Automatic Semantic Annotation: Towards Resolution
of WFIO Incompatibilities
Chahrazed Tarabet, Meriem Mouzai and Ali Abbassene
Centre de Développement des Technologies Avancées, Baba Hassen, Algiers, Algeria
Keywords: IOWF, NLP, Automatic Semantic Annotation, Ontology.
Abstract: Inter-organizational workflows (IOWF) allow for orchestration of processes between different
organizations, but the incompatibility they reveal poses a serious problem. Nevertheless, there are
approaches that can remedy this problem, notably the semantic annotation. In this position paper, we will
present a study whose objective is to address the detection and correction of these incompatibilities between
workflow partners. For this purpose, amelioration, optimization and automation are necessary for the
semantic annotation phase of inter-organizational workflows, in order to achieve the IOWF incompatibility
resolution.
1 INTRODUCTION
An organization is a coordinating unit, with
identifiable boundaries, working to achieve a goal
shared by its participating members. Nowadays, the
company is no longer industrial but commercial, it is
no longer located in a single place, it is extended, it
even happens not to be fully visible but to be (in part)
virtual. These companies need inter-organizational
cooperative systems, that is, across multiple network
organizations. In this context, information and
communication technologies play an essential role in
enabling enterprises to exchange all types of data.
The inter-organizational workflows go in the
same direction, allowing an orchestration of
processes between several organizations. However,
the incompatibility of the latter poses a serious
problem (Abbassene, Alimazighi and Aouachria,
2015). Nevertheless, there are approaches that can
remedy this problem, notably the semantic
annotation. This approach represents a mechanism for
linking a data to its semantic description represented
by a concept derived from an ontology. It is an
effective way to detect and correct these
incompatibilities between workflow partners. For this
purpose, amelioration, optimization and automation
are necessary for the semantic annotation phase of
inter-organizational workflows.
In this position paper, we will present a critical
review of the works that deal with the automation of
semantic annotation phase, using techniques such as
NLP (Natural Language Processing). Thus, we
propose a solution to the problematic posed according
to the results obtained.
2 LITERATURE REVIEW
In this section, we will present the works of the
semantic annotation domain found in the literature
that we judge interesting.
Authors in (Davis et al., 2009) use Controlled
Natural Languages (CNLs) which are subsets of
natural language whose grammars and dictionaries
have been restricted in order to reduce or eliminate
both ambiguity and complexity.
These languages prompt the novice user to
annotate, while simultaneously creating, their
respective documents in a user-friendly way, while
protecting them from the formalisms of
representation of the complex underlying knowledge.
CNLs have already been applied successfully in the
context of authoring ontology, but very little research
has focused on CNLs for semantic annotation. They
describe here a user-friendly semantic annotator,
based on CLIE (Controlled Language for Information
Extraction) tools.
However, this annotator only allows non-expert
users to write and annotate minutes of meetings and
semi-automatic status reports using controlled natural
language.
Also, this other work (semantator) (Tao et al.,
556
Tarabet, C., Mouzai, M. and Abbassene, A.
Automatic Semantic Annotation: Towards Resolution of WFIO Incompatibilities.
DOI: 10.5220/0006377505560562
In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 2, pages 556-562
ISBN: 978-989-758-248-6
Copyright © 2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
2013), which is an annotation tool that allows to
annotate documents with semantic Web ontologies
using a loaded free text document and an ontology,
users can annotate document fragments with classes
in the ontology to create instances and link instances
created with ontology properties. Thus, it provides:
1) Basic manual annotation features: creation /
deletion of ontology instance, creation / deletion of
relationship, binding of equivalent instances and
export / reload of existing annotations;
2) Automatic annotation by connecting to the
annotator NCBO and cTAKES;
3) Basic reasoning support based on the
underlying semantics of the preachers owl:
disjointWith and owl: equivalentClass.
In contrast, this tool only allows semi-automatic
annotation and focuses on clinical documents.
The main objective of the works (Kiyavitskaya et
al., 2006) and (Kiyavitskaya et al., 2005) is to present
a methodology supported by tools that semi-
automates the semantic annotation process for a set of
documents in relation to a semantic model (ontology
or conceptual schema). It is proposed to address the
problem using highly efficient and proven methods
and tools in the field of software analysis for
processing billions of source code lines of legacy
software.
