The document presents a new ontology matching system based on a multi-agent architecture. The system takes ontologies described in XML, RDF Schema, and OWL as input. It uses multiple matchers and filtering to generate mappings between ontology entities. The mappings are then validated. The system is implemented as a multi-agent system with different agent types responsible for resources, matching, generating mappings, and filtering/validating mappings. The architecture allows for robust, flexible, and scalable ontology matching.

1. International Association of Scientific Innovation and Research (IASIR)
(An Association Unifying the Sciences, Engineering, and Applied Research)
International Journal of Emerging Technologies in Computational
and Applied Sciences (IJETCAS)
www.iasir.net
IJETCAS 14-639; © 2014, IJETCAS All Rights Reserved Page 295
ISSN (Print): 2279-0047
ISSN (Online): 2279-0055
Towards a new ontology matching system through a multi-agent architecture
Jihad Chaker, Mohamed Khaldi and Souhaib Aammou
LIROSA, Faculty of Sciences, Abdelmalek Essaadi University,
B.P.2121, Mhanech II, Tetouan,
MOROCCO
Abstract: This paper presents a new method of ontology matching to improve semantic interoperability. This method takes as input ontologies described in XML, RDF Schema and OWL format. The proposed matching process involves several stages through the analysis of ontologies entities sources, calculates the terminological similarity with several matchers to maximize the discovery of many similar couples. Once the mapping hypotheses are generated, a filtering system is in place to ensure the quality of alignments. The system architecture is based on a multi agent system, each agent has its own behavior and communicates with the common environment to produce mappings between ontologies source.
Keywords: Ontology; Ontology Matching; Semantic Interoperability; Multi-Agent System; Matchers; Mappings
I. Introduction
The notion of ontology is related to the field of philosophy, it comes from the greek word (Ontologia), meaning speaking (logia) about being (onto), Ontology refers to the theory of being as being. In the context of artificial intelligence and more specifically in knowledge engineering, ontology is rich in definitions, the most commonly cited, is that given by Gruber [1] , an ontology is defined as an explicit specification of a conceptualization. Studer added to the generic definition of Gruber the sharing criterion: Ontologies are a formal, explicit specification of a shared conceptualization [2]. A more recent generic definition is given by Christopher ROCHE. [3]: « Ontology is a conceptualization of a domain which is associated with one or more vocabularies of terms. The concepts are structured in a system and participate in the meaning of terms. Ontology is defined for a given purpose and expresses a view shared by a community. An ontology is expressed in language (representation) based on a theory (semantics) that guarantees the properties of the ontology in terms of consensus, coherence reuse and sharing».
Ontology matching is a solution to the semantic heterogeneity problem. It finds correspondences between semantically related entities of ontologies. These correspondences can be used for various tasks, such as ontology merging, query answering, or data translation. Thus, matching ontologies enables the knowledge and data expressed with respect to the matched ontologies to interoperate [4]. L’objectif majeur est l’établissement de liens de correspondances entre les ontologies originales, précisément entre les concepts from the two ontologies ,the estimated similarity between the two concepts et le type des relations inter-ontologies.
Also according Euzenat [4], The matching process can be seen as a function f which, from a pair of ontologies to match o and o’, an input alignment A, a set of parameters p and a set of oracles and resources r, returns an alignment A’ between these ontologies: . This can be schematically represented as illustrated in Figure 1:
Figure 1: The ontology matching process.
Several systems of ontology matching have implemented, we quote: SAMBO [5], Falcon [6], OLA [7], QOM [8], DSsim [9], RiMOM [10], ASMOV [11]. The increasing number of methods available for schema or ontology

2. Jihad Chaker et al., International Journal of Emerging Technologies in Computational and Applied Sciences, 9(3), June-August, 2014, pp.
295-299
IJETCAS 14-639; © 2014, IJETCAS All Rights Reserved Page 296
matching mandate consensus for evaluation of these methods. The Ontology Alignment Evaluation Initiative is a coordinated international initiative (OAEI) to forge this consensus.
The paper is organized as follows. The second section discusses the agents and multi-agent systems. In the third section, the process of the new method for ontology matching is described. The fourth section shows the implementation of the system based on agents.
II. Agents and multi-agent systems
In the literature we find a multitude of definitions of agents. They all look alike, but they differ depending on the type of application for which the agent is designed. One of the first definitions of agent due to Ferber [12]: An agent is a real or abstract autonomous entity which is able to act on itself and its environment, which, in a multi- agent universe, can communicate with other agents, and whose behavior is the result its observations, knowledge and interactions with other agents. Russell added that the agent is an entity that senses its environment and acts upon it [13].
One of the most comprehensive definition of agents, that I particularly favor, is the one given by Wooldridge and Jennings [14].in which an agent is: “ a hardware or (more usually) a software-based computer system that enjoys the following properties: autonomy - agents operate without the direct intervention of humans or others, and have some kind of control over their actions and internal state; social ability - agents interact with other agents (and possibly humans) via some kind of agent-communication language; reactivity: agents perceive their environment and respond in a timely fashion to changes that occur in it; pro-activeness: agents do not simply act in response to their environment, they are able to exhibit goal-directed behaviour by taking initiative.”
The agent is capable of acting on its environment, act and control its own shares without the intervention of a third party (human or agent), take the initiative at the right time, respond in time and interact with other agents to perform tasks or help these agents to do theirs. Depending on the type of agent used in an application, there are systems of cognitive and reactive agents: The first is based on the cooperation of agents able alone to perform complex operations but the reactive agent systems have only a protocol and a very small communication language, respond only to the law "stimulus / response".
Multi-agent systems have emerged with the advent of distributed artificial intelligence. Unlike the traditional artificial intelligence, which models the intelligent behavior of a single agent, distributed artificial intelligence is concerned with intelligent behavior are the products of cooperative activity several agents.
A Multi-Agent System is a set of agents operating in a common environment, which means the real world or the virtual world.
According to Ferber [12] A Multi-Agent System is a system consisting of:
 E environment, a space with a generally metric.
 A set of objects O. These objects are located, that is to say that, for any object, it is possible, at a given time, to associate a position in E. These objects are passive; they can be perceived, created, destroyed and modified by the agents.
 A set A of agents, which are specific objects (A ⊆ O), which represent the active entities of the system.
 A set of relations R that unite objects (and thus agents) to each other.
 A set of operations Op allowing agents of A to perceive, produce, consume, transform and manipulate objects from O.
 Operators responsible for representing the application of these operations and the world's reaction to this attempt to change, that the laws of the universe will be called.
III. Our approach to ontology matching
A. Process of matching
Once the extraction of the basic concepts description languages (XML, RDF Schema, OWL) is made, ontologies target entities to make it usable for analysis during the calculation of similarity terminology, we analyze the target ontology entities to make them usable when calculating similarity terminology, it involves standardizing the entities. Preprocessing comments and labels as necessary to support the calculation of similarity, eliminating words that do not carry useful information.
The purpose of calculating similarity terminology is to maximize the discovery of many similar couples and reduce the number of those who are dissimilar. Our system uses multiple matchers, including syntactic and lexical matchers. Syntactic matchers calculate the similarity or dis-similarity between two strings using functions and methods of comparison based on a sequence of characters. this system uses the following matchers:
 N-gram [15] Many works have shown the efficacy of n-grams as a method for representing texts for their classification. This test takes as input two strings and calculates the number of common n-grams between them. Let ngram (s, n) be the set of substrings of s (augmented with n−1 irrelevant characters at the
beginning and the end) of length n, the n-gram distance is a dissimilarity
such that:
The normalized version of this function is:

3. Jihad Chaker et al., International Journal of Emerging Technologies in Computational and Applied Sciences, 9(3), June-August, 2014, pp.
295-299
IJETCAS 14-639; © 2014, IJETCAS All Rights Reserved Page 297
This function is quite efficient when characters are only missing.
