[/align] 个人主页：
http://dit.unitn.it/~kokash/
研究兴趣（Research interests）：
* Web Services and SOAs
* Business Process Modeling
* Software Engineering
* Networks and Mobile Systems
* Formal Methods
PhD Thesis：
Engineering Service-Oriented Systems: Modeling, Discovery and Quality 2008（M&D）
发表论文：
2009：
1、 Formal Behavioral Modeling and Compliance Analysis for Service- Oriented systems
2、 A Flexible QoS-aware Routing Protocol for Infrastructure-less B3G Networks
2008:
3、 Collaborative web service discovery with the Implicit Culture Framework
4、 Design Time Service Composition with Reo Coordination Tools
5、Towards Using Reo for Compliance-aware Business Process Modelling
2007
Kokash, N., D'Andrea, V.: "Evaluating Quality of Web Services: A Risk-driven Approach", Poznan, Poland, April 2007, LNCS 4439, pp. 180-194 (SC/Fail)
2006
Kokash, N.: "A Service Selection Model to Improve Composition Reliability", International Workshop on Artificial Intelligence for Service Composition (AISC), in conjunction with the ECAI, Riva del Garda, Italy, August 2006, pp. 9--14 (Service Selection)
本文服务（service）定义：
A service is an abstraction that encapsulates a certain software function.（软件功能）
服务的功能通过接口(interfaces)来实现的
One of the fundamental and distinctive characteristics of services is a separation of interfaces from actual implementation.
客户端应用程序可以通过接口查找和调用服务（服务的查找和调用）
The interfaces are exploited by client applications to locate and invoke service functionalities.
web services的定义：
web services as “loosely coupled, reusable software components that semantically encapsulate discrete functionality and are distributed and programmatically accessible over standard Internet protocols”.---http://www.perfectxml.com/Xanalysis/TSG/TSG_DefiningWebServices.pdf
SLA的定义与功能
Service Level Agreement (SLA) is the most notable achievement in formulating the reciprocal obligations（相互义务？） between a service and a client.
SLAs specify organizational aspects of interactions among customers and providers, technical information on service functionality such as application-, middleware-, or network-level QoS, and legal data that make the contracts valid such as partners' names and signatures.
----图片来自文件Michael P. Papazoglou et.al. Service-Oriented Computing: State of the Art and Research Challenges
其中主要功能部分：
1、Service publishing
Service publishing focuses on how providers can advertise their services and make them easily accessible.
2、Service description
Service description deals with machine-understandable ways to associate precise syntax and semantics to services and approach to automated service discovery.
3、Service discovery
Service discovery considers how customers can find services able to full their needs.
4、Service matching
In particular, service matching studies methods for meeting
functionalities required by a client with functionalities provided by available web services.
5、QoS-driven service selection
QoS-driven service selection aims at selecting services
with the best quality among those able to satisfy the client's needs.
6、Service composition
Service composition addresses the situation when there is no single service able to satisfy a client's goal, but the goal can be accomplished by combining a set of available services.
7、Service negotiation
Research on service negotiation targets at providing means for negotiating client requirements and service usage conditions.
For example, service providers may want their services to be securely delivered only to authorized clients while customers may care that their private information is not divulged.
8、Service monitioring
QoS negotiation and runtime service monitoring represent means for tackling this problem and maintaining a certain level of QoS.
9、Service-Oriented Systems Engineering (SOSE)
Service-Oriented Systems Engineering (SOSE) [145] strives for finding best practices and creating exible and reusable services as well as methods for avoiding bottlenecks in SOSs and making them reliable.
10、Service coordination
Service coordination refers to management of dependencies among service activities.
11、Service management
Service management deals with evolving conditions, realization of environment controls and implementation of self-healing and self-reconfigurable SOSs.
下面重点研究第五章：Quality-driven Web Service Selection
5.1 Background
首先给出了Quality of Service的定义：
Quality of Service as “a set of perceivable characteristics expressed in user-friendly language with quantifiable parameters that may be subjective or objective”.
可以把Quality of Service 分为两种类型：
1、Quantitative：
which can be precisely expressed in numerical values,如response time or execution cost
2、Qualitative：
which cannot be defined through a natural metric and usually refiect common views and opinions about the service.如：service reputation or privacy
根据Quality of Service的来源可以把Quality of Service 分为三种类型：
1、Real quality（实际的服务质量）
refers to the actual quality of a service component
2、Advertised quality（声明的服务质量）
are service properties published by their providers in public registries or directly shared with service consumers.
3、Observed quality（观察的服务质量）
the quality of a delivered service, or, in other words, the quality perceived on the client side.
Observed QoS parameters may significantly difier from actual or advertised parameters since they depend on the context information such as input data or network layout.
观察的服务质量由于输入数据或者网络延迟等上下文信息的原因很容易偏离实际的和声明的服务质量
For enabling efficient QoS management, the following questions are of key importance:
- How providers specify and advertise QoS;
- How QoS of delivered web services is measured;
- How clients can select good services in terms of QoS;
- How clients can predict quality of SOSs given QoS parameters of candidate web services.
5.1.1 How providers specify and advertise QoS(略)
5.1.2 Measuring Quality of Web Services :
1. collected automatically by runtime service monitors.
主要方法：Statistical data（统计）
缺点：（1）overhead on the SOA；
（2）unable to measure qualitative parameters；
2. assessed using client feedback.
缺点：（1）heterogeneous data difficult to process and reason about
（2）client not provide feedback
3. evaluated by specialized unbiased agencies.
方法：Independent organizations can exhaustively test available web services and assure prospect clients that these services behave correctly.
学术方法（略）
5.1.3 Predicting Quality of SOSs
方法1：Majithia et al. [100] advocate a schema for calculating reputation scores of composite web services.
-------S. Majithia, A.S. Ali, O.F. Rana, and D.W. Walker. Reputation-based semantic service discovery. 2004
The schema accounts for the application context and the time of computing the reputation of
involved web services.（没有阅读）
方法2：Cardoso et al. [32] introduce a set of models for QoS measurement in service-based workflows.
----------J. Cardoso, A. Sheth, J. Miller, J. Arnold, and K. Kochut. Quality of service for workflows and web service processes. Journal of Web Semantics, 1(3):281-308, 2004.
In particular, the authors evaluate expected response time, execution cost and reliability of a workflow applying sequential, parallel, conditional, loop and fault-tolerant reduction rules.
方法3：Lakhal et al. [91] extends this work by reviewing the evaluation of reliability and response time for fault-tolerant service compositions.
-----------N. Ben Lakhal, T. Kobayashi, and H. Yokota. A failure-aware model for estimating the efficiency of web service compositions. 2005.
Fault-tolerance in this context presumes a composition's resistance to individual service failures - if some service fails, another functionally equivalent web service is invoked to complete the task.
方法4：Jaeger and Ladner [68] consider identification of composition weak points.
-----------M.C. Jaeger and H. Ladner. Improving the qos of ws compositions based on redundant services.
For each weak point they identify alternative candidates that meet the functional requirements.
Three possible replacement patterns are proposed.
（1）In the first pattern, the original service is replaced by an alternative service.
（2）In the second pattern, a structure containing the alternative candidates in the AND split with 1-out-of-n-join pattern is involved instead of the original service.
（3）In the third pattern, structure only one of the available alternative candidates is invoked.
(注：Dr.-Ing. Michael C. Jaeger 个人主页：http://user.cs.tu-berlin.de/~michi/
Doctoral Thesis：Optimising Quality-of-Service for the Composition of Electronic Service 2007）
方法5：Dealing with the analogous problem, Stein et al. [140] apply an algorithm for defining workflows which achieve higher success probability in uncertain environments through varying the number
of services provisioned for each task.
--------[140] S. Stein, N.R. Gennings, and T.R. Payne. Flexible provisioning of service workflows.2006
（注： S. Stein个人主页：http://www.ecs.soton.ac.uk/people/ss2/，
Stein, S. (2008) Flexible Service Provisioning in Multi-Agent Systems. PhD thesis, University of Southampton. ）
5.1.4 Quality-driven Web Service Selection
方法1：Zeng et al（one of first approaches to the runtime service selection）consider QoS-driven web service selection as a global optimization problem and apply linear programming to determine the optimal service composition.
The target function is defined as a linear combination of five QoS parameters: availability, successful execution rate,response time, execution cost and service reputation. If global QoS characteristics such as total composition cost are not constrained, the optimal solution to this problem can be found using the Dijkstra algorithm on a graph of available compositions
-------L. Zeng, B. Benatallah, A.H.H. Ngu, M. