Web Services Discovery Research Based On Index

Web services, as an effective realization technology of SOA, have developed largely. Comparing with the traditional web pages, the information island problem existing in web services field is more serious. On the Internet, web services with specific functions not only submerge in the web services library, but also submerge in the traditional web pages library. As a result, finding web service rapidly and accurately becomes a problem needing solving, and this promots the research on the field of web services discovery.Basing on national 973 project "The Research on the Basis of Trusted Network and Pervasive services——Research on Theory, Model and Key Technology for Pervasive Service ", we take the Service Oriented Architecture as baisis, and have achieved some results. The work in the paper focuses on the following aspects:1. Selecting a proper web services formalizing model is the basis of web services discovery. Comparing with the initial web services description files, the current web services formalizing models exist infmation losing or information imposing. The information losing displays that lots useful information is missing when formalizing from the initial description files; and this apprears mainly in the web services discovery methods basing on traditional keywords matching. The information losing leads to the low accuracy in the matching process for lack of matching basis in the web services discovery methods. The information imposing displays that lots of useful information in the formalizing model can not obtain from the basic web service description files, or is very difficult to obtain from the basic web service description files. This apprears mainly in the web services discovery methods basing on sematics ontology matching. The information imposing leads to basis missing in the matching process of web service discovery methods when we apply these discovery methods on the Internet. In this paper, we propose a web services formalizing model. This model bases on the basic information of web service (compostion), such as wsdl file, bpel files and so on. Comparing with the traditional keywords matching formalizing models, our model contains more rich semantics information. Comparing with the sematics ontology matching formalizing models, all the information in our model can be extracted from the basic description files; this makes the web services discovery methods based on our formalizing model can better handle the web services on Internet, and thereby these methods can be easy deployed on the Internet.2. The accurancy of web services matching affects heavily on the semantics ontology. The current web services matching methods with high accurancy mainly depend on semantics ontology library; they realize the accurate matching based on descriptive information expressed by ontology. However, there is no authoritative ontology library on cuurent internet, and most ontology libraries are created by the provider of web services discovery system; this leads to difficult to achieve compatibility in different ontology library, which makes the criterions missing in web services matching basing on semantics ontology. In this paper, we propose a web services matching engine with high accurency, and there is no semantics ontology dependence in our matching engine. When matching in natural langurage description, we apply synonyms expansion and ontology expansion (optional); whether applying the ontology expansion dose not play a significant influence in our matching engine. Most effective information for web services contains in various web services naming, which are combinations of several semantics terms. When mathing these namings, under the supervision of concurrent probability, we apply synonyms expansion, abbreviations expansion, ontology expansion (optional) and semantics unit induction; this helps our matching engine act high accurance in the namings matching. The matching engine performs high precision rate and recall rate, while it gets rid of the dependence of semantics ontology.3. The web services on the Internet perform different performance. Some web services entity can not consistent with their advertisements; some web services perform a long responding time; some web services provide unfriendly interface; however, all these web services provide similar description files, and this leads to difficult to distinguish these web services. In this paper, basing on the tools and theories of traditional web pages, and combining with the characteristics of web service itself, we propose a web services performance evaluation method with high feasibility and low cost. We evaluate web services from information providing aspect and function sharing aspect to implements an effective evaluation, and the result shows that our method can improve the precision rate of web services discovery system significantly.4. The searching accurancy and the searching processing time are two most criterions in web services discovery field. However, there is no research result which shows high performance in the both evaluation criterions. The low searching processing time is always in web services discovery methods based on traditional keywords matching framework, while these methods show low precision rate and recall rate. When matching the interface naming, the traditional keywords matching acts inefficiently, especially when matching namings with different forms but expressing same semantics. The high searching accurance is always in web services discovery methods based on semantics ontology matching framework, while these methods cost lots searching processing time. To describe web service more detailed, they apply many attribute for each naming, and this increases the searching space; besides that, when matching users request, these methods always apply ontology reseaning machine, which increases the time complexity to O(n2) (where n is the attributes number in web services library). In this paper, we propose a web services discovery framework basing index library. The index library helps providing a low processing time in request searching. When creating the index library, we apply semantics mining to overcome the shortcoming of low searching accurancy in traditional index framework; we import web services compostions searching to impove the recall rate in our framework; and we apply a matching result evaluation process to help improving the precision rate. When processing users'web services discovery request, our framework shows high precision rate and recall rate. This framework provides a solution to the problem of persuiting both low searching processing time and high search accurancy in web service discovery field.