TW200844762A - A possibilistic petri-nets based method for web service matchmaking - Google Patents

A possibilistic petri-nets based method for web service matchmaking Download PDF

Info

Publication number
TW200844762A
TW200844762A TW96116402A TW96116402A TW200844762A TW 200844762 A TW200844762 A TW 200844762A TW 96116402 A TW96116402 A TW 96116402A TW 96116402 A TW96116402 A TW 96116402A TW 200844762 A TW200844762 A TW 200844762A
Authority
TW
Taiwan
Prior art keywords
service
model
unit
confidence
item
Prior art date
Application number
TW96116402A
Other languages
Chinese (zh)
Inventor
Jonathan Lee
Yao-Chiang Wang
Shang-Pin Ma
Jie Lee
Wen-Ting Lee
Mu Chun Su
Original Assignee
Univ Nat Central
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Univ Nat Central filed Critical Univ Nat Central
Priority to TW96116402A priority Critical patent/TW200844762A/en
Publication of TW200844762A publication Critical patent/TW200844762A/en

Links

Landscapes

  • Information Transfer Between Computers (AREA)

Abstract

In this patent, we devise a Web service matchmaking method based on the WSDL and PPN technologies. This method is realized by two main transform procedures and a calculation formula to get the matchmaking result for service matchmaking or service recommendation. A transformation procedure transforms a Web service represented by a service description written in WSDL format into a PPN model unit to capture important characteristics. Meanwhile, another one PPN model unit captures service requestors' desired needs. Two model units are combined into a bigger one to form a service matchmaking model with some extra confidence transformation units between them. These units are constructed through a confidence transformation procedure that leverages a primate data type mapping table. Finally, a matchmaking calculation formula utilizes possibilistic reasoning to get the similarity between service request and matched Web service which can also be useful for Web service recommendation.

Description

200844762 九、發明說明: 【發明所屬之技術領域】 網際方法;特別是’適用於 寸色與可能性派翠網路的推論機制上,透 結果計算方式,將服務需求與服務描述 ;型與提供圖形化表示,最後根據推 進仃服務晴相似度的計算,並兼具服務推薦的功能。 【先前技術】 服務導向架構(S〇A: Service 〇riented200844762 IX. Description of invention: [Technical field to which the invention belongs] Internet method; in particular, 'inference mechanism applicable to the color and possibility of the Pai network, through the results calculation method, the service demand and service description; type and provision Graphical representation, and finally based on the calculation of the similarity of the promotion service, and the function of service recommendation. [Prior Art] Service Oriented Architecture (S〇A: Service 〇riented

Architecture)提出以來,已經廣為π領域所接受,其相 關標準的制定使得網際服務(Webservice)漸漸在重^的 應用領域上,扮演核心的元件。這些熟知的標準包含·簡 單物件存取協定(S0AP : Simple Object Access Protocol)、網際服務描述語言(WSDL: Web ServiceSince its introduction, Architecture has been widely accepted by the π field, and its related standards have made Internet services (Webservice) gradually play a core component in the application field. These well-known standards include the Simple Object Access Protocol (S0AP) and the Internet Service Description Language (WSDL: Web Service).

Description Language )、以及 UDDI ( UniversalDescription Language ), and UDDI ( Universal

Description,Discovery and Integration)服務註冊機。 (service provider) | (implement) f部屬(deploy)網際服務實體,並透過網際服務描述語 言來描述該網際服務的介面規格(interface specification),並將之註冊(register)或公佈(publish) 在UDDI服務註冊機上。服務需求者(service request〇r) 則根據服務註,機所提供的搜尋功能,找尋所需要的網際 服務,並透過簡單物件存取協定與所需要的網際服務實體 進行溝通,完成服務遞送。 雖然有許多相關的研究努力在制定更完善的服務描述 7 200844762 語言(service description language),例如網際服務 知識本體(〇WL-S: Web Service Ontology f〇r feb Service)、語意化網際服務描述語言(WSDL-S: Semantic Web Service Description Language)、可能性派翠網路 代理人描述語言(PPM-ASDL: PPN-based Agent Service Description Language)或其他加入服務品質(QuaHty 〇f Web Service)資訊的服務描述語言等,其對應的服務配對 方法都要建立在這些複雜的描述語言上。例如,針對不同 的服務領域,需要有完善的知識本體(ontology)建置,以 提供服務配對所需的語意分析,來使得其配對結果更準 ,。$正因為這些描述語言是在網際服務實做(implement) 完成後才另外由服務提供者於註冊服務時手動增加描述, 造成服務提供者的不便利性,因而使這些服務描述語言不 如網際服務描述語言(WSDL)來的普及與流行。除此之外, 一些致力於發展網際服務相關技術的第三組織,如開發 Tomcat 應用程式伺服器(Applicati〇n Server)的Description, Discovery and Integration) service registration machine. (service provider) | (implement) fDepartment (deploy) Internet service entity, and describe the interface specification of the Internet service through the Internet Service Description Language, and register or publish (publish) the UDDI The service is registered on the machine. The service requester (service request〇r) finds the required internet service according to the service function provided by the machine, and communicates with the required internet service entity through the simple object access protocol to complete the service delivery. Although there are many related research efforts in the development of a more complete service description 7 200844762 language (service description language), such as Internet Service Ontology (〇WL-S: Web Service Ontology f〇r feb Service), semantic Internet Service Description Language (WSDL-S: Semantic Web Service Description Language), PPM-ASDL (PPN-based Agent Service Description Language) or other services that include Qualified Service Quality (QuaHty 〇f Web Service) information Descriptive languages, etc., the corresponding service pairing methods must be built on these complex description languages. For example, for different service areas, a complete knowledge ontology needs to be built to provide semantic analysis for service pairing to make the matching results more accurate. $Because these description languages are manually added by the service provider when registering the service after the Internet service implementation is completed, causing the service provider's inconvenience, thus making these service description languages inferior to the Internet service description. The popularity and popularity of language (WSDL). In addition, some third organizations that are committed to developing technologies related to Internet services, such as the development of the Tomcat application server (Applicati〇n Server)

Apache.org,其所提供的AXIS網際服務實做平台(目前最 普及且免費的),就提供許多的免費工具,如java2wsdl 等,方便程式設計師將所撰寫的應用程式自動轉換為網際 ,務以及所需的服務描述,也使得網際服務描述語言仍是 ^重服務描述語言。因此,如何(只)應用網際服務 犏述浯&的特色來達成服務配對的功能,就成了本發明的 重要精神。 本發明借用了可能性派翠網路(PPN: possibilistic Petri Nets )模型的可能性推論(possibiiisHc reasoning)機制以及圖形化(visualizati〇n)的表達方 ^來建,服務配對的理論模型。這使得本發明的配對機 I理淪依據,且服務配對的過程可以透過圖形化的方 式來表達,it是其他服務崎機制所欠缺的。本發明係透 8 200844762 過一系列的轉換程序(transfer procedure),來將服務描 述語言(WSDL)的元件轉換成可能性派翠網路模型單元,再 與另一個代表服務需求的模型單元進行服務配對的機制, 由可能性推論的結果,來整合整個服務配對的結果。 最後本發明的配對機制亦可以用於服務推薦上, 將-個服務描賴言當作服務需求,來找尋她的網際服 目 =纽務轉线所欠缺的雙錢制(服務 配對與服務推薦)。 【發明内容】 本;發、明設計一系列的轉換程序(transfer procedure),來將服務描述語言的元件轉換成可能性派翠 元,例如介面型態(P〇rtTyPe)轉換成階級式 可月b性派翠網路(HPPN: Hierarchical PPN)單元,服務 ,作(operation)及其對應的輸入資料(inputdata)盥 輸出貢料(output data)轉換成多個輸入元(Ip: input places)及其對應的前置條件(pre_c〇nditi〇ns)、一個信 心轉換(con f 1 deuce trans i t i on )單元、和多個輸出元(0P: output Places )及其對應的後置條件 conditions)。接著只要將服務需求的可能性派翠 、、罔,杈組单兀與之整合(透過額外的模型單元),即可進行 〒推並透過圖形化的表*方式,暸解服務配對的 又過程,最後藉由整合相關的推論結果,來得到服務 f依照可能性派翠網路的推論機制,其推論的 =在於相關信心轉換(⑺nfident transiti〇n)單元的 連及信心程度㈣的演算法,並透過⑼單元來表 過程的不確定,亦即配對中的網際服務能滿足 的服^求的不確定性,進而表達服務敲的程度。 ”、、了建立兩個模型的初始連結狀態以利可能性推論 9 200844762 提/服務f求模型,另—個表示服務 ί=亍推論:以求得在不二心』Ιίίΐ關 1 ,的輸入資料與服務需求所能提供的不一致(輸 服務提供所產出的結果與服務需求所期待的不一致) ίΐ认,際ί務能滿足使用者需求的最大可能性為多 二^貝料型g的不—致性主宰了服務配對過程中的所 不確疋性,本發明亦設計透過額外的基 咖vedata-type 卿Plngtable),來 杈1L中所有的信心程度單元的步驟。透過不斷的修正對應 表中的信心程度,將可以使得服務配對的結果更加準確了 本發明透過上述相同的機制,使服務配對技術亦可以 用於服務推薦系統上。由於可能性派翠網路模型單元可以 同時用來表達所需要尋找的服務特徵或所提供的服務特 徵丄因此本發明可以用相同的方式來將某個已知的網際服 務實體的服務描述語言轉換成一個服務需求模型單元y而 後來的步驟就如同前例所訴:丨·根據轉換程序,建立另一 個網際服務的服務提供模型單元;2·整合兩個模型單元· 3 進行可能性推論;4·取得配對結果。最終即可以此結果來· 表示兩個網際服務實體之間的相似程度,而做為推薦服務 的依據,例如當相似程度大於〇· 8時,即推薦該網際服務 給予服務需求者。 依照本發明的較佳實施例所述(請參考圖1),一個網 際服務(Web service)實體102可以有一個以上(通常只奋 產生一個)的服務描述文件104來描述其服務的特徵,而^ 文件係依據網際服務描述語言(WSDL: Web Sendee Description Language)規格所撰寫,並註冊(registei^ 200844762 在另一服務註冊機上,例如標準的UDDI服務註冊機106, 服務的需求者可透過其提供的服務介面,來搜尋所需要的 網際服務’並取得其詳細的服務描述,例如―虬服務描述 文件。 _依照本發明的較佳實施例所述,上述網際服務描述語 二(WSDL)為W3C組織所制定,用來描述網際服務的規格, 分為服務介面資訊(如何與該服務溝通)與服務實做資訊 (該服務的網路實體位址)。Apache.org, which provides the AXIS Internet service implementation platform (currently the most popular and free), provides many free tools, such as java2wsdl, which allows programmers to automatically convert their written applications into the Internet. As well as the required service description, the Internet Service Description Language is still a service description language. Therefore, it is an important spirit of the present invention to apply (only) the function of the Internet service to describe the features of the service pairing. The invention borrows the possibility of the Possibilistic Petri Nets (PPN: possibiiisHc reasoning) mechanism and the graphical representation (visualizati〇n) expression to construct a theoretical model of service pairing. This allows the pairing machine of the present invention to be relied upon, and the process of service pairing can be expressed in a graphical manner, which is lacking by other service mechanisms. The present invention is through a series of transfer procedures to convert a service description language (WSDL) component into a possibility to send a network model unit, and then to another model unit representing a service requirement. The mechanism of pairing, the result of the possibility of inference, to integrate the results of the entire service pairing. Finally, the matching mechanism of the present invention can also be used for service recommendation, and the service description is regarded as a service demand, to find her internet service = the double money system lacking in the new service transfer line (service matching and service recommendation) ). [Summary of the Invention] This is a series of transfer procedures designed to convert the components of the service description language into possibilities, such as the interface type (P〇rtTyPe) into a class type. The HPPN (Hierarchical PPN) unit, service, operation and its corresponding input data (output data) are converted into input elements (Ip: input places) and Its corresponding precondition (pre_c〇nditi〇ns), a confidence conversion (con f 1 deuce trans iti on ) unit, and a plurality of output elements (0P: output Places) and their corresponding postconditions). Then, as long as the possibility of service demand is sent to Cui, 罔, 杈 兀 整合 ( (through additional model units), you can push and understand the service pairing process through the graphical table* method. Finally, by integrating the relevant inference results, the inference mechanism of the service f according to the probability of the Pai network is obtained. The inference = lies in the algorithm of the connection of confidence ((7)nfident transiti〇n) and the degree of confidence (4), and Through the (9) unit, the uncertainty of the process, that is, the uncertainty of the service that the Internet service in the pairing can satisfy, expresses the degree of service knocking. ",, the establishment of the initial state of the two models to facilitate the possibility of inference 9 200844762 mention / service f seeking model, another - represents the service ί = 亍 inference: in order to obtain the input 』 ί ί ΐ The inconsistency between data and service requirements (the results of the service delivery are inconsistent with the expectations of the service). The maximum probability that the service can meet the user's needs is more than two. The non-consistency dominates the uncertainty in the service pairing process, and the present invention also designs the steps of all confidence levels in the 1L through the additional base vedata-type Plngtable. The degree of confidence in the table will make the result of the service pairing more accurate. The present invention enables the service matching technology to be applied to the service recommendation system through the same mechanism as above. Since the possibility of the Pai network model unit can be used simultaneously Expressing the service characteristics or service characteristics that need to be sought, so the present invention can use the same way to describe the service of a known Internet service entity. The language is converted into a service demand model unit y and the subsequent steps are as in the previous example: 丨·Based on the conversion procedure, establish another service model unit of the Internet service; 2. Integrate the two model units. 3 Perform the possibility inference 4) Obtain the pairing result. Finally, the result can be used to indicate the degree of similarity between the two Internet service entities, and as the basis for the recommendation service, for example, when the degree of similarity is greater than 〇·8, the Internet service is recommended. Service Requirer. In accordance with a preferred embodiment of the present invention (please refer to Figure 1), a web service entity 102 may have more than one (usually only one) service description file 104 describing its services. Features, and ^ files are written according to the WSDL: Web Sendee Description Language specification and registered (registei^ 200844762 on another service registry, such as the standard UDDI service registry 106, service requirements Through the service interface provided by them, they can search for the required Internet services' and obtain detailed service descriptions. For example, a service description file. According to a preferred embodiment of the present invention, the above-mentioned Internet Service Description 2 (WSDL) is defined by the W3C organization to describe the specifications of the Internet service, and is divided into service interface information (how to Communicate with the service) and the service implementation information (the network entity address of the service).

^照本發明的較佳實施例所述(參考圖1),服務提供 1^0疋以某WSDL服務描述語言文件來描述。當服務需求1〇8 S某服務描述語言文件來描料,即可進賴謂的服務 推薦(service rec⑽mendati〇n)的服務配對功能。 依,本發明的難實_所述(參考圖3),服務配對 二疋以可忐性派翠網路來表示,服務需求模型單元表 =矛=需求,服務提供模型單元則表示服務需求,而兩者 過麟的信,時鮮元連接,完成―個完整的可能 〇ί d?/此模型還包含:輸入元πρ)、輸出元 ⑴Ρ)、.§己(·)與信心程度(Ν,π)。 圭實施例所述(參考圖3),上述可能性派 配對f型中之每-信心程請η)係由一組 (DOSsih.l +’ t別代表必要性(neCeSSity)與可能性In accordance with a preferred embodiment of the present invention (see Figure 1), the service provider is described in a WSDL service description language file. When the service needs 1〇8 S service description language file to trace, you can enter the service recommendation function (service rec(10)mendati〇n). According to the difficulty of the present invention (refer to FIG. 3), the service pairing is represented by a scalable network, the service demand model unit table = spear = demand, and the service providing model unit represents service demand, And the letter between the two, the fresh element is connected, complete - a complete possibility 〇 ί d? / This model also includes: input element πρ), output element (1) Ρ), § 己 (·) and confidence level (Ν, π). As described in the example of the Gui (refer to Figure 3), each of the above-mentioned possibilities is paired with the f-type, and the confidence step η) is composed of a group (DOSsih.l +'t represents necessity (neCeSSity) and possibility

^上L I A又分別以“Ν * Π”來表示,於圖示表 達上即以α π)來表示信心程度。 Q 務提例所述(參考圖υ,上述的服 度轉換程序128t則是透過另—系列信心程 依照本發明的較佳實施例所述(參考圖…上述的信 200844762 心程度轉換程序128係參考另一基礎資料型態(primitive data type)對應表114來進行。該對應表紀錄不同的基礎 資料型態之間的不一致性,用作建置信心程度的數值。其 它簡單型恶(simple type)或複雜崔態(complex type)之間 的不一致性則可透過其組成結構與計算公式,由對應表求 依照本發明的較佳實施例所述(參考圖1),上述服務 配對結果係透過計算公式,來將上述的可能性派翠網路服^L I A is again expressed by "Ν * Π", and the degree of confidence is expressed by α π) on the graph. According to the example of the present invention (refer to the figure, the above-mentioned service conversion program 128t is according to the preferred embodiment of the present invention through another series of confidence steps (refer to the above-mentioned letter 200844762 heart degree conversion program 128 series) Referring to another primitive data type correspondence table 114, the correspondence table records inconsistencies between different basic data types and is used as a numerical value for constructing confidence. Other simple types (simple type) Or the inconsistency between complex complex types can be obtained through the corresponding structure and calculation formula according to the preferred embodiment of the present invention (refer to FIG. 1). Calculate the formula to bring the above possibilities to the Pai network service

務配對模型所推論的結果加以整合,並以一數值表示(介於 0·0 到 1.0) 〇 ' 依照本發明的較佳實施例所述(參考圖丨),上述服務 配,結果Π6,可用來代表服務需求與服務提供之間的相 似私度,並可以據此數值決定是否推薦該網際服務I%。 本發明因直接採用網際服務描述語言(WSDL)來者作服 務配對的資訊來源,以及利用可能性派翠 礎,提出-個適合網際服務領域的服務配對方:= 服務導向_巾,騎服絲對過料遇_不確定性’問 題,以提財等級的部分輯機制。亦可 翌 網路的理論模型,透形化的方式 務配對的語意,此項優點是衣與服 能。且依據本發明的實施步驟,根據^務^g的功 依據計算公絲整合服務配對結果,即·料再 務之間的相似程度,以達成服務推薦的效果。、4-桃服 為讓本發明之上述和其他目的、特 易懂,下文特舉較佳實施例,並配合 =‘、,、=更月顯 明如下。 口所附圖式,作詳細說 【實施方式】 以下我們舰本發日种的侧單喊行料(參考圖一)。 12 200844762 一、網際服務描述語言文件 由於網際服務領域中所使用的標準服務描述語言即 是W3C所制定的網際服務描述語言(WSDL),因此本發明即 利用該#言當作最佳的服務提供表達(represen恤i〇n)。 而根據發明内容所述,服務需求也可以用相同的語言來表 達。目别相關的網際服務工具’已經可以自動根據服務開 务者所使用的程式語言,來自動產生該文件,並可依照該 文件,來產生服務需求者在呼叫服務時,所需要的程式語 言片段,大大增加了其流行的程度。也因為如此,本發明 與其他相關服務配對方法比較,除了可以確保服務描述文 件來源外,也可以提供最高的相容度。 如圖2所示,一個標準的網際服務描述語言文件2〇〇 包含以下的元件: >型態(types)元件204:透過XML Schema技術所描述 的,包含簡單型態(simple type)或複雜型態 (complex type)的資訊格式。 >訊息(messages)元件206:用以描述該網際服務實 體經由網路與外部系統傳遞的資訊格式,其組成可 以是由型態(types)元件中所描述的基本資訊。 >輸入(input)元件212與輸出(output)元件214:用 以描述該網際服務的輸入與輸出各是什麼訊息 (messages) 〇 >操作(operation)元件210:表示該網際服務所提 供的一種功能,透過其中所包含的輸入與輸出 (input/output)元件,服務需求者即可了解該操作 13 200844762 所需的輸入訊息(input message)與會回應的輪出資 訊(output message) 〇 >接口型態(portType)元件208:由多個操作 (operation)元件組合而成,用以表示同類型的功 ~ 能。通常一個網際服務實體亦可提供多種類型的功 能,藉由接口型態元件的描述,服務需求者即可了 解該網際服務的服務介面(service interface)資 訊。 φ >連結(binding)元件216:包含訊息的傳輪協定 (transport pro toco 1)或編碼(enc〇d i ng)格式等其 他額外資訊,通常相同的接口型態元件可以有不同 的實做方法,每一種實作方法即可用一個連結元件 描述。 >接口(port)元件220:用來描述該網際網路實體的 URL位置與所實做的連結元件,讓服務需求者可以依 此資訊,透過正確的傳輸協定與編碼,進行服務的 呼叫,並獲得服務結果。 力 ^ >服務(service)元件218:用來群組多個接口元件, 服務需求者即可了解該網際服務的服務實做 (service implementation)資訊。 >定義(definition)元件:最上層元件,用來整合上 述所有元件。 一由上述介紹可知,服務介面(serviee interface)相 關貧訊主要用來描述該網際服務主要的特徵,如操作、輸 入與輸出等等,在服務配對的過程中,這此將祜來 建立配對模型。在服務配對結束後,使_=== 200844762 結果選擇所需要的網際服務及其描述語言文件,再根據該 網際服務的服務實做(service implementation)資訊,進 行服務呼叫的相關事宜,但該事宜非本發明的重點,故在 此並不多加钦述。 另外,在網際服務實體部署完成後,可透過工具來自 動產生服務描述120,接著我們可以透過UDDI註冊機1〇6 來註冊122網際服務,這也包含其服務描述語言(WSDL)文 件104的取得資訊。在服務配對進行前,即可透過UDDI 註冊機的搜尋124功能,來找尋所有的網際服務,因為UDDI 注冊機106所定義的服務泛指所有服務類型,網際服務只 疋其中的一種。藉由這樣的前置動作,將可過濾掉一些不 相關的網際服務,減少候選服務的數量,進而提升服務配 對的效能。 圖4與圖5為一股票查詢網際服務“St〇ck Qu〇te” 的WSDL文件104内容,該服務包含一個“St〇ck”接口型 態元件,與三個操作元件,分別為“getprice” 、 “getQuotedStock” 與 “getHighestPrice” 等功能,分別 用來表不(1)股價查詢:輸入某一股票的代號,並輸出該股 票的目前的股價;(2)查詢股票代號:無需輸入,直接輸出 所有股票的代號;以及(3)查詢最高股價:輸入一組欲查 詢的股票代號,並輸出其中最高者的股價。其他相關的訊 息、輸入/輸出或服務實做資訊亦可由圖中看出。 二、可能性派翠網路 因為可能性派翠網路模型Π2適合用以推論不確定性 與圖性化表達的能力,本發明採用可能性派翠網路來做為 15 200844762 底層的推論機制,並由其可能性推論結果130,來計算服 務配對的結果Π6。在說明本發明的主要内容前,我們先 針對本發明所使用的可能性派翠網路及其服務配對模型的 概念進行說明。 個可能性派翠網路模型係包含多個轉換 (transition )單元 t,如信心轉換(confidence transition)單元 ta ’ 整合轉換(aggregati〇n transiti〇n) 單元 ΐ ’ 複製轉換(duplication transition)單元 td,還The results inferred from the pairing model are integrated and represented by a numerical value (between 0·0 and 1.0) 〇' according to the preferred embodiment of the present invention (refer to Figure 丨), the above service is matched, and the result Π6 is available. To represent the similarity between service demand and service provision, and to determine whether to recommend the Internet service I% based on this value. The invention adopts the Internet Service Description Language (WSDL) to directly serve as a source of information for service pairing, and utilizes the possibility to send a foundation to propose a service suitable for the Internet service field: = service orientation _ towel, riding clothes silk A partial compilation mechanism for the issue of _uncertainty. It can also be the theoretical model of the network, the way of translating the way to match the semantics, this advantage is clothing and clothing. According to the implementation steps of the present invention, the pairing result of the male silk integration service, that is, the degree of similarity between the materials and the service, is calculated according to the power of the utility to achieve the effect of the service recommendation. 4 - Peach Service In order to make the above and other objects of the present invention well understood, the preferred embodiments are hereinafter described, and in conjunction with = ', , = = more months are shown below. The description of the mouth is described in detail. [Embodiment] The following is a summary of the ship's side-by-side (see Figure 1). 12 200844762 I. Internet Service Description Language File The standard service description language used in the field of Internet services is the Internet Service Description Language (WSDL) developed by the W3C. Therefore, the present invention utilizes the #言 as the best service provider. Expression (represen shirt i〇n). According to the invention, the service requirements can also be expressed in the same language. The related Internet service tool 'is automatically generated according to the programming language used by the service sponsor, and can be used according to the file to generate the program language fragment required by the service requester when calling the service. , greatly increasing the extent of its popularity. Because of this, the present invention compares with other related service pairing methods, and can provide the highest compatibility in addition to ensuring the source of the service description file. As shown in Figure 2, a standard Internet Service Description Language file 2〇〇 contains the following components: > types element 204: described by XML Schema technology, including simple type or complex The information format of the complex type. >messages element 206: used to describe the information format transmitted by the internet service entity via the network and the external system, the composition of which may be the basic information described in the types element. > Input element 212 and output element 214: used to describe what the input and output of the Internet service are (messages) 〇 > Operation element 210: indicates that the Internet service provides A function through which the service requester can understand the input message required by the operation 13 200844762 and the output message that will be responded to by the input/output component. Interface type (portType) component 208: is composed of a plurality of operation elements to represent the same type of work. Usually an Internet service entity can also provide multiple types of functions. By describing the interface type components, the service provider can understand the service interface information of the Internet service. φ > binding component 216: other information such as transport pro toco 1 or encoding (enc〇di ng) format, usually the same interface type component can have different implementation methods Each implementation method can be described by a link component. > Port component 220: used to describe the URL location of the Internet entity and the connected component, so that the service requester can make a service call through the correct transmission protocol and coding according to the information. And get the service results. Force ^ > service element 218: used to group multiple interface components, service demanders can understand the service implementation information of the Internet service. >definition component: The uppermost component used to integrate all of the above components. As can be seen from the above introduction, the service interface (serviee interface) related information is mainly used to describe the main features of the Internet service, such as operation, input and output, etc. In the process of service pairing, this will establish a pairing model. . After the service pairing is completed, the _=== 200844762 result selects the required Internet service and its description language file, and then performs service call related information according to the service implementation information of the Internet service, but the matter It is not the focus of the present invention, so it is not mentioned here. In addition, after the deployment of the Internet Service Entity is completed, the service description 120 can be automatically generated by the tool, and then we can register the 122 Internet service through the UDDI Registry Machine 1〇6, which also includes the acquisition of its Service Description Language (WSDL) file 104. News. Before the service pairing is performed, all the Internet services can be found through the DFDI registrar's search 124 function, because the services defined by the UDDI registrar 106 refer to all service types, and the Internet service is only one of them. With such pre-actions, some unrelated Internet services can be filtered out, and the number of candidate services can be reduced, thereby improving the efficiency of service matching. Figure 4 and Figure 5 show the contents of a WSDL file 104 of a stock query internet service "St〇ck Qu〇te", which contains a "St〇ck" interface type component and three operating elements, respectively "getprice" , "getQuotedStock" and "getHighestPrice" and other functions, respectively, used to indicate (1) stock price query: enter the code of a stock, and output the current stock price of the stock; (2) query stock code: no input, direct output The code name of all stocks; and (3) query the highest stock price: enter a set of stock codes to be queried and output the stock price of the highest one. Other relevant information, input/output or service implementation information can also be seen in the figure. Second, the possibility of Pai Cui network because of the possibility of the Pai network model Π 2 is suitable for inferring the ability of uncertainty and graphical expression, the present invention uses the possibility of Pai Cui network as the underlying inference mechanism of 2008 200844762 And by inferring the result 130 from its likelihood, the result of the service pairing is calculated Π6. Before explaining the main contents of the present invention, we first explain the concept of the possibility of using the network and its service matching model. The possibility of the Pai network model system includes a plurality of transition units t, such as a confidence transition unit ta 'integration conversion (aggregati〇n transiti〇n) unit ΐ 'duplication transition unit td ,also

有其他輸入元(IP:;[叩utPlace)與輸出元(〇p: 〇utput Place),符記(token)鲁與信心程度(Ν,π)等構成。轉 換單元可與多個輸入元Ιρ與輸出元〇ρ透過箭頭連接,並 f長條型表示。輸入元ΙΡ與輸出元即以空心圓表示,若 前頭指向轉換單元t,那麼連接在該箭頭上一端的空心圓 就代輸入元IP(相較於t來說);反之若箭頭指向空 則,接在該箭耻_端的空心圓就代輸出元㈤(相較於七 輸入兀IP與輸出元〇p其内可以放 ⑽⑻•,符記上又可以帶有一個信心程二的己 一種資源允許被使用代表某 就可用來表示這樣的條件被滿足m:n) 該轉換元即可_^(fire),觸記時, 整個網路中沒有任何—㈣換元可㈣觸到 16 200844762 二、可能性派翠網路之服務配對模型 透過上義概念,油即可用可能性派翠網路 服務需求模型單元&與服務提供模型單元SP。如圖广所 二=^將輸入f lpl(rl)〜m(r4)當作是服務進行時 t (pre_ iGn)’而輸人以P1(ql)〜肌⑹則 ,作服務進行後的後置條件(PQS卜咖ditiQn),而tl声There are other input elements (IP:; [叩utPlace) and output elements (〇p: 〇utput Place), tokens (trust) and confidence levels (Ν, π). The conversion unit can be connected to the input element Ιρ and the output element 〇ρ by an arrow, and f is a long strip type. The input element and the output element are represented by a hollow circle. If the front head points to the conversion unit t, then the hollow circle connected to the upper end of the arrow is substituted for the input element IP (relative to t); otherwise, if the arrow points to the space, Connected to the hollow circle of the arrow _ _ end of the output element (5) (compared to the seven input 兀 IP and the output element 〇 p can be placed inside (10) (8) •, the token can also have a confidence of the second resource allows Being used on behalf of a can be used to indicate that such a condition is satisfied m:n) The conversion element can be _^(fire), when the note is recorded, there is nothing in the entire network - (4) The exchange element can be (4) touched 16 200844762 Possibility of the service matching model of Pai Cui Network Through the concept of Shangyi, the oil can use the possibility to send the network service demand model unit & and the service to provide the model unit SP. As shown in Fig. 2, the input f lpl(rl)~m(r4) is regarded as the service time t (pre_ iGn)' and the input is P1(ql)~muscle (6), after the service is performed. Condition (PQS Bu coffee ditiQn), and tl sound

不執行該服務。因此,從需求者的角度看來,則是“卡^ 在中壢壞了,執行該服毅,能將針轉好”,The service is not executed. Therefore, from the perspective of the demander, it is "the card ^ is broken in the middle, the implementation of the service, can turn the needle well",

務Ϊ供的角度來看則是“修轉位於台北,且能修好外掉 的相型車” ’接著細想知道與服務提供麵單元S 滿足服務需求模型單元Sr。 我們可以建置如圖3所示之服務配對模型,該模型於 ,務需求模型與服務提供模型間另外建置了多個信心轉換 早兀,如.....“,用來表達服務S求的前置條件盘 服務提供的前置條件可能產生的不確定性。其連接方向/由 服務需求模型單元Sr中的輸入元IPn+i(rn+i)〜Ιρ_(“)連接 到服務提供模型單元&的輸入元IPl(n)〜ΙΡη(Γη)上,或是 由服,提供杈型單元SP的輸出元0Ρΐ(Φ)聯接到服務需求 模型單Sr元的輸出元〇P2(q2)上。其個別對應關係由其位置 兀所代表含意進行連結,在此不予已限制,需注意的是若 有某她於,務提供模型單元Sp上的輸人元無法對應到服 務需求模型單元Sr中的任一輸入元時,此服務配對模型將 無法觸發可能性推論,亦即表示兩者之間的滿足度(相似度) 為 0· 0 〇 17 200844762 生推論:可能性派翠網路的命題推論公式 ,=的可能性派翠網路模型中,有三種轉 , 刀別表不不同的推論模式,以下將簡單說明。 ansition)單元t1及其可能性 (Α)仏心轉換(confidence tr 推論:From the perspective of the supply of goods, it is “repairing the phase car that is located in Taipei and can be repaired out”. Then we carefully know that the service provider unit S satisfies the service demand model unit Sr. We can build a service matching model as shown in Figure 3. This model creates a number of confidence conversions between the service demand model and the service provision model, such as ....., to express the service S. The uncertainty that may be generated by the preconditions provided by the precondition disk service. The connection direction / connected by the input element IPn+i(rn+i)~Ιρ_(") in the service requirement model unit Sr The input element IP1(n)~ΙΡη(Γη) of the model unit & or the output element 0Ρΐ(Φ) of the 杈-type unit SP is connected to the output element 〇P2 of the service demand model single s element (q2) )on. The individual correspondences are linked by the meanings represented by their positions, and are not limited here. It should be noted that if there is a certain woman, the input elements on the model providing unit Sp cannot correspond to the service demand model unit Sr. When any input element is used, this service pairing model will not trigger the possibility inference, that is, the satisfaction degree (similarity) between the two is 0·0 〇17 200844762 The birth inference: the proposition of the possibility to send the network Inference formula, the possibility of = in the Pai Cui network model, there are three types of inference, the different models of inference are different, the following will be briefly explained. Ansition) unit t1 and its possibility (Α) heart transformation (confidence tr inference:

如賴述,-個信心程度可以表示為(ν,π),分別 表π必要性(necessary)與可能性(p〇ssibimy)。為 I進一步朗,我舰解釋(N,Π)的涵義。假設有一個 ^ (proposition) (r, (Ν. Π〇 ) :示至夕有1一 Nr的谷易程度(degree ⑽從)說明厂 疋,的,同樣的,有至多nr的容易程度說明r是對的。透 過故樣的定義,我們可以定義出下列的公式:(此公式將 在後來的可能性推論上用到,其詳細證明步驟在此將 明)For example, the degree of confidence can be expressed as (ν, π), respectively, π necessity and possibility (p〇ssibimy). For I to further lang, my ship explained the meaning of (N, Π). Suppose there is a ^ (proposition) (r, (Ν. Π〇): indicating that there is a degree of 1 Nr in the valley (degree (10) from) to explain the factory, the same, there is at most nr easy to explain r It is correct. Through the definition of the original, we can define the following formula: (This formula will be used in the subsequent possibility inference, the detailed proof steps will be explained here)

^5 ΓΙ (Π 八广 2 八··· ) —> q 厂2,(N ”,Π,2) MN〜,ru) = min|N(nAr2A...A.„)^ = Π(π An a ... Arn) q ,夕個(η, (Nri, Πri))已知的情形下,即可求得 回到圖3的服務配對模型中,#我們在每個服務需求 18 200844762 模型單元Sr的輸入元IPn+1(rn+1)〜 ip f 心程度·α i)後,且已知輸入元;PE之;記::: 心程度後(參考圖6,例如“車子在中堤,,夕 北”兩個前魏狀_滿足度),亦即表補轉換單^ t” twv “ 的信心程度(Ν,π)、(Νι,π〇、(N2, ιω、(Νη,π„)都已知後,我們將可進行可能性推論,求 得最後位於0P2(q2)上的符記鲁所帶的信心程度Λ,π q) ’而Nq就是該服務需求1〇8與服務提供11〇帛的相似程 度。^5 ΓΙ (Π八广2 八···) —> q Plant 2, (N ”, Π, 2) MN~, ru) = min|N(nAr2A...A.„)^ = Π( π An a ... Arn) q , 夕 (η, (Nri, Πri)) is known, can be obtained back to the service pairing model of Figure 3, # 我们在服务要求18 200844762 The input element of the model unit Sr is IPn+1(rn+1)~ip f heart level·α i), and the input element is known; PE; note::: after the heart level (refer to Figure 6, for example, "the car is at Zhongdi,, Xibei "two pre-wei _ satisfaction", that is, the degree of confidence (Ν, π), (Νι, π〇, (N2, ιω, (Νη) After π„) is known, we will be able to make a possible inference to find the degree of confidence 符, π q) ', and Nq is the service demand 1 〇 8 at the last 0P2(q2). Similarity to the service provided by 11〇帛.

(B)整合轉換(aggregati〇n transiti〇n)單元栌及其可能 性推論: ^ 其推論方式和f類似,但是主要用來整合相同命題^ 的所有k心程度,以求得最佳的信心程度。其用途在當有 多個輸出元0P存在,而其所表達的是相通一個命題, 我們即可透過ta整合其信心程度,以求得最佳的信心程 度。(可參考圖18的ta) Φ 以公式來表示的話,就如以下所示: = max {Nr}(B) Integral transformation (aggregati〇n transiti〇n) unit 栌 and its possibility inference: ^ Its inference is similar to f, but mainly used to integrate all the k-degrees of the same proposition ^ to obtain the best confidence degree. Its use is when there are multiple output elements 0P, and what they express is a proposition, we can integrate the confidence level through ta to get the best confidence level. (Refer to ta in Figure 18.) Φ is expressed as a formula, as shown below: = max {Nr}

Ilg = max {IL} (C)複製轉換(duplication transition)單元1^及其可能 性推論: 其主要用來將一個輸入元IP複製成多個表示相同命 題與信心程度的輸入元IP,這對可能性派翠網路的建置特 別重要。因為推論後的信心程度會由符記Φ所帶走,而根 19 200844762 據轉換單元觸發的條件必須為其所有輸入元Ip都有符記 鲁情形下才能觸發’因此會造成競爭的問題。為避免這種 形造成該服務配對模型的推論結果,因為競爭的問題而 . 有不同的結果,則必須於服務配對模型中,加入此轉換單 • 元,產生足夠的符記•與相同的信心程度,以確保推論的 進行。(可參考圖18的td) 五、轉換程序··由WSDL文件元件建立相關可能性派翠網 • 路模型單元, 接下來將介紹一旦取得WSDL描述文件1〇4後(可透過 UDDI 5主冊機的搜尋功能),如何透過轉換程序來建立對應 的服務提供模型單元。由於WSDL文件1〇4中所描述的每一 個操作(operation)元件210皆可以是一個獨立的使用功 月匕例如本貝施例中所舉用的Stock Quote”網際服務 γ芩考圖4與圖5),即提供三種不同的服務操作,但都歸 類於相同的接口型態(portType)元件2〇8 “St〇ck”中。 _ 此外服務需求的描述方式通常為“找尋一個能提供 XXX功此的服務,而該功能大多以其功能介面描述,例 如該功能需要輸入多筆Y,資訊,輸出多筆乙資訊。根據這 樣的概念,任一 WSDL操作元件210,都可以表達為一個獨 立的且最小化的服務提供模型單元。例如圖13即表示該服 矛力長:供一個功能,讓服務需求可以輸入一股票代號,其資 料型恶為string,透過成功的服務呼叫,即可回應一股 核’其資料型態為integer,其中string與integer都是 一種簡單型態(simple type),每一個輸入元ΙΡι代表一前 置條件(pre-condition),而每一輸出元〇Pl代表一後置條 20 200844762 件(post-condition),條件的内容則是透過命題 (proposition)方式描述在當中。而圖14所示與圖 • 13所示的差異在於透過況你這個特殊方式來表示該服務 • 操作的前置條件之命題,亦即該服務操作不需要提供任何 輸入資料,即可達成呼叫服務的前置條件。圖15所示的模 型單元不同處則在於其資料型態為array of string,是 一種複雜型態(complex type)。 由於所配對的網際服務是由WSDL文件所描述而成, • 因此其配對模型相較於圖3所示的最精簡版服務配對模型 有所差異,且需要以整個網際服務來思考。為了簡化模型 的複雜度,同時提供一定程度的抽象化,我們用階級式可 能派翠網路(HPPN)來進一步隱藏過多的模型單元資訊。如 圖16所示,當轉換一個接口型態(p〇rType)元件2〇8成為 一個更大的服務提供模型單元Sp時,我們透過1(χχχ ) 這樣的HPPN來代表圖13、14與15中的各別小型服務提供 模型單元SP,而XXX的命名規則是參考該操作元件21〇的 命名空間(namespace),yyy則是該操作元件21〇的名稱, 1^則是用來表示該HPPN是某個操作元件21〇的抽象化模 組單元。因此圖16中的服務提供模型單元即可了解該接口 型態元件·包含三個操作元件,另外各制輸入了輸出 與命題條件皆可由圖形中瞭解。特別注意圖16中加入一 tU 1)信心轉換單元,用來確保當進行可能性推論時,輸 入兀IP2(n)—定會有符記(token)_流入,如此才能計算 該操作το件206所代表的服務操作能滿足服務需求&的卞 心程度。 "" 目理,@ Π為最後的抽象化的服務提供模型單元&, 21 200844762 :徂;T中^元件2Q8々Μ’所表達的小型服務 直,、果里早元又進一步被抽象化為HPt(xxx,yyy)的HPPN。 :與的命名方式同前例所述,Hpt則特別用來表示 ;。:象化ί為一接口型態元件208的服務提供模型單 二ΛΛ別注意其輸入元只剩下兩個,因為該網際服務雖然 :、二個操作元件2()6,但只需要兩種輸入,即可 ^網際服務的所有服務操作。Ilg = max {IL} (C) Duplication transition unit 1^ and its possibility inference: It is mainly used to copy an input element IP into multiple input element IPs representing the same proposition and confidence level. The possibility to build a network is particularly important. Because the degree of confidence after inference will be taken away by the token Φ, and the root 19 200844762 according to the conditions of the conversion unit must be signed for all of its input elements Ip can be triggered in the case of the situation, thus causing competition problems. In order to avoid this kind of inference result of the service pairing model, because of the competition problem, there are different results, you must add this conversion sheet in the service matching model, generate enough tokens and the same confidence. Degree to ensure that the inference is carried out. (Refer to td in Figure 18.) V. Conversion Program················································································· Machine search function), how to establish a corresponding service provision model unit through the conversion program. Since each of the operation elements 210 described in the WSDL file 1.4 can be an independent use function, for example, the Stock Quote" used in the example of the embodiment of the present invention. 5), that is, three different service operations are provided, but they are all classified into the same interface type (portType) component 2〇8 “St〇ck”. _ In addition, the service requirement is usually described as “finding one can provide XXX” This service is mostly described by its function interface. For example, this function requires inputting multiple pens Y, information, and output multiple B information. According to this concept, any WSDL operational component 210 can be expressed as an independent and minimized service provisioning model unit. For example, Figure 13 shows that the service is long: for a function, the service demand can input a stock code, and the data type is a string. Through a successful service call, it can respond to a nuclear 'its data type is integer Where string and integer are both simple types, each input element ΙΡι represents a pre-condition, and each output element 〇P1 represents a post-block 20 200844762 pieces (post- Condition), the content of the condition is described by proposition. The difference between Figure 14 and Figure 13 is that you can use the special way to indicate the proposition of the service and operation preconditions, that is, the service operation does not need to provide any input data to complete the call service. Preconditions. The difference between the model elements shown in Figure 15 is that the data type is array of string, which is a complex type. Since the paired Internet services are described by WSDL files, • the pairing model is different from the most streamlined service pairing model shown in Figure 3 and needs to be considered by the entire Internet service. To simplify the complexity of the model while providing a degree of abstraction, we use the class-like possible network (HPPN) to further hide too much model unit information. As shown in Fig. 16, when converting an interface type (p〇rType) element 2〇8 into a larger service providing model unit Sp, we represent Figures 13, 14 and 15 by HPPN such as 1(χχχ). The individual small service in the model provides the model unit SP, and the naming rule of XXX refers to the namespace of the operating element 21〇, yyy is the name of the operating element 21〇, and 1^ is used to represent the HPPN. Is an abstraction module unit of a certain operating element 21〇. Therefore, the service providing model unit in Fig. 16 can understand the interface type component, including three operating elements, and the input and output conditions of each system can be understood by the graphic. Pay particular attention to the addition of a tU 1) confidence conversion unit in Figure 16 to ensure that when a likelihood inference is made, the input 兀IP2(n) will have a token_inflow so that the operation can be calculated 206 The service operations represented can meet the service needs & level of care. "" Vision, @ Π provides the model unit for the final abstraction service &, 21 200844762 :徂;T中^Component2Q8々Μ' expressed the small service straight, and the fruit was further improved by Abstracted as HPt (xxx, yyy) HPPN. : The naming scheme is the same as the previous example, and Hpt is especially used to indicate ; The image of the service provider model of 208 is an interface type component 208. Note that there are only two input elements, because the Internet service is: two operating elements 2 () 6, but only two Enter, you can ^ all service operations of the Internet service.

^與圖8則是上述所有轉換程序的詳細步驟說明, 广建置步驟的描述是由上而下,但其實際的建置的結 ^共上述介紹相同。其步驟的簡單敘述為:依 的Sp ’並分縣之抽象化為 pt、早疋後14服務需求模型單元&整合來建置多個ppN 服務配對模型。(2)>ή: 3查g Am ΛΑ X^r At 化H媪刑二士 接口型態元件208的抽象^ and Figure 8 are detailed steps of the above all conversion procedures, the description of the extensive implementation steps is from top to bottom, but the actual implementation of the combination is the same as described above. The simple description of the steps is as follows: According to the Sp' and the abstraction of the county is pt, the service requirement model unit & integration is built to build multiple ppN service pairing models. (2) > ή: 3 check g Am ΛΑ X^r At H Hendant 2 interface type element 208 abstraction

Pt、早兀日令’則先建置每一個操作元件206的小型服 矛曰提供㈣單TL Sp ’並將之抽象為各㈣‘模型單元3 最後將額外的信d元與I模型單元整合,用以確保可 ”的進行。⑷將前-步驟的模型單it都整合進Hpt 模型單元中,即完成轉化程序的相關步驟。 p 六、信心程度轉換程序:建立完整的可能性派翠 服務配對模型。 有了服務需求·單元&與服務提供模組單元 仍需額外賴料元來連接彼此,赠立—個完整的服務 配對模型,® 9鱗細的步驟描述。由於可能性派翠網路 的特性,符記(token)·為有限的資源,任何—個轉換 加ansition)單元,包含信心轉換(c〇nfide· 22 200844762 ^^〇^元t、整合轉換(aggregati〇n t刪出⑽) 早70斤二、稷衣轉換(duphcation transition)單元 td,皆 而要符來進行推論動作。為避免來自服務需求模型中 ㈣始小型可能性網路單元所消耗,而 &成其他小型可能性網路單元皆無法進行可能性推論,因 此有必要麵—型可能_路單元都可以有符記· 通過。 刑W如所不’為達到上述目的’我們在服務需求模 1早70與服務提供模型單元sP的輸人元⑽連接部份, 是先透過複製轉·uplieatiQn transitlQn)單元td來進 行輸入兀1Ρ1(η)複製的動作。根據配對模型建置的需要, 需求模型單元&的輸入元ΙΡι⑹先進行複製 =Λ H70 IP2(ri)與輸入元1P4(n),再與 信心轉換單以(Ν1,Π1)與輸人幻ρ 輸= m⑹透過信心轉換單^表Π2)與輸人元^2 相同的,因為推論的結果必須加以整, 的輸出元0Ρ對應部份,亦透過整合轉換 ta來整合減絲。·先依據輸出元 0P對應的兩要,複製服務需求模型單元& :r^:4i^"op5(q〇^ ㈣q〇透過信心轉換單元tWN3,n3)與輸出緣== 應,輸出70 〇Ρ3(φ)透過信心轉換單元h (Ν4, Π4)與輸出 兀〇Μφ),輸出元〇Ρ4(φ)透過信心轉換單元ti 、、Pt, early 兀日令' then provide a small suit spear for each operating element 206 to provide (4) a single TL Sp 'and abstract it into each (four) 'model unit 3 and finally integrate the additional letter d element with the I model unit (4) Integrate the pre-step model single it into the Hpt model unit, ie complete the relevant steps of the conversion process. p 6. Confidence level conversion procedure: establish a complete possibility of Pai Cui service Pairing model. With service requirements, unit & and service provider module units still need additional elements to connect with each other, a complete service pairing model, a 9-scale step description. Due to the possibility of Pai Cui The characteristics of the network, token (for a limited resource, any conversion plus ansition) unit, including confidence conversion (c〇nfide· 22 200844762 ^^〇^元t, integrated conversion (aggregati〇nt deleted) (10)) 70 kg 2, dupcation transition unit td, all of which are required to perform inference actions. To avoid consumption from the (4) small-possible network unit from the service demand model, & The possibility network unit can not carry out the possibility of inference, so it is necessary to face-type possible _ road unit can have a token to pass. Criminal W if not to achieve the above purpose 'we are in service demand model 1 early 70 The input part of the input unit (10) of the service providing model unit sP is an operation of copying the input 兀1Ρ1(η) by copying the turnup/uplieatiQn transitlQn) unit td. According to the needs of the pairing model construction, the demand model unit & The input element ΙΡι(6) is first copied = Λ H70 IP2 (ri) and input element 1P4 (n), and then converted with confidence (以1, Π1) and input illusion ρ = m (6) through confidence conversion list ^ table Π 2) The same as the input element ^2, because the result of the inference must be integrated, the corresponding part of the output element 0Ρ, and also integrate the conversion ta to integrate the minus wire. · First copy the service demand model according to the two elements corresponding to the output element 0P Unit &:r^:4i^"op5(q〇^ (4) q〇 through confidence conversion unit tWN3, n3) and output edge == should, output 70 〇Ρ 3 (φ) through confidence conversion unit h (Ν4, Π 4) With the output 兀〇Μφ), the output element 〇Ρ4(φ) passes through the confidence Ti ,, conversion unit

與輸出元〇ρ7(ςι)。 , U 23 200844762 圖18中的服務配對模型可以進行可能性推論前,我 =將相關的信心轉換單元代的信心程度(队,π。計 二每一個信心程度(Νη,Πη)都表示該輸入元IP之 1或疋,出兀0P之間的滿足度。舉例來說,假設η表示 料却^料型悲為lnteger _表示為“〈_Ul:1&gt;資 枓id為stnng” ,我們想知道在服務需求模型單元&amp; 所,述的前置條下,_前置條件能符合服務提供模 里單元SP所需的前置條件的可能性,而該可能性的計算方 式,就是此兩資料型態之間的相似程度,我們透過以 算公式來求得。 ° 定義1:令typei與type2為兩個資料型態(data type),則我們可以建立一個基礎資料型態對應表114,又 稱為矩陣Μ(如圖10所示),其中的每一個數值則表示其對 應的行(column)與列(row)簡單資料型態之間的相似程 度。因此,令typei表示行,而type?表示列的話,則可得 到以下公式:With the output element 〇ρ7(ςι). , U 23 200844762 The service matching model in Figure 18 can be used before the possibility of inference, I = the degree of confidence in the unit of confidence conversion unit (team, π. 2) each confidence level (Νη, Πη) represents the input The identity of the meta-IP is 1 or 疋, and the degree of satisfaction between the 0Ps. For example, suppose that η indicates that the material is sad and the lnteger _ is expressed as "<_Ul:1> 枓 id is stnng", we want to know Under the preamble of the service demand model unit &amp;, the preconditions can meet the possibility of providing the preconditions required by the service module unit SP, and the calculation of the possibility is the two data. The degree of similarity between the types is obtained by calculating the formula. ° Definition 1: Let typei and type2 be two data types, then we can create a basic data type correspondence table 114, and Called the matrix Μ (shown in Figure 10), each of which represents the degree of similarity between its corresponding column (column) and the row's simple data type. Therefore, let typei represent the row, and type If you indicate the column, you can get the following formula:

Mtypel, type2 二 similarity y 0 ^ similarity ^ 1 定義2:令typei與type2為兩個基本資料型態(pdt: Primitive Data Type),則 typei 對應到 type2 的相似程度 計算公式的定義如下:Mtypel, type2 2 similarity y 0 ^ similarity ^ 1 Definition 2: Let typei and type2 be two basic data types (pdt: Primitive Data Type), then typei corresponds to the similarity of type2. The calculation formula is defined as follows:

MatchpDT {typei -&gt; type2 ) = M^ 定義3:令type!i type2為兩個資料型態,可能為簡 單資料型態(SDT: Simple Data Type)或基本資料型態 24 200844762 (PDT) ’則type!對應到type2的相似程度計算公式的定義 如下: • MatchsDT (typei~^type2) = 1, if typei. name equals typG2. nQine.MatchpDT {typei -> type2 ) = M^ Definition 3: Let type!i type2 be two data types, which may be SDT: Simple Data Type or Basic Data Type 24 200844762 (PDT) ' The formula for calculating the similarity degree of type! to type2 is as follows: • MatchsDT (typei~^type2) = 1, if typei. name equals typG2. nQine.

MatchsoT (typei-&gt;type2) - Matchpoi (styPeiy &quot; where S typei is either type, (if typei is a PDT) or the restriction PDT type/ used to define typei.MatchsoT (typei-&gt;type2) - Matchpoi (styPeiy &quot; where S typei is either type, (if typei is a PDT) or the restriction PDT type/ used to define typei.

定義4:令type為一資料型態,則功能 “getElements(type)” 的定義如下: 1. getElements( type) = {typei), where typei is the directed sub-element of type. 2. getElements(type) = {type), if type is either PDT or SDT. 定義5:令typei與type2為兩個資料型態(可能是 PDT、SDT 或 CDT: Complex Data Type),則 typei 對應到 type2的相似程度計算公式的定義如下: 1. MatchDT {typei-&gt;type2) = 1, if typei. name equals type2. name. 2. MatchDT (typei-&gt;type2) = typei-&gt;type2), if both typei and type2 are not CDT. 3. MatchDT (typei-&gt;type2) =Avg {MatchDT (SUbtypel, SUbtype2) }, for each subtyPei, where subtyPei in getElements(⑽ 由於WSDL文件中的型態(types)元件所定義的相關資 25 200844762 料型悲係利用XML Schema技術來達成,而根據其設計,所 有資料都是由内建的基本資料型態(PDT)所構成。簡單資料 型態(SDT)則是透過別名(alias)或限制(restricti〇n)等 方式由基本資料型態(PDT)所定義,例如只是單純的換各名 • 稱或是重新限制PDT的值域(value range),來產生新的資 料型態。而複雜資料型態(CDT)則是透過各種組合結構,來 將PDT、SDT、或是CDT等作進一步的組合來產生新的資料 型態。 、 ⑩ 根據上述的定義,我們即可以得到下面的資料型態配 對相似度: (1) MatchDT {string -&gt; array_of_string) ^ Avg { MatchDT (string -&gt; string)} =MatchsDT {string -&gt; string) -Msir//2^·, string ~ 1.0。 (2) MatchDT (array_of_string -&gt; integer) =Avg { MatchDT {string -&gt; integer)} ~ MatchsDT {string -&gt; integer、 I ~~ yistring, integer — 0. 5 ° 而上述的資料型態配對相似度,即是每一個信心轉換 單元(Νη,Πη)中,Nn的實際值。由於Πη代表該推論的 可能性,其值皆為1(我們不考慮為0的情形)。因此,假 設圖18中的服務需求模型單元S”為“找尋依網際服務, 該服務提供一種功能,該功能將輸入一個代表股票代號的 〈stock symbol〉,其型態為string,且會輸出該股票代號 的股價〈price〉’其型態為integer。”那我們即可求出每 個信心轉換單元t、(Νι,Πι)〜t、(Ns,lb)中的Νι〜的 26 200844762 值··Definition 4: Let type be a data type, the function "getElements(type)" is defined as follows: 1. getElements( type) = {typei), where typei is the directed sub-element of type. 2. getElements(type ) = {type), if type is either PDT or SDT. Definition 5: Let typei and type2 be two data types (possibly PDT, SDT or CDT: Complex Data Type), then typei corresponds to type 2 similarity calculation The formula is defined as follows: 1. MatchDT {typei-&gt;type2) = 1, if typei. name equals type2. name. 2. MatchDT (typei-&gt;type2) = typei-&gt;type2), if both typei and type2 Are not CDT. 3. MatchDT (typei-&gt;type2) =Avg {MatchDT (SUbtypel, SUbtype2) }, for each subtyPei, where subtyPei in getElements((10) Correlation defined by the type elements in the WSDL file资 25 200844762 The sadness of the material is achieved by using XML Schema technology, and according to its design, all data is composed of built-in basic data type (PDT). Simple data type (SDT) is through alias (alias) ) or restrictions (restricti〇n), etc. by basic data type PDT) is defined as, for example, simply changing the name or re-limiting the value range of the PDT to generate a new data type. The complex data type (CDT) is through various combinations. To further combine PDT, SDT, or CDT to generate a new data type., 10 According to the above definition, we can get the following data type pairing similarity: (1) MatchDT {string -> ; array_of_string) ^ Avg { MatchDT (string -&gt; string)} =MatchsDT {string -> string) -Msir//2^·, string ~ 1.0. (2) MatchDT (array_of_string -> integer) =Avg { MatchDT {string -> integer)} ~ MatchsDT {string -> integer, I ~~ yistring, integer — 0. 5 ° and the above data type pairing The similarity is the actual value of Nn in each confidence conversion unit (Νη, Πη). Since Πη represents the possibility of this inference, its value is 1 (we do not consider the case of 0). Therefore, assuming that the service demand model unit S" in Fig. 18 is "finding the Internet-based service, the service provides a function that inputs a <stock symbol> representing the stock symbol, whose type is string, and will output the The stock price of the stock code <price> is of type integer. "Then we can find the value of each confidence conversion unit t, (Νι,Πι)~t, (Ns, lb) 26 200844762 value ··

Ni = MatchDi {string -&gt; string) = l.〇 N2- Matchoi {string -&gt; arrays fairing) - ie〇Ni = MatchDi {string -> string) = l.〇 N2- Matchoi {string -> arrays fairing) - ie〇

Na ^ MatchDi {integer -&gt; integer) = 1.0 N4 - MatchoT {array_of^string -&gt; integer) - 〇e 5Na ^ MatchDi {integer -> integer) = 1.0 N4 - MatchoT {array_of^string -> integer) - 〇e 5

Ns = MatchDi {integer -&gt; integer) - l.〇 七、服務配對結果計算與服務推薦應用 有了服務需求模組單元Sr與服務提供模組單元s 以及兩者之_推論連結後u所有服務需求槿 &amp;的輪入it IP上都放置符記與信心程度Φ(11),接 = 可進行可能性推論13G,取得推論結果(即最後留在 = 求,組單元Sr的輸出元0Ρ上的符記所帶的信心程度):: 由這些結果來計算最後的服務配對結果116。 亚 本發明所設計之網際服務的服務配對目的,即在“ 尋一網際服務,而該網際服務提供至少一個以上的服務操 作以最高的相似程度來滿足服務需求所需的功能,,。、 由圖18可知(其他相關信心程度計算請參考前一章矿 之敘述)’該可能性推論的結果為(1. G,L Q),也就說明^ 服務至少提供-個以上的服務操作可以滿足該服務須求 108,若以各別操作(operationk件來看,其各別的相似 度分別為: “getPrice” 服務操作:1.0 “getQuotedStocks” 服務操作:ο』 “getHighestPrice” 服務操作:1〇 因此,該服務提供110 “st0ckQU0te,,滿足的程度即 27 200844762 為1· 0,亦即最後的服務配對結果116。此外當一網際服務 描述語言文件中有多個以上的接口型態(portType)元件 208被定義時,我們仍可針對個別的接口型態(p〇r1Uype) 元件進行服務配對動作,並紀錄最高的服務配對結果,而 最後所記錄的最高配對程度,即是整個服務配對的結果。 詳細的計算公式說明請參考圖11的步驟說明。 另外,若服務需求本身亦透過服務描述語言(WSDL)來 描述的話,亦即使用者想要進行所謂的推薦服務132功能 B寸’其進行方式和上述方式相同,不同的地方在於,其服 務配對的目標變成:“網際服務A所定義的相關接口型態 元件滿足網際服務B中所有操作元件的平均值,,。实今鐵 際服務A為服務提供,而網際服務B為服務需求,而以平 均值(average value)代替最佳值(best value)的用意在於 避免採用單一操作元件的配對所計算出的配對結果。由一 個網際服務找尋(或推薦)另一個相似的網際服務,其目的 則在比較兩者之間的整體相似程度,故每一種單一操作元 件間的配對所計算出來的配對結果皆要考慮,才能完整的 衡量整體的相似度,詳細的計算方法請參考圖17所描述的 計算步驟。 回到圖18上’若該服務需求reqUest是由一個服務 描述文件所描述104,而該網際服務只定義一個接口型態 (portType)元件 208 叫作 PTre quest ,且該元件内亦只定義一 個操作(operation)元件210叫作Operationre_st,則網際 服務“Stock Quote”針對網際服務“reqUest”的即為:Ns = MatchDi {integer -&gt; integer) - l.〇7, service pairing result calculation and service recommendation application With service requirement module unit Sr and service provision module unit s and both _ inference link u service The demand 槿 &amp; the round of it IP is placed on the IP and the degree of confidence Φ (11), then = can be inferred 13G, the inference result is obtained (that is, finally left in = seeking, the output unit 0 of the group unit Sr The degree of confidence carried by the signature):: The final service pairing result 116 is calculated from these results. The service matching purpose of the Internet service designed by the invention is to find an internet service, and the internet service provides at least one service operation to meet the service requirements with the highest degree of similarity, Figure 18 shows that (for the calculation of other relevant confidence levels, please refer to the description of the mine in the previous chapter.) The result of this possibility is (1. G, LQ), which means that the service provides at least one or more service operations. The service requires 108. If the operations are different, the respective similarities are: “getPrice” Service operation: 1.0 “getQuotedStocks” Service operation: ο” “getHighestPrice” Service operation: 1〇 Therefore, The service provides 110 "st0ckQU0te," the degree of satisfaction is 27 200844762 is 1.0, which is the final service pairing result 116. In addition, there are more than one interface type (portType) elements 208 in an internet service description language file. When defined, we can still perform service pairing actions for individual interface types (p〇r1Uype) components and record the highest service allocation. As a result, the highest degree of pairing recorded last is the result of the entire service pairing. For a detailed description of the calculation formula, please refer to the step description of Figure 11. In addition, if the service requirement itself is also described by the Service Description Language (WSDL), That is, the user wants to perform the so-called recommendation service 132 function B"'s way of doing the same as the above, except that the goal of the service pairing becomes: "The relevant interface type component defined by the Internet service A satisfies the Internet. The average value of all operating elements in service B. The actual interstellar service A is provided for the service, while the internet service B is the service demand, and the purpose of replacing the best value with the average value is to avoid Pairing results calculated using pairing of single operating elements. Another similar internet service is sought (or recommended) by an internet service, the purpose of which is to compare the overall similarity between the two, so each single operating element The pairing results calculated by the pairing must be considered in order to fully measure the overall similarity. For the calculation method, please refer to the calculation steps described in Fig. 17. Returning to Fig. 18, if the service requirement reqUest is described by a service description file 104, the internet service defines only one interface type (portType) element 208 called As PTre quest, and only one operation element 210 is defined as Operationre_st in the component, the Internet service "Stock Quote" for the Internet service "reqUest" is:

Average { Match (an operation in “request” , “StockAverage { Match (an operation in “request” , “Stock

Quote” ) } 28 200844762 =Match (Operatiorirequest, “Stock Quote” )/1 二 1·0 / 1 二 1.0。Quote" ) } 28 200844762 =Match (Operatiorirequest, "Stock Quote" ) / 1 2 1 / 0 / 1 2 1.0.