The semantic annotation method of documents
uses generalized syntactic analysis and the TXL
structural transformation system, the basis of the LS /
2000 (Kiyavitskaya et al., 2006) automated system.
TXL is a programming language specially designed
to allow, for example, rapid prototyping of language
descriptions, tools and applications. The system
accepts as input a grammar and a document,
generates an analysis tree for the input document and
applies transformation rules to generate an output in a
target format (Kiyavitskaya et al., 2006), this
transformation phase is not yet implemented in the
work (Kiyavitskaya et al., 2005).
The disadvantage of these approaches is that it
deals only with the semi-automatic annotation, so the
results of the document (Kiyavitskaya et al., 2005)
are adopted only in the tourism sector.
Antti Vehvilainen in this work (Vehvilainen,
Hyvonen and Alm, 2008) dealt with applications of
semantic Web technologies to help office services.
They focus on QA (Questions / Answers) support
services, where the service database is composed of
answers to the previous questions, that is, QA pairs.
They propose a semi-automatic semantic annotation
of natural language text for the question-answer (QA)
pairs annotation and case-based reasoning techniques
to find similar questions. The methodology consists
of using semantic Web technologies in content
annotation, using the QA repository and integrating
the information available online on the Web with the
creation process and responses. They consider here
the usefulness of CBR (Case-Based Reasonning) in
the indexing and the extraction of informations since
the similar pairs of QAs reproduce in the services of
QA. Case-based reasoning (CBR) is a problem-
solving paradigm in artificial intelligence where new
problems are solved based on previously experienced
similar problems. The CBR cycle consists of four
phases:
1) Retrieve the most similar case (s),
2) Reuse the recovered cases to solve the problem,
3) Revise the proposed solution, and
4) Retain the solution as a new case in the base
of cases.
Nevertheless, this approach is not generalized in
all domains and only allows the semi-automatic
annotation.
They treat here (Qacim and Salih, 213), the
semantic similarity between sentences based on
WordNet semantic dictionary. The proposed
algorithm will be based on a number of resources,
including Ontology and WordNet.
The goal of this research (Qacim and Salih, 213)
is to create an efficient automatic annotation platform
that develops a way to automatically generate
metadata to semantically annotate Web documents
that improve information retrieval. The proposed
system should be easily understood by non-technical
users who may not be familiar with the technical
language used to create ontologies. The proposed
system provides ontological similarity to determine
the relationships between words in sentences and
concepts in ontology. It has been found that the
meaning of the term similarity is ambiguous because
of its use in many different contexts, such as
biological, logical, statistical, taxonomic,
psychological, semantic, and many other contexts, to
resolve ambiguities, WordNet Must be used to
provide a lexical ontology.
The semi-automated annotation of texts in natural
language was approached in this work (Erdmann et
al., 2000) by designing an information extraction-
based approach for semi-automated annotation, which
was implemented on SMES (Saarbrucken Message
Extraction System), (Erdmann et al., 2000) which
includes a tokenizer based on regular expressions. It
is a generic component that respects several
principles that are crucial to its objectives. (i) it is fast
and robust, (ii) it maps terms to ontological concepts,
(iii) produces dependency relationships between
terms, and (iv) is easily adaptable to new domains. As
Automatic Semantic Annotation: Towards Resolution of WFIO Incompatibilities
557
in this approach, finite state technologies support
lexical acquisition as well as semantic marking. The
goal of the global process is the generation of so-
called lexical networks that can be used to enable
automatic and semi-automatic construction of texts
on the Web. Incoming documents are processed using
the SMES information retrieval system. SMES
associates simple words or complex expressions with
a concept of ontology, linked by the domain lexical
link. Recognized concepts and relationships between
concepts are underlined as suggested annotations.
This mechanism has the advantage that all relevant
informations in the ontology document are
recognized and proposed to the annotator, but it’s
applied only on text in natural language and is not
fully automated.
Aurélie Névéol
in this article (Névéol, Doğan and
Lu, 2011) dealt with the production of annotated data
which is a necessary step for many natural language
processing (NLP) or information processing tasks.
They have studied the semantic annotation of a
large number of biomedical requests. This study
shows that automatic pre-annotations are considered
useful by most annotators to speed up the annotation
rate and improve the consistency of the annotations
while maintaining a high quality of the final
annotations. The disadvantage of this work is that its
field of application is restricted to the biomedical
domain.