 Edit distance [15]: Given a set Op of string operations , and a cost function , such that for any pair of strings there exist a sequence of operations which transforms the first one into the second one (and vice versa), the edit distance is a dissimilarity , such that , is the cost of the less costly sequence of operations which transform s in t.
 Jaccard similarity: calculates the similarity between two sets of elements by comparing the number of common elements to the total number of elements belonging to both sets. A value between 0 and 1 is obtained, which corresponds to the identical assemblies 1.
To find associations between entities or classes, linguistic matchers are used based on the external resource, mainly WordNet dictionary [16].
The results of individually previous matchers are combined to generate a single mapping between each pair of concepts. After a filter based on a similarity threshold is applied to reduce the number of false assumptions mapping
After the filter similarity, structural and semantic matchers intervene to find new relations similarity. Structural methods determine the similarity between two entities based on structural information. Indeed, the entities are connected together by links of the semantic or syntactic, this process provides the use of:
 Internal structural Methods: operate only the information describing the attributes of entities, more specifically, it uses the information contained in the internal structures of the entities for calculating similarity (eg, value interval, cardinality of attributes, etc.).
 External structural methods: compare relationships with other entities
The technical structural techniques implement various heuristics and are based on the hypothesis [17]: “if two entities both ontologies are similar, their neighbors are also somehow”, we propose the calculation of the structural similarity between entities in the ontologies, one inspired by the work of Abolhassani [18].
Still with the aim of improving the quality of the matching, we thought of using both approaches semantic methods. The first approach is based on logic models, while the second approach includes methods of deduction to derive the similarity between two entities. The filter system and validation intervenes once again after the generation of mappings.
B. Comparison of our system with other systems of ontology matching
The following table compares our system with other existing systems, based on a set of key criteria of alignment methods such as input formats, the outputs of alignments between concepts and relationships, the validation system of the mappings generated, also the extensional methods used and semantic filtering.
Table I: Comparison table between our system and other systems
System
Input
Output
Validation
Extensional
Semantic
DSsim
OWL, SKOS
1 :1 alignments
expert
-
yes
RiMOM
OWL
1 :1 alignments
expert
Vector distance
-
ASMOV
OWL
n :m alignments
expert
Object similarity
yes
AgreementMaker
XML, RDFS , OWL and N3
n :m alignments
expert
-
-
Our system
XML, RDFS and OWL
n :m alignments
Expert and automatic (agent)
-
yes
IV. Agents architecture of our system
Multi-agent systems are now a technology of choice for the design and implementation of distributed applications and cooperatives. The proposed architecture is based on four types of agents, namely: resource Agent (RA), the matchers Agent (MA), agent of Generating the mappings (MGA), Agent Filtering hypothesis and Validation (FVA). The system is not centralized, and each agent has its own behavior with his entourage (which can be an agent or an external user). Transmitting and / or receiving results as messages.
Figure 2 illustrates the general behavior of a multi-agent system, presented in the form of a agent interaction protocol (AIP) tell defined in AUML [19], and that the messages provides standardized communication, we chose the FIPA agent communication language (ACL).

4. Jihad Chaker et al., International Journal of Emerging Technologies in Computational and Applied Sciences, 9(3), June-August, 2014, pp.
295-299
IJETCAS 14-639; © 2014, IJETCAS All Rights Reserved Page 298
Figure 2: Interactions between agents.
V. Conclusion
We describe in this paper a new method of matching of anthologies. It is based on a process comprising the similarity calculation, and generation of mappings, filtering and validation. One of the highlights of our system is the ability to integrate several matchers, the filtering system and semi-automatic validation, which reflects positively in values of quality metrics alignment (precision measurements, recall Fallout and Fmesure) and subsequently ensure the quality of matching.
The implementation as a system-based agents, obviously inherits the benefits of these systems such as robustness, flexibility and scalability.
The next job is to evaluate the performance of our algorithm, through a series of tests it using a few basic tests provided in the Benchmark at the disposal of the international community by EON competition [20], as a comparison with other methods
VI. References
[1] T. R. Gruber, “A Translation Approach to Portable Ontology Specifications”, Knowledge Acquisition, 1993.

