Introduction to the Special Issue: Semantic Matchmaking and Resource Retrieval on the Web

Eugenio Di Sciascio, Francesco M. Donini, and Tommaso Di Noia, Guest Editors
International Journal of Electronic Commerce,
Volume 12, Number 2, Winter 2007-08, pp. 5.

The promise of the Semantic Web is to make machine understandable all the information available on the Web. The knowledge on any specific domain can be stored in an explicit and reusable format by means of ontology languages. Moreover, exploiting the formal semantics of ontology languages, implicit knowledge can be elicited through automated reasoning mechanisms.

Semantic Web technologies open new scenarios and suggest new approaches to classical problems. The envisaged applications are obvious in e-commerce, Web services, and peer-to-peer interaction, to mention a few. The formalization of machine-understandable annotations facilitates interoperability among heterogeneous resources, while avoiding usual drawbacks of unstructured data. Having an explicit semantics associated to queries and resource descriptions inherently allows a mechanized system to perform matchmaking and, subsequently, retrieval—based on the meaning of “what the resource is”—that is smarter than (whatever enhanced version of) pure text matching. We all know that the Web is an “open environment.” However, this is true not only for the technological infrastructure, but also for the information content. New information is continuously added to already existing resources, and old data are deleted. We may never own all the knowledge related to a resource; there can always be some pieces of missing, under-specified, information. Using classical data models like the ones behind modern databases, it is not possible to deal with such characteristics of an informational open environment. As a matter of fact, the huge research effort devoted to semi-structured data witnesses the existence of this problem from the point of view of database researchers. Unfortunately, databases always assume a closure of the data when answering queries. Semantic Web technologies are able to cope with informational openness by adopting the so-called open-world assumption (OWA). In other words, the system assumes that missing information can always be filled later on if needed, which is to say, a Semantic Web system does not treat the absence of a datum as evidence of absence and can distinguish it from negative information.

Regardless of the intuitive advantages introduced by Semantic Web, its expansion is still far from being a fact. There are several reasons for this situation:

1. the annotation effort is considerable (although promising results are being obtained on automated extraction and ontology mapping and merging);

2. the computational effort is often demanding, even for simple reasoning tasks; and

3. interaction with semantic-based systems is often cumbersome and requires skills that most end-users do not have—and are not willing to learn.

Semantic matchmaking can be defined as an information-retrieval task whereby queries and resources advertisements are expressed with reference to a shared specification of a conceptualization for the knowledge domain at hand, that is, an ontology. The results should be ordered lists of the resources descriptions that best fulfill the query, regardless of syntactic differences, and therefore of promising match candidates.

It is therefore intuitive that semantic matchmaking plays a relevant role in all scenarios where Semantic Web technologies can be fruitfully applied. Nevertheless, the effort of annotation should be rewarded with inferences smarter than purely deductive services, such as classification and logical compatibility. These inferences allow only for a coarse partition of candidate matches and show their limits in scenarios, as in e-commerce, where an approximate search is often useful and ranking methods are needed in order to retrieve most promising resources with respect to a user request.

This special issue present the best five papers—of 26 submissions—on semantic matchmaking and resource retrieval. Topics include formal reasoning for matchmaking, ontology matching, Web service discovery, matchmaking, and retrieval in e-commerce. The papers address some of the above-mentioned challenges from various perspectives.

“A Flexible Model for the Location of Services on the Web,” by Rubén Lara, Miguel Ángel Corella, and Pablo Castells, addresses the problem of semantic Web service discovery. Eschewing pure semantic approaches, they propose a framework for semi-automated service discovery wherein both structured and unstructured information is used to satisfy user requests. Another novel element is the use of two-phase matchmaking. The first phase uses a centralized matchmaker, in this case a UDDI service registry, while the second phase is performed on the consumer side, exploiting user personal information. The proposed approach is grounded in the emerging WSMO framework.

“DIANE: A Matchmaking-Centered Framework for Automated Service Discovery, Composition, Binding and Invocation on the Web,” by Ulrich Küster, Birgitta König-Ries, Michael Klein, and Mirco Stern, focuses on the automation of all tasks related to the use of a Web service. The authors propose DSD, a new service description language, and a matchmaking algorithm exploiting its capabilities. Particular emphasis is given to the asymmetry between the user request and the service description. While the request has to meet and represent user preferences, the service description has to be more general in order to represent a category of services (or resources). Furthermore, the language to describe the service is expressive enough to allow for the composition of more than one service in case a single service is not able to fit a request. Fuzzy sets are used to manage inexact matches.

What happens if resource descriptions refer to different schemas? Is it possible to provide one matchmaking process for a single request? A possible answer to this question is provided in “Ontology Mapping Between Heterogeneous Product Taxonomies in an Electronic Commerce Environment,” by Sangun Park and Wooju Kim. Their approach concentrates on taxonomies for e-commerce catalogs. User requests are automatically mapped in on-line shopping catalogs. The approach consists of three main steps. In the first step, WordNet is adopted to perform a word-sense disambiguation. The following two steps exploit the structure of the product taxonomies in terms of paths from the root product category and hierarchical relations within the taxonomies.

The strong relationship between the Semantic Web initiative and the realm of knowledge representation is witnessed by the use of description logics (DLs) as the formalism behind OWL (ontology Web language), the W3C official language proposed to build on and to reason with ontologies. Classical reasoning tasks in OWL (and then DLs), classification (subsumption), and logical compatibility (satisfiability) are based on OWA. Unfortunately, using approaches based on deduction in OWA has a twofold disadvantage. On the one hand, via subsumption and satisfiability, only whole match classes can be discovered, and it is very tricky to rank discovered resources within these classes. On the other hand, in a pure OWA approach, every bit of information missing in either the request or the resource description is considered to be underspecified. Possible ways to deal with these issues are proposed in the last two papers.