因為網際服務“request”只有一個操作元件 Operatiorwiest,而網際服務“Stock Quote”能滿足該操 作元件的配對程度就同上述例子中所述。反之,若是網際 服務“request”變成服務提供,而網際服務“stock Quote”變成服務需求,其結果就跟前面不同,而變成· Average { Match (an operation in &quot;Stock Quote” , “request” ) } ={ Match (getPrice, “request” ) + Match (getQuotedStocks, “request” ) + Match (getHigestPrice, “request” ) } / 3 =(1.0 + 0·0 + 1.0) / 3 二 0·67〇 根據上述的說明,即可由一個網際服務χ當做服務需 求目標,來找尋能滿足該需求的其他網際服務γ。當兩者 的相似程度(即服務配對結果116)大於某一 ^界值 (threshold)時,即可進行所謂的推荐服務132,提供使用 者選擇上的參考。 八、總結 透過轉換程序,服務需求與服務提供都可以利用現行 的網際服健賴言⑽DL)純贿,从增加了本發明 的實用性。相較於其他技術應用目前Μ 務描述語言,本發雜易於與現行的相關技術做整合。 透過利用ΡΡΝ為基礎的服務配對方法,可以藉由掌握 29 200844762 在配對過程中所解決的不確定,以促進部分配對的實現, 亦即服務配對的結果有等級之分。另外PPN的理論模型, 更有助於於透過圖形化方式觀看服務需求或是服務提供所 代表的涵義,並透過推論機制,暸解服務配對的過程。將 了月b性推論的結果加以整合,即可用於服務相似度的計 算’並可進一步利用於服務推薦系統的實做上。Since the Internet service "request" has only one operating element Operatiorwiest, the Internet service "Stock Quote" can satisfy the degree of pairing of the operating elements as described in the above example. On the other hand, if the Internet service "request" becomes a service offering, and the Internet service "stock Quote" becomes a service demand, the result is different from the previous one, and becomes "Eight { Match (an operation in &quot; Stock Quote", "request") } ={ Match (getPrice, “request” ) + Match (getQuotedStocks, “request” ) + Match (getHigestPrice, “request” ) } / 3 =(1.0 + 0·0 + 1.0) / 3 2·67〇 In the above description, an Internet service can be used as a service demand target to find other Internet services γ that can satisfy the demand. When the similarity between the two (ie, the service matching result 116) is greater than a certain threshold (threshold) The so-called recommendation service 132 can be performed to provide a reference for the user's choice. VIII. Summary Through the conversion process, both the service demand and the service provision can utilize the current Internet service health (10) DL) to purely bribe, and the invention has been increased. Practicality. Compared with other technology applications, the current description language is easy to integrate with the current related technologies. The pairing method can promote the realization of partial pairing by grasping the uncertainty solved in the pairing process of 29 200844762, that is, the result of service pairing is graded. In addition, the theoretical model of PPN is more conducive to Graphically view the meaning of service requirements or service provision, and understand the process of service matching through inference mechanism. Integrate the results of monthly inferences, which can be used for service similarity calculations' and can be further utilized. In the implementation of the service recommendation system.

雖然本發明已以較佳實施例揭露如上’然其並非用以 限定本發明,任何熟習此技藝者,在不脫離本發明之精神 ^範圍内,當可作些許之更動與潤飾,因此本發明之保護 範圍當視後附之申請專利範圍所界定者為準。 【圖式簡單說明】 圖1繪示為依照本發明所述實施方法之方塊示意圖。 圖2纟胃示為依照本發明一較佳實施例所列舉之一 WSDL文件# 格的方塊示意圖。 ' 圖3繪示為依照本發明一較佳實施例所列舉之一透過可能性 派翠網路模型進行服務配對的抽象模型示意圖。 圖‘4與圖5缘不為依照本發明一較佳實施例戶斤列舉之一網際服 務“Stock Quote”的網際服務描述語言(WSDL)實 發明—較佳實施綱鱗之—服務需求與 服務七供透過可能性派翠網路表達後 r實明一較佳實施例所列舉二 佳實關_舉之-信心程度轉 本發明—触實施漸列舉之-制資料形 態對應表之實際内容。 牛心暴礎貝枓型 圖11與圖12緣示為依照本發明一釦與 配對結果的計算方法之實際計算條之一服務 30 200844762 圖13繪示為依照本發明一較佳實施例所列舉之一 WSdl操作元 件“getPrice”所轉換而成的PPN模型單元範例。 圖14繪示為依照本發明一較佳實施例所列舉之一 WSDL操作元 件“getQuotedStocks”所轉換而成的PPN模型單元範例。 圖15繪示為依照本發明一較佳實施例所列舉之一 WSDL操作元 件“getHighestPrice”所轉換而成的PPN模型單元範例。 圖16繪示為依照本發明一較佳實施例所列舉之一 WSDL接口型 態元件“Stock”所轉換而成的PPN模型單元範例。 圖17繪示為依照本發明一較佳實施例所列舉之一網際服務 “Stock Quote”所轉換而成的PPN模型單元範例。 圖18繪示為依照本發明一較佳實施例所列舉之Request與網 際服務“Stock Quote”所轉換而成的一 ppn服務配對模型範 例0 【主要元件符號說明】 100:網際服務配對方法 102 :網際服務實體 104 : WSDL服務描述語言文件 106 : UDDI註冊機 108 :服務需求 110:服務提供 112 :可能性派翠網路配對模型 114 :基礎資料型態對應表 116 :服務配對結果 120 :產生服務描述 122 :註冊 124 =搜尋 126 :轉換程序 31 200844762 128 :信心程度轉換程序 130 :進行可能性推論 132 :推薦服務 200 : WSDL文件規格 202 ··定義(definition)元件 204 :型態(types)元件 206 ··訊息(messages)元件 208 :接口型態(portType)元件 210 :操作(operation)元件 212 :輸入(input)元件 214 :輸出(output)元件 216 :連結(binding)元件 218 :服務(service)元件 220 :接口(port)元件 222 :參照 ΙΊ〜Γη+m :前置條件 qi〜q4 :後置條件 ΙΡί (ΙΊ)〜IPn+m (Γη+m):輸入元 OPi (qi)〜OP7 (q7) ··輸出元While the present invention has been described in its preferred embodiments, the present invention is not intended to limit the invention, and the present invention may be modified and modified without departing from the spirit and scope of the invention. The scope of protection is subject to the definition of the scope of the patent application. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention. Figure 2 is a block diagram showing one of the WSDL file # cells listed in accordance with a preferred embodiment of the present invention. FIG. 3 is a schematic diagram showing an abstract model for service pairing through a Pocy network model according to a preferred embodiment of the present invention. FIG. 4 and FIG. 5 are not the Internet Service Description Language (WSDL) of one of the internet services “Stock Quote” according to a preferred embodiment of the present invention. Seventh through the possibility of sending a network of expressions, after the expression of a better embodiment of the two best-selling _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ FIG. 11 and FIG. 12 are diagrams showing one of the actual calculation bars of the calculation method of a buckle and pairing result according to the present invention. 30 200844762 FIG. 13 is a diagram illustrating a preferred embodiment of the present invention. An example of a PPN model unit converted from a WSdl operating element "getPrice". Figure 14 is a diagram showing an example of a PPN model unit converted from a WSDL operation element "getQuotedStocks" according to a preferred embodiment of the present invention. Figure 15 is a diagram showing an example of a PPN model unit converted from a WSDL operation element "getHighestPrice" according to a preferred embodiment of the present invention. Figure 16 is a diagram showing an example of a PPN model unit converted from a WSDL interface type element "Stock" according to a preferred embodiment of the present invention. FIG. 17 is a diagram showing an example of a PPN model unit converted from an internet service "Stock Quote" according to a preferred embodiment of the present invention. FIG. 18 is a diagram showing a PPN service pairing model converted from a Request and an Internet service “Stock Quote” according to a preferred embodiment of the present invention. [Main Component Symbol Description] 100: Internet Service Pairing Method 102: Internet Service Entity 104: WSDL Service Description Language File 106: UDDI Registrar 108: Service Requirement 110: Service Provision 112: Possibilities Pai Network Matching Model 114: Basic Data Type Correspondence Table 116: Service Pairing Result 120: Generating Service Description 122: Registration 124 = Search 126: Conversion Procedure 31 200844762 128: Confidence Degree Conversion Program 130: Probability of Regression 132: Recommendation Service 200: WSDL File Specification 202 • Definition element 204: Types element 206·messages element 208: interface type element 210: operation element 212: input element 214: output element 216: binding element 218: service (service) Element 220: Port element 222: Reference ΙΊ~Γη+m: Precondition qi~q4: Postcondition ΙΡί (ΙΊ)~IPn+m (Γη+m): Input OPi (qi) ~OP7 (q7) ·· output element

Sp :服務提供模型單元Sp : service provider model unit

Sr :服務需求模型單元 (1,1)、(N,Π)、(Nn,IL·):信心程度 • (1,1):符記與信心程度 t1、(1,1 )〜ίΛ ( Νη,Π η)、tS 〜、 t request (1,1)、t1 getPrice (1,1)、t1 getQiiotedStocks (1,1) t getHighestPrice (1,1):信心轉換單元 32 200844762Sr: service demand model unit (1,1), (N,Π), (Nn,IL·): degree of confidence • (1,1): degree of confidence and confidence t1, (1,1)~ίΛ ( Νη , Π η), tS 〜, t request (1,1), t1 getPrice (1,1), t1 getQiiotedStocks (1,1) t getHighestPrice (1,1): confidence conversion unit 32 200844762

ta:整合轉換單元 td:複製轉換單元Ta: integrated conversion unit td: copy conversion unit

Hop (ns, getPrice)、Hop (ns, getQuotedStocks)-H〇P (ns,getHighestPrice)、IU(ns,Stock):階層式 PPM 模型單元 33Hop (ns, getPrice), Hop (ns, getQuotedStocks)-H〇P (ns, getHighestPrice), IU(ns, Stock): Hierarchical PPM model unit 33

Claims (1)

