1. Introduction
The development of various ubiquitous computing service technologies with a new technological system which enables “the computing environment any time, any place and for anything” in which human centered interface application technologies are added on top of conventional web service technology is progressing. This research is aimed at establishing the fundamentals to implement in the future a Ubiquitous Computing Architecture by developing an intelligent algorithm which can integratedly manage and connectively operate individual applications which are composed of diverse platforms and components, in accordance with semantic web standards and web service standards, using intelligent web service technology. A Ubiquitous Computing Services System was also developed on the basis of a differentiated technology to complement the problem that most web service related research in the past was insufficient in suggesting the general and overall direction as it tended to focus on methodological aspects.
As potential breakthrough solutions to the problems of the current system we have InfoSleuth [
1], which has adopted the brokering method and Larks [
2], applying the matchmaking method. InfoSleuth [
1], which uses broker agents, is an agent based information searching system where the broker takes charge of both discovering and relaying services for efficient web service searching. As a downside, bottlenecks can occur during data transmission since the system is centered on the broker. On the other hand, Larks [
2], which is based on the dynamic matchmaking method developed by Carnegie Mellon University's Robotics Institute, uses LARKS language for service advertisements and requests to enable more flexible and efficient matching, but no mention is made of the QoS functions.
The proposed system solves the problematic point of existing UDDI by providing an Extended UDDI Search Module and can also provide a more exact examination of users' demands and reflect the demands in web searches by applying a new QoS Based Matchmaking Algorithm to perform quality evaluations. The proposed system also makes possible provision of a more credible service to users by reflecting the private preferences of respective service demanders through a Rule Based Search in which the services results generated from the Matchmaking engine and the users' preferences are reflected. This system is different from the existing Semantic Web Services Systems in the following aspects:
First, the meaning of the search keyword(s) used within the domain is defined accurately by constructing an ontology server. Additionally, based on the constructed ontology, similar words are searched for and interpreted.
Second, an extended UDDI search module is provided in order to solve problems with the existing UDDI. For this purpose, a matchmaking engine using the QoS based matchmaking algorithm was designed in this research.
Third, rule-based searching is executed in order to reflect the users' preferences. Rule-based searching is a method for converting rules in the user profile registry and each returned Web Services result is mapped into Jess script through XSLT.
Fourth, ranking is executed for more accurate information searches. The newly developed ranking algorithm was applied to rank search results and provide more accurate and reliable information.
Finally, Intelligent Web Services Integrated System was developed along with a step-by-step design method for efficient Web Services search and composition. Its efficiency and accuracy are verified by comparing it with the existing systems.
This paper is organized in the following order. In section 2 there will be an overview of the Semantic Web Services and Ubiquitous Computing Services. This is also analyzed and compared with current systems. In Section 3, the notion of an Intelligent Web Services Algorithm is suggested and in Section 4, the architecture and implementation of an improved Ubiquitous Computing Services System is explained, along with the functions, characteristics of its modules and execution results. It is also compared with current systems. Conclusions are provided in the final section, along with plans for further studies.
3. Intelligent Web Services Algorithm
Matchmaking is a process of finding a service provider that satisfies the service requester's requests [
7]. Matchmaking is executed based on whether the web service request and web service advertisement match or not [
7]. The match between requests and advertisements is determined based on whether the service input and output among the functional description match or not. The matchmaking system must support input and output through the repository and enable service browsing, correction and cancellation [
7]. Ultimately the exact correspondence between service request and service advertised and the reflection on QoS elements should be achieved for an efficient Semantic Web Services search. Though some matchmaking systems were proposed in past research, they have some problematic points like those mentioned below.
First, the past research conducted matching by stressing certain partial elements as they failed to treat the identity elements in web services comprehensively. For [
4], which conducts a matchmaking algorithm based on DAML-S, an exact web service search is unavailable as it performs web service matching using only input and output information. Second, the QoS matching mechanisms used in the past research have the problem that even high total QoS marks cannot guarantee providing good service in response to demands, as one particular QoS element decides the entire similarity. For example, a service for which the reliability or accessibility is very low but the price is very high could get high marks and selected as the result service in preference to other services. Conclusively, this research proposes solutions for the problematic points of the matchmaking algorithms suggested in the prior research. Integrated Matching Algorithm procedures are as shown in
Figure 1 below.
First the web service identity elements were classified by item and algorithms for the respective items were proposed and next, the integrated matching algorithm was described. That is, this research classified the web service identity elements like in
Table 2 for efficient matching.
Second, the reference values of the respective QoS elements were calculated after the respective QoS elements values were adjusted. Afterwards, all the QoS marks were counted and put into the Integrated Matching Algorithm for application. The algorithm measuring method for the respective identity elements are as shown in
Table 3.
The respective identity elements` values were standardized and formalized to a value between 0-1. The entire similarity value which is composed of the sum of the respective identity elements is like one given by the formula
(1) below:
The respective correspondence elements were standardized and normalized to a value between 0 to 1. The entire similarity value resulting from the sum of the respective correspondence elements can be expressed as formula
(1) and the detailed description of the algorithms of the mentioned definitions are as given below.
3.4 QoS Measurement Algorithm
For more accurate and reliable web service results, the provided matching method must reflect QoS factors such as time, price, reliability and performance. The current systems such as Larks [
2] or Infosleuth [
1] do not apply these factors in their matchmaking methods. Though METEROS [
3] developed by the University of Georgia does reflect QoS factors in matchmaking, no adjustments are made with regards to each individual QoS factor. Instead, the overall QoS similarity is determined by only one of the factors. Thus, very relevant and meaningful web services could be evaluated lower than their actual ranking.
For such reasons, this study classifies quality evaluation factors and suggests separate algorithms for each factor. After applying algorithms to each QoS measurement factor value, the adjusted value and standard value are calculated. Then, the QoS score is calculated by multiplying all the QoS measurement factors. The QoS measurement procedures are as shown in
Figure 2 below.
The QoS measurement algorithm based on the sum of standard values (Definition 4) of each QoS factor is as indicated below.
■ Definition 4 |
The QoS measurement algorithm, the multiplication of the standard values of each QoS factor including Availability (Au), Response Time (Tr), Throughput (Ths), Reliability (Rs), Accessibility(Rs) and price (Ps) is as indicated below. |
|
4. Ubiquitous Computing Services System
As a solution to the problems of the current Web Services, the present study suggests Ubiquitous Computing Services System based on a Matchmaking Engine and a Rule-based Search Engine. Section 4 will describe the overall structure of the suggested system, the functions of each module and the evaluation of its system.
4.1 Scenario
John who is driving in a car, wants to find the most famous pasta restaurant in Manhattan. As the precondition, the restaurant should be located within 10 miles from his current location. With this situation and conditions, the analysis on the user should be conducted through the user information profiles in the user information registry first to satisfy the user's requirement. By analysis of the user information profiles using the privatized agent, the information that John ordinarily orders meals at restaurants within a 60$∼100$ price range and prefers restaurants which provide various services relative to the food price is gathered. Then, the web service system finds the restaurant best fitting these criteria and provides the information to John by comparing the fact that John prefers restaurants in the 60$∼100$ price range providing diverse services relative to the price, which is found out through the user profiles, and the service search result analyzed by the Matchmaking Engine. The web service also analyzes the location of John and guides him to a restaurant which is within the radius of 10 miles from his location. As above, it is necessary to find out the information which best fits the request of the service requester by analysis on the service search results and the user information profiles. For this, the technology to determine the user's preferences using a rule based search engine is needed and the service for context awareness and location awareness on the user should be provided simultaneously.
4.2 System Architecture
Matchmaking describes the processes by which parties that are interested in participating in an exchange of economic value are put in contact with potential counterparts [
5].
Figure 3 illustrates the process of returning optimal results to a service requester using a Matchmaking and Rule-based Search Engines.
- Step 1:
First, the requester submits a request to the Search Manager.
- Step 2:
The Search Manager determines the current context and location information of the user using the Context Container and request answers from the Advertisement Database and UDDI Registry on the basis of the results. The Context Container provides the information on the context or the environment of a service requester, that is, the context based privatized information. Finally, the Search Manager retrieves all the advertisements from the Advertisement Database and UDDI Registry.
- Step 3:
Search Manager sends it to DAML-S/UDDI Translator that constructs a UDDI service description. Then Reasoner computes the match level and ranks the results.
- Step 4:
Search Manager sends results to the Web Service QoS checker using the QoS measurement algorithm to determine if it is verified or not.
- Step 5:
the system conducts a rule based search which compares the service result produced by the Matchmaking Engine and the user rules taken from the user profile registry and thereupon selects the appropriate information for service.
- Step 6:
Finally, Search Manager sends the final results to the requester.
4.3 Functions of Each Module
4.3.1 Matchmaking Engine
Figure 4 illustrates the system structure of the Matchmaking Engine. DAML-S/UDDI Translator stores DAML-S Profile expression in UDDI Registry.
Web service requesters use the UDDI request Port to execute the questionnaire sequence which the UDDI registry provides. However this questionnaire mechanism has the fault that it can not execute the semantic information of web service advertisements and the capability description which are provided by DAML-S Profile. DAML-S/UDDI Translator solves the above problem by executing the extension of UDDI Registry through the mapping of UDDI Registry and DAML-S Profile. In DAML-S/UDDI Translator not only the input and output functions for web services are implemented, but also the web service search which uses the function of pre-condition and post-effect is available, as it supports capability functionality extensively on top of the existing UDDI method. The Reasoner takes the role of efficiently matching web service requests and web service advertisements using the Matchmaking Algorithm (Chapter 2) and provides prioritization.