“Semantic Matchmaking of Web Resources with Local Closed-World Reasoning,” by Stephan Grimm and Pascal Hitzler, proposes the use of two non-monotonic extensions to DLs to overcome problems related to matchmaking in OWL under OWA. First, the authors show examples in which OWA yields counterintuitive results on matchmaking. Then they show how two closed-world extensions to DLs—namely, Autoepistemic DLs and DLs with Circumscription—can be applied to improve the matchmaking results.

In the last paper of this special issue, “A Nonmonotonic Approach to Semantic Matchmaking and Request Refinement in E-Marketplaces,” by Simona Colucci, Tommaso Di Noia, Agnese Pinto, Michele Ruta, Azzurra Ragone, and Eufemia Tinelli, the use of two other non-monotonic reasoning tasks, namely Concept Abduction and Concept Contraction, is proposed to match user requests and resource descriptions under OWA. The foundations of logical matchmaking in e-marketplaces are presented as well as the properties that ranking functions should have. The approach is initially presented in the framework of propositional logic and then is extended to DLs. The use of concept abduction for the refinement of user requests is outlined, and a tool exploiting the entire theoretical framework is presented.

We are confident that the papers selected for publication are representative of the state of the art for semantic matchmaking and resource retrieval, and we hope that they will contribute to a better understanding of the problems and possibilities related to this challenging research field.

All the papers were reviewed by two or three reviewers, either members of the Program Committee or external experts in the field. We are grateful to all the authors for their submissions and to the members of the Program Committee and external reviewers for their valuable work that helped make this special issue a valuable reference for both experienced researchers and interested newcomers. Special Issue Program Committee Members and Reviewers Sudhir Agarwal, Karlsruhe University, Germany Rama Akkiraju, IBM Research, United States Pietro Bonatti, University of Naples ‘Federico II,’ Italy Alex Borgida, Rutgers University, United States Andrea Calì, Free University of Bozen-Bolzano, Italy Fabio Casati, University of Trento, Italy Simona Colucci, Technical University of Bari, Italy Giuseppe De Giacomo, University of Rome “La Sapienza,” Italy John Domingue, The Open University, United Kingdom Stephan Grimm, Forschungszentrum Informatik, Germany Bin He, IBM Research, United States Tomasz Kaczmarek, Pozna´n University of Economics, Poland Rubén Lara, Tecnologá, Información y Finanzas, Spain Alain Leger, France Telecom, France David Martin, Artificial Intelligence Center, SRI International, United States Gregroris Mentzas, National Technical University of Athens, Greece Andrea Omicini, University of Bologna, Italy Massimo Paolucci, DoCoMo Labs, Germany Bijan Parsia, University of Manchester, United Kingdom Terry R. Payne, University of Southampton, United Kingdom Axel Polleres, University Rey Juan Carlos. Spain Riccardo Rosati, University of Rome “La Sapienza,” Italy Marie-Christine Rousset, University of Grenoble, France Ulrike Sattler, University of Manchester, United Kingdom Giovanni Semeraro, University of Bari, Italy Pavel Shvaiko, University of Trento, Italy Evren Sirin, University of Maryland, United States Umberto Straccia, ISTI-CNR, Italy Evgeni Zolin, University of Manchester, United Kingdom.

Eugenio Di Sciascio ( has a laurea degree with honors from the University of Bari and a Ph.D. from Politecnico di Bari (Technical University of Bari), where he is full professor of information technology engineering and chairs the Unified Council on Information Engineering. Formerly, he was been an assistant professor at the University of Lecce and associate professor at the Technical University of Bari. He is the scientific coordinator of SisInfLab, the Information Systems Laboratory of the Technical University of Bari. His current research interests range from knowledge representation to include theoretical and practical aspects of e-commerce, mobile and ubiquitous applications, Web services discovery and composition, multimedia information retrieval, and knowledge management. He is involved in several national and European research projects related to his research interests and has co-authored papers that received awards at the ICEC’04, IEEE CEC EEE ’06, and ICEC’07 conferences.

Francesco M. Donini ( is a full professor at the Università della Tuscia–Viterbo, Italy. Formerly, he was an assistant professor at the Università La Sapienza, Rome, and associate professor at Politecnico di Bari. His Ph.D. thesis was on description logics; he has subsequently worked on many aspects of knowledge representation in artificial intelligence, both on theoretical issues and practical applications. His research interests range from description logics to non-monotonic reasoning, abduction, algorithms, and complexity of reasoning in KR formalisms to their application in a variety of practical contexts. He co-authored papers that won awards at the IJCAI-1991, ICEC-2004, and IEEE CEC EEE -2006 conferences.

Tommaso Di Noia ( is an assistant professor of information technology engineering at the Technical University of Bari (Politecnico di Bari), Italy. He earned his Ph.D. at the Technical University of Bari. His main scientific interests include description logics (theoretical and practical aspects); resource matchmaking; knowledge representation systems for electronic commerce; automatic Web services discovery and composition; knowledge representation systems and applications for the Semantic Web. He co-authored papers that received best paper awards at the ICEC-2004 and IEEE CEC EEE -2006 conferences, co-chaired workshops on semantic matchmaking at SMR 2006 and SMR2 2007, and has been a program committee member of several international conferences and workshops in areas related to his research interests