This work (Dingli, Ciravegna and Wilks, 2003)
proposes a methodology to learn to automatically
annotate specific domain information from large
repositories with minimal user intervention. The
methodology is based on a combination of
information retrieval, information integration and
automatic learning. Learning is sown by extracting
information from structured sources. The retrieved
information is then used to partially annotate
documents. These annotated documents are used for
learning bootstrap for simple information extraction
(IE). It will be used to form more complex IE engines
and the cycle will continue to repeat until the required
information is obtained. User intervention is limited
to providing an initial URL and correcting
information if this is the case when the calculation is
completed. The revised annotation can then be re-
used to provide additional training and thus obtain
more information and / or more precision.
The methodology was fully implemented in
Armadillo, a system for extracting and integrating
unsupervised data from large collections of
documents.
This other work (Kiryakov et al., 2004) seeks to
create an efficient, robust and scalable architecture
for automatic semantic annotation, and to implement
this architecture in a component-based platform for
indexing and semantic search on large collections of
documents. What is considered a primary innovative
contribution, is the fact that it offers an end-to-end
extensible system that processes the complete cycle
of creating metadata, storing and semantic search, for
online use that provide navigation semantically
improved. Another approach presented in (Tulasi et
al., 2017) deals with automatic semantic annotation
based on ontologies, where all documents are
collected from the Web and a database is created. The
documents are then given as an input to make a
semantic annotation on the ontology.
In this paper (Kiryakov et al., 2004) authors
present a holistic architecture view for automatic
semantic annotation with references to classes in
ontology and instances, on the basis of these semantic
annotations, it has indexed and retrieved documents.
A system (called KIM), implementing this concept
(Popov et al., 2003), provides a new infrastructure
and knowledge and information management services
for automated semantic annotation, indexing and
document retrieval, it provides also a mature
infrastructure for scalable and customizable
information extraction (IE) and annotation and
document management, based on GATE.
GATE (General Architecture for Text
Engineering) (Ranganathan, Biletskiy and
Kaltchenko, 2008), developed by the University of
Sheffield, is an efficient tool used to perform some
Natural Language Processing (NLP) operations. It
has many features, such as manual annotation,
automatic annotation using a variety of
nomenclatures, information retrieval, ontology-based
processing, and so on. GATE has played a major role
in text annotation, which can be presented in different
formats, such as: eml, mail, text, dhtm, pdf, xhtml,
xml, rtf, txt, sgm, Sgml, htm, etc. Annotations are
mainly performed to communicate semantics in
electronic documents and / or underlining text that
provides better human understanding.
From a technical point of view, the platform
allows KIM-based applications to use it for automatic
semantic annotation, retrieval of content based on
semantic restrictions, and querying and modifying
ontologies and underlying knowledge bases.
In this article (Leopold et al., 2015), they present
an approach to automatically annotate process models
with the concepts of a taxonomy. At this point, the
focus is on business-based taxonomies, such as the
Supply Chain Operations Reference Model (SCOR),
the MIT Process Manual, and the Process
Classification Framework (PCF). To do this, they
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
558
propose an approach that combines the measurement
of semantic similarity with probabilistic optimization.
In particular, they use different types of similarity
between the process model and the taxonomy and the
distance between the concepts of taxonomy to guide
the matching with a formalization of the Markov
logic.
The semantic annotation is also used in the S-
CREAM project. The approach is interesting for the
strong implication of the automatic learning
techniques for an automatic extraction of the relations
between the annotated entities (Handschuh, Staab and
Ciravegna, 2002).
A similar approach is also taken in the MnM
project, where semantic annotations can be placed
online in the content of the document and refer to an
ontology and a KB (WebOnto) server, accessible via
an API (Vargas-Vera et al., 2002).
The QuizRDF module, used to provide
improvements to a standard full-text search with the
metadata extracted from OntoBuilder. QuizRDF is an
important source of inspiration for the design of KIM.
An interesting indexing of the named entity and a
system of questions / answers. Once indexed, the
content is queried via NL questions, with the NE
mark on the question used to determine the type of
response expected (Davies, Fensel and Van
Harmelen, 2003).
AeroDAML takes a similar approach to KIM,
but implements it as a prototype of research on a
much smaller scale (Kogut and Holmes, 2001).
