Survey

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 3

A Semantic Similarity Measure Between Web Services Based on Google Distance - Huirong YANG, Pengbin FU, Baocai YIN,

Mengduo MA,Yanyan TANG 2011 IEEE The emergence of semantic web services has enticed researchers to discover how to search the service precisely by accurately calculating the matching degree of web services. In the semantic matching of web services, traditional methods calculate the matching degree between input and output parameters by judging the semantic similarity between ontology concepts related with input and output parameters, thus obtain the semantic similarity measure between web services. There are several semantic matching algorithms developed in recent years, they mainly focus on calculating the semantic similarity between concepts within ontology domain. Such as Massimo Paoluccis algorithm in[1], they propose a matching algorithm of semantic web services based on the filiatory order relationship between the requests and the registered services. The results of semantic matching degree can be categorized as four levels: exact match, plugin match, subsumes match and fail match. This algorithm has always been quoted and laid a foundation for the comparison between different algorithms. However, the limitation of this algorithm is the lack of quantitative measures for specifying the extent of similarity, and it cannot be applied to comparison of large amounts of web services. Peng and Shi [2] define a stage function to calculate the similarity of two concepts in the same field of ontology and compare their tests results with Massimo Paoluccis. They define the exact match as 1, fail match as 0, plug-in match as real numbers between 0.5 and 1, subsume match as real numbers between 0 and 0.5. This method takes 0.5 as the threshold between plug-in match and subsumes match, which is still unable to describe the similarity between concepts precisely. Wu [3] use semantic similarity between words to detect the similarity between web services, and provide a method to calculate the semantic similarity between words. While this method doesnt take the principle that similarity of plug-in match should be greater than that of subsumes match proposed by Massimo Paolucci [1]. The common feature of the achievements above is that they make use of taxonomy tree to calculate the semantic similarity between advertised and requested services. As for the problems with limited algorithm accuracy, taking filiatory order relationship, depth and local density into consideration may improve the performance of algorithm to some degree. We developed similar improved algorithm [4]. The test results show such algorithm does enhance the recall and precision of the service while increase the complexity of calculation. While in most cases, the requesters may come from an amateur, and it is difficult for him/her to provide professional ontology to support the search of services. Thus the efficiency and accessibility of such algorithm is greatly limited, even if we make use of ontology domain like WordNet or HowNet. With a slower speed in updating content compared with the rapid development of internet. Therefore, if we apply such algorithm into the search of increasingly developed web services on the internet, the accuracy and precision will be discounted. Taking all these factors above into consideration and based on the lasted development of semantic web services, this paper propose a novel method of calculating similarities of web service A Semantic Matching Algorithm of Web Services Based on Google Semantic Distance. The algorithm uses the Google massive terms and open Google search engine to calculate the Normalized Google Distance (NGD) of web services input and output concepts, and then obtains the approximate degree between web services. Sheth et al. distinguish four different kinds of semantics for Web services [87, 209]: Data semantics formally define data in input and output messages of Web services. Functional semantics formally define the capabilities of Web services, i.e., by defining preconditions and effects and semantically annotating interfaces and operations. Non-functional semantics reference QoS and general policy requirements/constraints. Execution semantics describe the execution of services and operations. In order to define the meaning of distinct service components by semantic annotations or enhancements of a service description, it is necessary to have a domain model which can be used as a knowledge base. Most probably, the bestknown knowledge base format are ontologies. However, there are other formats, too: However, the following quote by Gruber is widely accepted as a common definition of an ontology from a more technical view [94]: An ontology is a formal, explicit specification of a shared conceptualisation. A more comprehensive explanation of the intended purpose of an ontology in computer science is provided by Lacy Computer science ontologies serve a similar function as database schemas by providing machine processable semantics of information sources through collections of terms and relationships. The semantics support a shared and common understanding of a domain that can be communicated between people and software. Matchmaking With regard to (semantic) Web services, matchmaking is the process of finding suitable service offers to fulfill a given service request [61]. Accordingly, a matching engine is a tool which implements such a process.

Web Service Discovery Research: A Study of Existing Approaches


Vaneet Sharma1, Mukesh Kumar2

Fig.2. Taxonomy of Web service discovery systems

B. Semantic Based Systems 1)Ontology based approach Recent researchers have focused on performing semantic matching to enhance the accuracy of Web service discovery. The predominant problem in UDDI matching is the restrictions imposed by keyword matching that do not allow retrieval of Web services with similar functionality. For example, two WSDL descriptions can be used to describe the same service but with different words such as car and automobile. Without a search that supports semantic matching, both of these Web services will not be returned when only one of the terms is used in Web service discovery. The semantic matching can be performed by exploiting the semantic representation of concepts and their relations in Web. A very important direction of the current Web service research is the semantic Web services group, such as OWLS and WSDL-S [11].Semantic descriptions of Web services can be obtained with the use of OWL-S languages. Most of the efforts presume prebuilt ontology. However, constructing ontology as a semantic backbone for a large number Web services is really not easy. Firstly, there is difficulty of the construction of a commitment to a common ontology. Even if the consensus of a common ontology is achieved, it may not be able to catch the fast pace of change of the targeted Web service or the change of users vocabularies in their applications. Secondly, using ontology for manual annotation requires that the annotator have some skills in ontology engineering which is a quite high requirement for

normal consumers. Actually, to the best of our knowledge, there are very few real-world Web Services described by OWL-S. 2)Concept Based Approach Another important work worth mentioning is Concept based Web service discovery. This approach employs an unsupervised approach to retrieve the underlying semantics from WSDL itself and measure the similarity between operations and input/output parameters.

You might also like