200844762 十、申請專利範圍: 1·、種應料能性㈣網路模型於網際服務的服務配對方 法’包括: . (a) 一個可能性派翠網路之服務配對模型,用以圖形化表達 •.絲需求與服務提朗服務崎涵義,以及贿配對過程 的推論; (b)轉換程序,絲將服務冑求或是服務提供轉換成對應 的服務需求模型單元與服務提供模型單元; ⑷-基礎資料型態對應表,絲計算配對模型中所需 心程度; ° (d) —信心程度轉換程序,用來結合服務需求模型單元與服 務提供模型單元,並產生服務配對模型; ^ (e) —服務配對結果計算公式,用來根據服務配對模型推論 的、、、口果计异服務需求與服務提供之間的相似程度,用以 當作服務配對結果或服務推薦之參考依據。 2·如申請專利範圍第1項所述之服務提供,係用一標準的網 際服務描述語言(WSDL)文件來描述。 、 籲3·如=請專利範U第1項所述之服務服務需求,在進行服務 推薦功能時,亦係用一標準的網際服務描述語言文件來描 述。 4·如申請專利範圍第1項所述之可能性派翠網路之服務配對 模型,包括服務需求模型單元與服務提供模型單元,另外 也包含數個k心轉換(confident transition)單元、敕 合轉換(aggregation transition)單元、複製轉二 (duplication transition)單元等用以連接前二者、以及 相關單元的信心程度(confident ievei)。 34 200844762 5·如申請專利範圍第4項所述之服務需求模型單元與服務提 供模型單元’都包含數個輸入元(input place)、輸出元 (output Place)、與信心轉換單元,分別用來表示欲搜尋 的網際服務或已註冊的網際服務。 7· 6·如申請專利範圍第5項所述之每一輸入元,其代表意義為 服矛々執行的各別前置條件(pre-C〇ndifi〇n),亦即該服^务 執行所需的各種輸入資料(i叩ut data)的型態限制。乃200844762 X. The scope of application for patents: 1. The application of the network (4) The network model service matching method for Internet services' includes: (a) A possibility to match the network service model for graphical expression • Silk demand and service to promote the service of Qi Hanyi, and the inference of the bribe matching process; (b) conversion procedures, silk service request or service provision into the corresponding service demand model unit and service delivery model unit; (4)- The basic data type correspondence table, the silk computing the required degree of the pairing model; ° (d) - the confidence degree conversion program, which is used to provide the model unit in combination with the service demand model unit and the service, and generate a service matching model; ^ (e) - The service pairing result calculation formula is used to refer to the service matching model, the degree of similarity between the service demand and the service provision, and is used as a reference for service matching results or service recommendation. 2. The provision of services as described in item 1 of the patent application is described in a standard Internet Service Description Language (WSDL) file. , 3, such as = the service service requirements described in the first paragraph of the patent model U, when using the service recommendation function, is also described by a standard Internet service description language file. 4. The service matching model of the possibility of the Pui Cui network, as described in item 1 of the patent application scope, includes the service demand model unit and the service provision model unit, and also includes several confidential transition units. An aggregation transition unit, a duplication transition unit, and the like are used to connect the former two and the confidence ievei of the relevant unit. 34 200844762 5. The service demand model unit and the service provision model unit described in item 4 of the patent application scope include a plurality of input places, output places, and confidence conversion units, respectively. Indicates the Internet service you are searching for or the registered Internet service. 7·6·If each input element mentioned in item 5 of the patent application scope, the representative meaning is the pre-condition (pre-C〇ndifi〇n) executed by the spear, that is, the service execution The type restrictions of the various input data (i叩ut data) required. Nai 如申請專利範圍第5項所述之每一輸出元,其代表意義為 服務執行的各別後置條件(p〇st_c〇nditi〇n),亦^該服 知執行後所需的各種輸出資料⑺社口此data)的型態限制。 8.如^明專利範圍弟5項所述之每一信心轉換單元,其代表 意義為在各別前置條件下,若執行該服務所能產生各別 置條件的可能性。 9·如申請專利範圍第5項所述之服務提供模型單元亦可由數 個輸入元j輸出元與一個結構式可能性派翠網路模型(HppN: Hierarchical ppN)單元表達,方便降低圖形化的複雜度, 以提高圖形模組的可讀性。 又 10·如申請專利顧第9項所述之結構式可能性派翠網路 (HPfN)單元,含數個輸入元、輸出元、與信心轉換單元, 同時亦可包含另一個結構式可能性派翠網路模型。 11·如申料利範圍第〗項所述之轉換程序,制來將描述服 務需求或服務提供的網際服務描述語言文件轉換成對鹿 服務需求與服務提供模型單元。 “、 12.如申請專利範圍第卜4項所述之信心程度轉換程序,係用 來將服務需求模型單元與服務提供模型單透過多個額 轉換(t—ation)單元、輸入元或輸出元結合,並 35 200844762 含相關信心程度數值的計算。 13·如申請專利範圍第1項所述之服務配對結果計算公々 ^ 利用可能性推論後的最大信心程度或平均信心程度,做為 ; 服務需求與服務提供之間滿足的相似程度依據。 _ ·- 14.如申請專利範圍第13項所述之相似程度,係透過計算後, 用一小數數值所表示,其值介於〇· 〇〜1· 〇之間。數值越大, 表示服務配對結果越好,亦表示推薦該服務提供給予使用 者’越能滿足其需求。For each of the output elements described in item 5 of the patent application, the representative meaning is the respective post-conditions for service execution (p〇st_c〇nditi〇n), and the various output data required after execution are known. (7) The type restriction of this data). 8. Each confidence conversion unit as described in item 5 of the patent scope is representative of the possibility that each of the conditions can be generated if the service is executed under respective preconditions. 9. The service provision model unit described in item 5 of the patent application scope may also be expressed by a plurality of input element j output elements and a structural possibility Pai network model (HppN: Hierarchical ppN) unit, which is convenient for reducing the graphical Complexity to improve the readability of graphics modules. 10) The structural possibility of the GPU network (HPfN) unit, as described in the application patent patent item 9, includes several input elements, output elements, and confidence conversion units, and may also contain another structural possibility. Pai Cui network model. 11. The conversion procedure described in item ‧ of the application scope is to convert the Internet Service Description Language document describing the service requirement or service provision into a model unit for the deer service demand and service provision. ", 12. The confidence level conversion procedure described in item 4 of the patent application scope is used to transmit the service demand model unit and the service provision model list through a plurality of t-ation units, input elements or output elements. Combine, and 35 200844762 contains the calculation of the value of the relevant confidence level. 13·If the service pairing result described in item 1 of the patent application scope is calculated, the maximum confidence level or average confidence level after the possibility of inference is used; The degree of similarity between demand and service provision. _ ·- 14. The similarity degree mentioned in item 13 of the patent application scope is expressed by a decimal value after calculation, and its value is between 〇· 〇~ 1. Between 〇. The higher the value, the better the service pairing result. It also means that the service is offered to the user to 'satisfy their needs.' 3636
TW96116402A 2007-05-08 2007-05-08 A possibilistic petri-nets based method for web service matchmaking TW200844762A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW96116402A TW200844762A (en) 2007-05-08 2007-05-08 A possibilistic petri-nets based method for web service matchmaking

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW96116402A TW200844762A (en) 2007-05-08 2007-05-08 A possibilistic petri-nets based method for web service matchmaking

Publications (1)

Publication Number Publication Date
TW200844762A true TW200844762A (en) 2008-11-16

Family

ID=44822709

Family Applications (1)

Application Number Title Priority Date Filing Date
TW96116402A TW200844762A (en) 2007-05-08 2007-05-08 A possibilistic petri-nets based method for web service matchmaking

Country Status (1)

Country Link
TW (1) TW200844762A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103646061A (en) * 2013-12-02 2014-03-19 东南大学 Service-oriented demand analysis method
TWI473028B (en) * 2012-05-09 2015-02-11 Univ Nat Pingtung Sci & Tech Method for recommending a web service and computer product thereof
CN110261159A (en) * 2019-07-31 2019-09-20 上海应用技术大学 Flexible manufacturing cutter subsystem fault diagnostic method
CN113343507A (en) * 2021-07-07 2021-09-03 广州昇谷科技有限公司 Web service combination discovery method for water conservancy surveying
TWI805394B (en) * 2021-06-03 2023-06-11 游宗憲 Method for analyzing reachability of petri nets

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI473028B (en) * 2012-05-09 2015-02-11 Univ Nat Pingtung Sci & Tech Method for recommending a web service and computer product thereof
CN103646061A (en) * 2013-12-02 2014-03-19 东南大学 Service-oriented demand analysis method
CN103646061B (en) * 2013-12-02 2017-02-15 东南大学 Service-oriented demand analysis method
CN110261159A (en) * 2019-07-31 2019-09-20 上海应用技术大学 Flexible manufacturing cutter subsystem fault diagnostic method
CN110261159B (en) * 2019-07-31 2021-06-29 上海应用技术大学 Fault diagnosis method for flexible manufacturing cutter subsystem
TWI805394B (en) * 2021-06-03 2023-06-11 游宗憲 Method for analyzing reachability of petri nets
CN113343507A (en) * 2021-07-07 2021-09-03 广州昇谷科技有限公司 Web service combination discovery method for water conservancy surveying
CN113343507B (en) * 2021-07-07 2024-05-14 广州昇谷科技有限公司 Web service combination discovery method for water conservancy survey

Similar Documents

Publication Publication Date Title
Mayer et al. An extensible discovery service for smart things
CN101958848B (en) There is the answer system that the social community of collaboration constraints generates
CN107274186A (en) The method and apparatus that intelligent contract-defined interface is obtained in block chain
US20030036917A1 (en) Service provision system and method
CN103329098B (en) document workflow architecture
TW200844762A (en) A possibilistic petri-nets based method for web service matchmaking
CN109104354A (en) Enter group, group building method and its equipment
CN103699667B (en) Web service multi-dimensional semantic model building method
Ma et al. A semantic QoS-aware discovery framework for web services
CN105453058B (en) Directory service is found and/or study
CN104320396A (en) Data interaction device and method
US20230259938A1 (en) Blockchain-based data processing method and apparatus, device, readable storage medium and computer program product
CN103699391A (en) Web service transformation method from traditional web services to multi-dimensional semantic models
TWI289261B (en) System and method for dynamically generating a HTTP query
CN102984297B (en) Its system of IDN register method realized based on domain name group
Huang et al. A semantic enhanced service exposure model for a converged service environment
Berardi Automatic Service Composition. Models, Techniques and Tools.
CN103164221B (en) Service modeling device and service modeling method
CN105354270A (en) User medical data query method and distributed system
CN105516269B (en) The configuration method of application and the configuration device of application
WO2020015404A1 (en) Internet customized information publishing and search service system
CN102394874B (en) TUXEDO (Transaction for UNIX has been Extended for Distributed Operation) data protocol conversion method
TW200941348A (en) Dynamic intermediate language modification and replacement
Harillo et al. LoRaCoin: Towards a blockchain-based platform for managing LoRa devices
Wombacher et al. Finding trading partners to establish ad-hoc business processes