SemTag is a platform closer in terms of
objectives and architecture to KIM, which performs
the semantic annotation on a large scale compared to
the TAP ontology which is very similar in size and
structure to the KIM and KB ontologies. SEMTAGS
gradually creates a first-order markov model based on
existing annotations and proposes a semantic
annotation, new syntactic trees. It first performs a
search phase, annotating all possible references to
instances of the TAP ontology. In the second phase of
disambiguation, SemTag uses a vector space model to
assign the correct ontological class or to determine
that this statement does not correspond to a class in
TAP. Disambiguation is performed by comparing the
context of the current statement (10 words on the left
and 10 on the right) with the case contexts in TAP
with compatible aliases (Dill et al., 2003) (Guha and
McCool, 2001).
With regard to semi-automatic semantic
annotation mechanisms, Pustejovsky describes the
approach for semantic indexing and typed hyperlinks
(Pustejovsky et al., 1997).
Another approach to semantic annotation of data
has improved the retrieval of information and
improved interoperability where a new approach
based on NLP ontology has been proposed and
applied on annual reports (Wang et al., 2009).
The results presented in this section are the result
of a detailed study on the state of the art on the
techniques and approaches used for the improvement,
optimization and automation of the semantic
annotation phase.
We have synthesized the results obtained in the
following table:
Table 1: Comparison of the presented works.
Semi-
automatic
Automatic Application Approach used
Suitable for
WFIO
(Davis et al., 2009) yes no
Administration (meeting
minutes - status report)
CNL no
(Tao et al., 2013) yes no Clinic Ontology / NLP no
(Kiyavitskaya et al.,
2006) (Kiyavitskaya
et al., 2005)
yes no Tourism Ontology / TXL no
(Vehvilainen,
Hyvonen and Alm,
2008)
yes no Help Desk Service CBR no
(Qacim and Salih,
213)
- yes Various fields Ontology / WordNet no
(Erdmann et al.,
2000)
yes no Text in natural language
Regular Expressions
/ SMES
no
(Névéol, Doğan and
Lu, 2011)
- yes Biomedical
NLP / pre-
annotations
no
Automatic Semantic Annotation: Towards Resolution of WFIO Incompatibilities
559
Table 1: Comparison of the presented works (cont.).
(Dingli, Ciravegna
and Wilks, 2003)
- yes Various fields
IE / Automatic
learning
no
(Kiryakov et al.,
2004) (Popov et al.,
2003)
- yes
Any type of text (web
page / regular document
(non web))
Ontology / KIM / IE no
(Leopold et al.,
2015)
- yes Various fields Markov Logic no
Based on the results obtained, we consider that the
works (Davis et al., 2009), (Tao et al., 2013),
(Kiyavitskaya et al., 2006), (Kiyavitskaya et al.,
2005), (Vehvilainen, Hyvonen and Alm, 2008) and
(Erdmann et al., 2000) partially automate the
semantic annotation phase while it is completely
automatic in (Qacim and Salih, 213), (Névéol, Doğan
and Lu, 2011), (Dingli, Ciravegna and Wilks, 2003)
and (Kiryakov et al., 2004). Thus, we note that some
works use common approaches including, NLP and
ontology. As for the application, most of the works
focus on a specific field. Finally, we consider
adaptation to inter-organizational workflows an
important criterion because they allow to answer the
problem posed in section 1, they also allow the
collaboration between organizations with a better
exchange of data while being flexible and effective
(Semar-Bitah and Boukhalfa, 2016). However, the
approaches used in the cited works in this paper do
not cover the notion of inter-organizational
workflows.
3 CONTRIBUTION
The aim of this study is to explore the different works
related to the domain of automation of semantic
annotation in order to define a solution whose
objective is to detect and to correct the
incompatibilities between the workflow partners.
To this end, so that organizations can collaborate
with better compatibility, we propose an approach
that aims to automate the semantic annotation phase
for inter-organizational workflows (IOWFs) using the
NLP approach that has proved to be successful. We
recommend the adoption of methods to improve and
optimize the IOWF semantic annotation, namely:
1) Hierarchy: It allows to improve the annotation by
providing a formal framework that allows to argue
on the consistency of the extracted information. In
particular, semantic hierarchies have proved to be
very useful in reducing the semantic gap. Three
types of hierarchies for image annotation and
classification have been recently explored:
1) Hierarchies based on textual knowledge;
2) Hierarchies based on visual (or perceptual)
information, i.e. low-level characteristics of the
image;
3) Hierarchies that we call semantic based on both
textual and visual information (Bannour and
Céline, 2013).
