject an action introducing a redundancy instead of silently
ignoring it because this often permits the author of the action to detect a mistake, a
bad interpretation or a lack of precision (on his part or not). At the very least, this
reminds the users that they should check what has already been represented on a
subject before adding something on this subject.
Adding, modifying or removing a term is done by adding, modifying or removing
at least one statement (generally, one relation) that uses this term. A new term can
only be added by specializing another term (e.g., via a definition), except for
process types which for convenience purposes can also be added via
subprocess/superprocess relations. A new statement is automatically added by
WebKB-2 into the extended specialization hierarchy via graph matching or, for
informal statements, solely based on the extended specialization between the
words they include). An automatic categorization may be "corrected in a loss-less
way" by any user. A new informal statement must also be connected via an
argumentation relation to an already stored statement. In summary, all objects are
manually or automatically inserted in the extended specialization hierarchy and/or
the subprocess hierarchy, and hence are easy to search and compare.
5. If adding, modifying or removing (a statement defining) a term T introduces an
inconsistency involving statements created or believed by other users (i.e., users
different from the one having performed this action), T is automatically cloned to
ensure that its interpretation by these other users is still represented. In the case of
term removal, term cloning simply means changing the creator's identifier in this
term to the identifier of one of the other users (if this generated term already
exists, some suffix can be added). In a cbwoKB server, since statements point to
the terms they use, changing an identifier does not require changing the
statements. In a global virtual cbwoKB, identifier changes in one server need to be
replicated to other servers using this identifier.
In a cbwoKB, it is not true that beliefs and formal terms (or their definitions, as
well as what they refer to, e.g., concepts) "have to be updated sooner or later".
Indeed, in a cbwoKB, every belief must be contextualized in space and time, as in
` `75% of bird can be agent of a flight' in place
France and in period 2005 to 2006'_[u3], even though such
contexts are not shown in the other examples of this article. If needed, u3 can
associate the term u3#75%ofbirdsflyinFrancefrom2005
to2006 with this last belief. Due to the possibility of contextualizing beliefs it
is rarely necessary to create formal terms such as u2#Sydney_in_2010. Most
common formal terms, e.g., u3#bird and wordnet1.7#bird never need to
be modified by their creators. They are specializations of more general formal
terms, e.g., wn#bird (the fuzzy concept of bird shared by all versions of the
WordNet ontologies). What certainly evolves in time is the popularity of a belief
or the popularity of the association between an informal term and a concept. If
needed, this changing popularity can be represented by different statements
contextualized in time and space.
6. If adding, modifying or removing a belief introduces an implicit inconsistency
involving beliefs created by other creators, it is rejected. However, a user may
"loss-less correct" a belief (that he does not believe in) by connecting it to a belief
(that he believes in) via a corrective relation. E.g., here are FE statements by u2
that correct a statement made earlier by u1:
` `every bird is agent of a flight'_[u1] has for
correctiverestriction `most healthy flyingbird are
able to be agent of a flight' '_[u2] and
` `every bird can be agent of a flight'_[u1] has for
correctivegeneralization `75% of bird can be agent
of a flight' '_[u2].
If instead of the belief `every bird can be agent of a flight', u1 entered the
definition `any bird can be agent of a flight', i.e., if he gave a definition to the type
named "bird", there are two cases:
• u1 originally created this type (u1#bird); then, u2's attempt to correct the
definition is rejected, or
• u1 added a definition to another source's type - say wn#bird since this type
from WordNet has no associated constraint preventing the adding of such a
definition - and hence i) the types u1#bird and u2#bird are
automatically created as clones (and subtypes of) wn#bird, ii) the
definition of u1 is automatically changed into `any u1#bird is agent
of a flight'_[u1], and iii) the belief of u2 is automatically changed
into `75% of u2#bird can be agent of a flight'_[u2].
In WebKB-2, users are encouraged to provide argumentation relations on
corrective relations, i.e., a meta-statement using argument/objection relations on
the statement using the corrective relation. However, to normalize the shared KB,
they are encouraged not to use an objection relation but a "corrective relation with
argument relations on them". Thus, not only are the objections stated but a
correction is given and may be agreed with by several persons, including the
author of the corrected statement (who may then remove it). Even more
importantly, unlike objection relations, most corrective relations are transitive
relations and hence their use permits better organization of argumentation
structures, thus avoiding redundancies and easing information retrieval.
The use of corrective relations makes explicit the disagreement of one user with
(his interpretation of) the belief of another user. Technically, this also removes the
inconsistency: an assertion A may be inconsistent with an assertion B but a belief
that "A is a correction of B" is technically consistent with a belief in B. Thus, the
shared KB may (and should) remain consistent.
For problem-solving purposes, i.e., for an application, choices between
contradictory beliefs must be made. To make them, an application designer can
exploit i) the statements describing or evaluating the creators of the beliefs, ii) the
corrective/argumentation and specialization relations between the beliefs, and
more generally, iii) their evaluations via meta-statements (see the next point). For
example, an application designer may choose to select only the most specialized or
restricted beliefs of knowledge providers having worked for more than 10 years in
a certain domain. Thus, this approach is unrelated to defeasible logics and avoids
the problems associated with classic "version management" (furthermore, as above
explained, in a cbwoKB, neither formal terms nor statements have to evolve in
time).
This approach assumes that all beliefs can be argued against and hence be
"corrected". This is true only in a certain sense. Indeed, among beliefs, one can
distinguish "observations", "interpretations" ("deductions" or "assumptions"; in
this approach, axioms are considered to be definitions) and "preferences";
although all these kinds of beliefs can be false (their authors can lie, make a
mistake or assume a wrong fact), most people would be reluctant to argue against
self-referencing beliefs such as "u2 likes flowers"_[u2] and "u2 is
writing this sentence"_[u2]. Instead of trying to formalize this into
exceptions, the editing protocols of WebKB-2 rely on the reluctance of people to
argue against such beliefs that should not be argued against.
7. Like all descriptions of techniques, statement/creator evaluation techniques are
considered as term definitions and are automatically organized into the extended
specialization hierarchy. To support more knowledge filtering or decision making
possibilities and lead the users to be careful and precise in their contributions, a
cbwoKB server must propose "default measures" deriving a global evaluation of
each statement/creator from i) users' individual evaluations of these objects, and
ii) global evaluations of these users. Details are given in the next section. These
measures should not be hard-coded but explicitly represented (and hence be
executable by the cbwoKB) to let each user specialize them for its goals and
preferences. Indeed, only the user can find the criteria (e.g., originality, popularity,
acceptance, ..., number of arguments without objections on them) and weighting
schemes that suit him. Then, since the results of these evaluations are also
statements, they can be exploited by queries on the objects and/or their creators.
Furthermore, before browsing or querying the cbwoKB, a user should be given the
opportunity to set "filters for certain objects not to be displayed (or be displayed
only in small fonts)". These filters may set conditions on statements about these
objects or on the creators of these objects. They are automatically executed queries
over the results of queries. In WebKB-2, like conceptual querying, filtering is
based on a search for extended specializations. Filters are useful when the user is
overwhelmed by the amount of information in an insufficiently organized part of
the KB.
8. The approach described by the previous points is incremental and works on semi-
formal KBs. Indeed, the users can set corrective or specialization relations between
objects even when WebKB-2 cannot detect an inconsistency or redundancy. As
noted above, a new informal statement must be connected via an argumentation
relation (e.g., a corrective relation) to an already stored statement. For this relation
to be correct, this new statement should generally not be composed of several sub-
statements. However, allowing the storing of (small) paragraphs within a statement
eases the incremental transformation of informal knowledge into (semi-)formal
knowledge and allows doing so only when needed. This is necessary for the
general acceptance of the approach.
With these editing protocols, each object is connected to at least another object via
relations of specialization/generalization, identity and/or argumentation. They permit
a loss-less information integration, since no knowledge selection has to be made. They
can be seen as enabling a precise asynchronous dialogue between knowledge
providers. To sum up, they permit, enforce or encourage people to interconnect their
knowledge into a shared KB, while keeping the KB consistent but without having to
discuss and agree on terminology or beliefs.