5. Jihad Chaker et al., International Journal of Emerging Technologies in Computational and Applied Sciences, 9(3), June-August, 2014, pp.
295-299
IJETCAS 14-639; © 2014, IJETCAS All Rights Reserved Page 299
[2] R. Studer, R. Benjamins and D. Fensel, “Knowledge engineering: Principles and methods Data & Knowledge Engineering”, March 1998.
[3] C. Roche, “Terminologie et ontologie ”.In Revue Langages, 2005.
[4] J. Euzenat and P. Shvaiko, “Ontology matching”. Springer, 2007.
[5] P. Lambrix and H. Tan, “SAMBO – a system for aligning and merging biomedical ontologies”, Journal of Web Semantics, vol. 4, no. 1, pp. 196–206, 2006.
[6] W. Hu, Y. Qu, and G. Cheng, “Matching large ontologies: A divide-and-conquer approach”, Data and Knowledge Engineering, vol. 67, no. 1, 2008.
[7] Euzenat J., Loup D., Touzani M., Valtchev P., “Ontology Alignement with OLA ”, Proceedings of the 3rd International Workshop : Semantic Web Conference EON, Hirochima, Japan, p. 341-371, November, 2004.
[8] Ehrig M., Staab S., “QOM : Quick Ontology Mapping ”, Proceedings of The 3rd ISWC, GI Jahrestagung (1), Hiroshima, Japon, p. 356-361, November, 2004.
[9] M. Nagy, M. Vargas-Vera, and P. Stolarski, “Dssim results for OAEI 2009”, in Proc. 4th International Workshop on Ontology Matching (OM) at the International Semantic Web Conference (ISWC), pp. 160–169, 2009.
[10] J. Li, J. Tang, Y. Li, and Q. Luo, “Rimom: A dynamic multistrategy ontology alignment framework”, IEEE Transactoins on Knowledge and Data Engineering, vol. 21, no. 8, pp. 1218–1232, 2009.
[11] Y. R. Jean-Mary, E. P. Shironoshita, and M. R. Kabuka, “Ontology matching with semantic verification” Journal of Web Semantics, vol. 7, no. 3, pp. 235–251, 2009.
[12] Ferber, J., “Les systèmes multi-agents: Vers une intelligence collective”, InterEditions, 1995.
[13] Russell, S.J. , “Rationality and intelligence”. Artificial Intelligence, Vol. 94, p.57-77,1997.
[14] Wooldridge,M and N. R. Jennings. “Agent theories, architectures, and languages”. In Wooldridge and Jennings, eds. Intelligent Agents, Springer Verlag, p.1-22,1995.
[15] EUZENAT J., BACH T., BARRASA J., BOUQUET P., BO J. D., DIENG R., EHRIG M., LARA R., MAYNARD D., NAPOLI A., STARMOU G., STUCKENSCHMIDT H., SHVAIKO P., TESSARIS S., ACKER S. V. and ZAIHRAYEU I., “ State of art on ontology alignment, Technical Report ”KWEB/2004/D2.2.3/v1.2, Knowledge WebConsortium, August, 2004
[16] George A. Miller, “WordNet: A lexical database for english”. communication of the ACM. pp. 39-41. 1995.
[17] Euzenat J., Valtchev P., “Similarity-based ontology alignment in OWL-lite”. In Proceedings of the European Conference on Artificial Intelligence (ECAI), pages 333–337, 2004.
[18] Abolhassani H. , B.B. Hariri, S. H. Haeri, “On Ontology Alignment Experiments”, Webology, Volume 3, Number 3, September, 2006.
[19] BAUER, B., J.P. MÜLLER and J. ODELL , “Agent UML: a formalism for specifying multiagent interaction”, International Journal of Software Engineering and Knowledge Engineering, 11 (3), 207–230. 2001.
[20] Eon, « EON 2007 : Evaluation of Ontology for the Web », Proceedings of the 5th International EON Workshop, Ontology Alignment Evaluation Initiative Test library ,2007.