2) Indexing: It makes it possible to document the
knowledge represented by the semantic
annotations, or even to keep them up to date when
the reference texts evolve. In general, it is a
question of using the semantic structure
constructed by the annotations to identify
elements in a document and to navigate semantic
elements to fragments of text or vice versa.
However, an indexing process consists of
annotating text and gathering it by following a
semantic organization (Lévy, Nazarenko and
Guissé, 2010) (Guissé et al., 2010).
3) Learning: A learning process is characterized by
an interaction between the learner and the
environment by setting up a system capable of
learning how to annotate a given corpus. The goal
of this method is to acquire better or new
knowledge and / or a mechanism or procedure
(inference engine and knowledge) by deducing a
set of rules. There are three techniques of learning
in particular, learning patterns, digital learning
and active learning (Bannour and Audibert, 2012).
4 CONCLUSION AND FUTURE
WORKS
In this paper, we have presented the different
approaches to semantic annotation found in the
literature to remedy the problems of incompatibility
of inter-organizational workflows. On the basis of the
study carried out, we have recommended approaches
that can contribute to the automation phase of
semantic annotations while improving and optimizing
the semantic annotations.
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
560
In the future, we want to concretise our vision by
adopting the approaches cited in Section 3 to ensure
better collaboration among workflow partners.
REFERENCES
Abbassene, A., Alimazighi, Z., and Aouachria, M. (2015).
Towards an open framework for inter-organizational
workflow semantic annotation. Algiers, Algeria:
ISPS'15 - 12th International Symposium on
Programming and Systems, USTHB.
Davis, B., Varma, P., Handschuh, S., Dragan, L., and
Cunningham, H. (2009). Controlled Natural Language
for Semantic Annotation: The Semantic Web: Research
and Applications Volume 5554 of the series Lecture
Notes in Computer Science, pp. 816-820.
Tao, C., Song, D., Sharma, D., and Chute, C. G. (2013).
Semantator: Semantic annotator for converting
biomedical text to linked data: Journal of biomedical
informatics Volume 46, Issue 5, pp. 882-893.
Kiyavitskaya, N., Zeni, N., Mich, L., Cordy, G. R. and
Mylopoulos, J. (2006). Text Mining Through Semi
Automatic Semantic Annotation: Practical Aspects of
Knowledge Management Volume 4333 of the series
Lecture Notes in Computer Science, pp. 143-154.
Kiyavitskaya, N., Zeni, N., Cordy, J. R., Mich, L. and
Mylopoulos, J. (2005). Semi-Automatic Semantic
Annotations for Web Documents: Proc. SWAP 2005,
2nd Italian Semantic Web Workshop.
Vehvilainen, A., Hyvonen, E. and Alm, O. (2008). A Semi-
Automatic Semantic Annotation and Authoring Tool for
a Library Help Desk Service: Emerging Technologies
for Semantic Work Environments: Techniques,
Methods, and Applications.
Qacim, A. and Salih, M. (2013). Toward From Manual to
Automatic Semantic Annotation : Based on Ontology
Elements and Relationships: International Journal of
Web & Semantic Technology; Vol. 4 Issue 2, p. 21.
Erdmann, M., Maedche, A., Schnurr, H. P. and Staab, S.
(2000). From manual to semi-automatic semantic
annotation: about ontology-based text annotation tools:
Proceedings of the COLING-2000 Workshop on
Semantic Annotation and Intelligent Content, pp. 79-
85.
Névéol, A., Doğan, R. I. and Lu, Z. (2011). Semi-automatic
semantic annotation of PubMed queries: A study on
quality, efficiency, satisfaction: Journal of Biomedical
Informatics 44, pp. 310-318.
Dingli, A., Ciravegna, F. and Wilks, Y. (2003). Automatic
Semantic Annotation using Unsupervised Information
Extraction and Integration: ceur-ws.
Kiryakov, A., Popov, B., Terziev, I., Manov, D. and
Ognyanoff, D. (2004). Semantic annotation, indexing,
and retrieval: Web Semantics: Science, Services and
Agents on the World Wide Web 2, pp. 49-79.
Tulasi, R. L., Rao, M. S., Ankita, K. and Hgoudar, R.