4.3.2 Rule Based Search Module
The module executes rule based searches to compare produced results with the user's rules drawn from the user profile registry and selects only the appropriate service information. In this research, SweetJess [
6] is used for executing rule based searches. SweetJess executes reasoning after expressing rules with DAML+OIL and converting them into JESS's rule language. With SweetJess, the rules indicated in the user information profile and the returned web service results are converted to Jess script through XSLT. Here the Jess engine [
10] uses XSLT to convert rules expressed by RuleML to a format that can be used by the existing automatic reasoner and then draws the results through forward inference [
6].
Figure 5 is the system architecture of the Rule Based Search Module based applied methodologies.
4.4 Implementation Method
The intelligent web service integrated system implemented in this study is a system that fetches the information based on keywords, keyword explanations, price keywords, QoS information (network speed, credibility, availability, accessibility) and enterprise related items from the databases of the four major domestic portal sites (Naver, Yahoo Korea, Daum, Empas) and displays the information.
This system includes in its internal structure all the items suggested in this study, that is, the Match-Making Algorithm, the Korean character (Hangeul) using search engine and the Ontology Server. With this implementation method, the real data which were being provided at the major portal sites were fetched and the information from the data were processed to fit the intelligent web service concept. In this system, a ‘server program’ which can contact the portal sites of Overture, Daum, Naver, Yahoo Korea and Empas and fetch keywords, price information and QoS information from the sites, is operating continually and this server contacts web pages accompanying the search engine programs and fetches web pages. Subsequently the system analyzes the fetched web pages and stores the analysis result in the database. In the process of web page analysis the system only takes highly related information using the Extraction Agent suggested in this study.
At this time the system utilizes ‘analysis rules’ according to the respective portals. The ‘analysis rules’ according to the respective portal sites are made to be modified in correspondence with the portal sites as the portal sites tend to modify the web pages of search results. Finally the system carries out analysis on the web pages by applying the the Ranking Measuring Algorithm and the Match-making Algorithm, and subsequently displays the analysis results.
Figure 6 shows the implementation method of the system.
4.5 Execution Results
Figure 7 shows the service search screen illustrating the system in which web service site information, keyword price information and advertisement sponsor information can be searched, for matchmaking between the service requester and the service provider.
The information in the search here are ones taken from the real data of four internet portal sites in Korea (Empas, Yahoo Korea, Naver, Daum) and processed, mainly on the basis of keywords, price information, demand frequencies and QoS information.
Figure 8 is the result screen obtained from the Ubiquitous Computing Services System after the user input the search keywords and sent the query. Having applied the Rule based Search Algorithm and the QoS Matchmaking Algorithm, the result differentiates from that of the current system.
4.6 An Experimental Evaluation
In this research the experiment was conducted after credible statistics were secured by pre-operation for 7 months. For the analysis on the performance of proposed system, the real data of NateOn, an ordinary internet portal site was taken and used for the comparison. The data described around NateOn is the result directly taken from the site, not via the server program of the proposed system. As experimental measured items, the service explanation information, input and output information, quality consistency rate for the collection of the keywords which users had used for searching, from February 26th, 2006 to Aug 12th, 2006 was used as standard, and the backup data was made into a statistical numeric by a server program developed with UNIX as standardized test marks do not exist as of now. The result of the experiment was as shown in
Table 4 and
Figure 9, below.
From the experimental result, it is found that the similarity marks of the “Matchmaking Factor” of the portal sites, Empas, Yahoo Korea, Naver, Daum, which is proposed in this research, is enhanced higher compared to the similarity marks of other sites.
5. Conclusion and Future Studies
“Ubiquitous” enables a computing environment which is accessible from any place, at any time while it also enables capturing of the user's location and provision of an optimal service. The ubiquitous computing services system suggested in this paper, in which “context awareness” and “location awareness” is available, enables searching and treating of information that users want, not being bound by the limitations of physical space. This paper also applied semantic web technology to web services and enabled the web to perform ubiquitous computing service composition and execution. The proposed system makes efficient matching between service requesters and providers possible through a system based on QoS. The proposed system also makes examining users' request and reflecting it to the service possible by the Rule Based Search Engine in which users' preferences are reflected. Finally, a quantitative experimental evaluation measuring the rate that service features such as information, input and output information, service function expression information and quality information corresponded to the users' demands based on the keyword collection selected by users, was conducted using a server program which was developed by us. In the result it was identified that the similarity marks of the proposed system was improved by 15-20% compared to that of ordinary internet sites. As a future research task, the adding of screen interface which can mirror demanding items by users and limitation conditions in more detailed divisions and the design of a methodology which can examine the efficiency and exactness of the proposed system objectively is needed.