Since the techniques described in this article work on semi-formal KBs and are not
particularly difficult for information technology amateurs - since the minimum these
techniques require is for the users to set the above mentioned relations from/to each
term or statement - they can be used in (semantic) wikis to avoid their governance
problems cited in the introduction and other problems caused by their lack of
structure. More generally, the presented approach removes or reduces the file-based
approach problems listed in the previous section, without creating new problems. Its
use would allow merging of (the information discussed or provided by the members
of) many communities with similar interests, e.g., the numerous different
communities working on the Semantic Web. From an application viewpoint, the
approach seems interesting to allow the collaboratively building of states of the art in
scientific domains, corporate memories, catalogues, e-learning, e-government, e-
science, research, etc.
The hypotheses of this approach are that i) conflicts can always be solved by adding
more precision (e.g., by making their sources explicit: different "observations",
"interpretations" or "preferences"), ii) solving conflicts in a loss-less way most often
increases or maintains the precision and organization of the KB, and iii) different,
internally consistent, ontologies do not have to be structurally modified to be
integrated (strongly inter-related) into a unique consistent semantic network. None of
the various kinds of integrations or mappings of ontologies that I made invalidated
these hypotheses.
4 Evaluating Objects and Sources
Many information repositories support free-text/numerical evaluations on objects or
files by people and then display them or statistical measures on them. For example,
Knowledge Zone [8] allows each of its users to i) rate ontologies with numerical or
free text values for criteria such as "usage", "coverage", "correctness" and "mappings
to other ontologies", ii) rate other users' ratings, and iii) use all these ratings to retrieve
and rank ontologies. Such evaluations have several problems: i) the evaluations are not
organized into a semantic network, ii) the above examples of criteria and their
numerical values are not about objects in the ontologies and hence do not help in
choosing between objects, iii) multi-criteria decision making is difficult since two sets
of (values for) criteria are rarely comparable (indeed, one set rarely includes all the
criteria of the other set and, at the same time, has higher values for all these criteria),
and iv) similarity measures on criteria only permit retrieval of possibly "related"
ontologies: the work of understanding, comparing or merging their statements still has
to be (re-)done by each user.
In a cbwoKB, these problems are strongly reduced, since evaluations are on objects
and are themselves objects: they are managed/manageable like other objects and are
integrated into a network of specialization, correction and argumentation relations. As
previously noted, a cbwoKB should provide "default global measures" for the
evaluation of each statement/creator (based on each user's individual evaluations) and
allow the users to refine it. Here are comments (general ones due to space restrictions)
on the global measures that are currently being implemented in WebKB-2.
A global measure of how consensual a belief is should take into account i) the
number of times it has been re-used or marked as co-believed, and ii) its
argumentation structure (i.e., how its arguments/objections are themselves
(counter-)argued). A simple version of such a measure was implemented in the
hypertext system SYNVIEW [9]. The KB server Co4 [4] had protocols based on
peer-reviewing for finding consensual knowledge; the result was a hierarchy of
KBs, the uppermost ones containing the most consensual knowledge while the
lowermost ones were the private KBs of contributing users. Establishing "how
consensual a belief is" is more flexible in a cbwoKB: i) each user can design his
own global measure for what it means to be consensual, and ii) KBs of consensual
knowledge need not be generated.
A global measure of how interesting a statement is should be based on its type (if
it has one, e.g., observation, deduction, assumption, preference, ...), on its
relations (especially those arguing for/against it or representing its originality,
acceptance, ...), and on the usefulness of the authors of these relations (see below).
A global measure of the usefulness of a statement should exploit (at least) the
above two measures.
A global measure of the usefulness of a user U should incorporate the global
measures of usefulness of U's statements and, to encourage participation in
evaluations, the number of objects he evaluated.
Given these comments, the motivation for enabling end-users to adapt the default
measures is clear. However it is done, taking into account the above cited elements
should encourage information providers to be careful and precise in their
contributions and give arguments for them. Indeed, unlike in traditional discussions or
anonymous reviews, careless statements here penalize their authors. This may lead
users not to make statements outside their domain of expertise or without verifying
their facts. (Using a different persona when providing low quality statements does not
seem to be a helpful strategy to escape the above approach, since this reduces the
number of authored statements for the first persona.) For example, when a belief is
objected to, the usefulness of its author decreases and he is therefore led to deepen the
argumentation structure on its belief or remove it.
[6] describes a "Knowledge Web" to which teachers and researchers could add
"isolated ideas" and "single explanations" at the right place, and suggests that this
Knowledge Web could and should "include the mechanisms for credit assignment,
usage tracking and annotation that the Web lacks" (pp. 4-5). [6] did not give hints on
what such mechanisms could be. This article gives a basis for them.
6 Conclusion
This article aimed to show that a cbwoKB - and hence a cbwoKB based ontology
repository - is technically and socially possible, and - in the long term or when
creating a new KB for general knowledge sharing purposes - provides more
possibilities, with on the whole no more costs, than the mainstream approach [14][1]
where knowledge creation and re-use involves searching, merging and creating (semi-)
independent (relatively small) ontologies. However, research on these two approaches
are complementary: i) results on knowledge extraction or merging may ease the
creation of a cbwoKB, ii) the results of applying these techniques with a cbwoKB as
one of the inputs would be better and they would not be lost if stored in a cbwoKB.
This article showed that a cbwoKB can be collaboratively built and evaluated
without a selection committee and without forcing the users to discuss or agree on
terminology and beliefs. However, to guide users into collaboratively representing
knowledge in a normalized and organized way, and hence inserting it "at the right
places", other elements are also needed: expressive and normalizing notations,
methodological guidance, a large general ontology, and an initial cbwoKB core for the
application domain of the intended cbwoKB. WebKB-2 proposes research results for
all these elements. One explored application domain is the "Semantic Web related
techniques".
7 References
1. Bizer, C., Heath, T., Berners-Lee, T.: Linked data - the story so far. International Journal on
Semantic Web and Information Systems, 5, vol. 3, pp. 1-22 (2010)
2. Casanovas, P., Casellas, N., Tempich, C., Vrandecic, D., Benjamins, R.: Opjk and diligent:
ontology modeling in a distributed environment. Artificial Intelligence Law, 15, vol. 2, pp.
171-186 (2007)
3. Chein, M., Mugnier, M.-L.: Positive nested conceptual graphs. In: ICCS 1997, LNAI 1257,
pp. 95-109 (1997)
4. Euzenat, J.: Corporate memory through cooperative creation of knowledge bases and hyper-
documents. In KAW 1996, pp. (36)1-18 (1996)
5. Euzenat, J., Mbanefo, O., Sharma, A.: Sharing resources through ontology alignment in a
semantic peer-to-peer system. In "Cases on semantic interoperability for information systems
integration: practice and applications", pp. 107-126 (2009)
6. Hillis, W.D.: Aristotle (the knowledge web), Edge Foundation, Inc., 138 (2004)
7. Lee, J., Park, J., Park, M., Chung, C., Min, J.: An intelligent query processing for distributed
ontologies. Systems and Software, 83, vol. 1, pp. 85-95 (2010)
8. Lewen, H., Supekar, K.S., Noy, N.F., Musen, M.A.: Topic-specific trust and open rating
systems: An approach for ontology evaluation. In EON 2006 at WWW 2006.
9. Lowe, D.: Co-operative structuring of information: The representation of reasoning and
debate. International Journal of Man-Machine Studies, 23, pp. 97-111 (1985)
10. Martin, Ph., Eboueya, M.: For the ultimate accessibility and re-usability. Chapter 29 (14
pages) of the Handbook of Research on Learning Design and Learning Objects: Issues,
Applications and Technologies, IGI Global (2008)
11. Martin Ph.: Exploitation de graphes conceptuels et de documents structurés et hypertextes
pour l'acquisition de connaissances et la recherche d'informations. Ph.D. thesis (378 pages),
University of Nice - Sophia Antipolis, France (1996)
12. Noy, N.F., Tudorache, T.: Collaborative ontology development on the (semantic) web. In:
AAAI Spring Symposium on Semantic Web and Knowledge Engineering (SWKE 2008)
13. Palma, R., Haase, P., Wang, Y., d'Aquin, M.: Propagation models and strategies. Deliverable
1.3.1 of NeOn - Lifecycle Support for Networked Ontologies; NEON EU-IST-2005-027595
(2008)
14. Shadbolt, N., Berners-Lee, T., Hall, W.: The semantic web revisited. IEEE Intelligent
Systems, 21, vol. 3, pp. 96-101 (2006)
15. Sowa, J.: Theories, models, reasoning, language, and truth.
http://www.jfsowa.com/logic/theories.htm (2005)