(2017). Ontology-Based Automatic Annotation: An
Approach for Efficient Retrieval of Semantic Results of
Web Documents: Proceedings of the First International
Conference on Computational Intelligence and
Informatics Volume 507 of the series Advances in
Intelligent Systems and Computing, pp 331-339.
Popov, B., Kiryakov, A., Kirilov, A., Manov, D.,
Ognyanoff, D. and Goranov, M. (2003). KIM –
Semantic Annotation Platform: The Semantic Web -
ISWC 2003 Volume 2870 of the series Lecture Notes in
Computer Science, pp. 834-849.
Ranganathan, G. R., Biletskiy, Y. and Kaltchenko, A.
(2008). Semantic annotation of semi-structured
documents: Electrical and Computer Engineering:
CCECE.
Leopold, H., Meilicke, C., Fellmann, M., Pittke, F.,
Stuckenschmidt, H. and Mendling, J. (2015, June).
Towards the automated annotation of process models :
In International Conference on Advanced Information
Systems Engineering : Springer International
Publishing, pp. 401-416.
Handschuh, S., Staab, S. and Ciravegna, F. (2002,
October). S-CREAM—semi-automatic creation of
metadata: In International Conference on Knowledge
Engineering and Knowledge Management. Springer
Berlin Heidelberg, pp. 358-372.
Vargas-Vera, M., Motta, E., Domingue, J., Lanzoni, M.,
Stutt, A. and Ciravegna, F. (2002, October). MnM:
Ontology driven semi-automatic and automatic support
for semantic markup: In International Conference on
Knowledge Engineering and Knowledge Management.
Springer Berlin Heidelberg, pp. 379-391.
Davies, J., Fensel, D., & Van Harmelen, F. (Eds.). (2003).
Towards the semantic web: ontology-driven knowledge
management. John Wiley & Sons.
Kogut, P. A. and Holmes III, W. S. (2001, October).
AeroDAML: Applying Information Extraction to
Generate DAML Annotations from Web Pages: In
Semannot@ K-CAP 2001.
Dill, S., Eiron, N., Gibson, D., Gruhl, D., Guha, R.,
Jhingran, A., Kanungo, T., Rajagopalan, S., Tomkins,
A., Tomlin, J. A. and Zien, J. Y. (2003, May). SemTag
and Seeker: Bootstrapping the semantic web via
automated semantic annotation: In Proceedings of the
12th international conference on World Wide Web.
ACM, pp. 178-186.
Guha, R. and McCool, R. (2001). Tap: Towards a web of
data.
Pustejovsky, J., Boguraev, B., Verhagen, M., Buitelaar, P.
and Johnston, M. (1997, March). Semantic indexing
and typed hyperlinking: In Proceedings of AAAI Spring
Symposium, NLP for WWW.
Wang, B., Huang, H., Wang, X., & Chen, W. (2009,
December). An Ontology-Based NLP Approach to
Semantic Annotation of Annual Report: In
Computational Intelligence and Security, 2009. CIS'09.
International Conference : IEEE, Vol. 1, pp. 180-183.
Semar-Bitah, K. and Boukhalfa, K. (2016). Towards an
Inter-organizational Collaboration Network
Characterization. In Modelling and Implementation of
Complex Systems: Springer International Publishing,
pp. 233-248.
Automatic Semantic Annotation: Towards Resolution of WFIO Incompatibilities
561
Bannour, H. and Céline, H. (2013). Construction de
hiérarchies sémantiques pour l'annotation d'images :
Revue des Sciences et Technologies de l'Information-
Série RIA: Revue d'Intelligence Artificielle, 27(1), 11-
37.
Lévy, F., Nazarenko, A. and Guissé, A. (2010). Annotation,
indexation et parcours de documents numériques :
Document numérique, 13(3), 121-152.
Guissé, A., Lévy, F., Nazarenko, A., & Szulman, S. (2009).
Annotation sémantique pour l’indexation de règles
métiers : In L’Homme, M.-C. and Szulman, S., editors.
Conférence Internationale sur la Terminologie et
l’Intelligence Artificielle (TIA 2009), page (electronic
medium). Université Paul Sabatier-Toulouse.
Bannour, S., & Audibert, L. (2012, June). Vers une
approche interactive pour l'annotation sémantique : In
IC2012.
ICEIS 2017 - 19th International Conference on Enterprise Information Systems
562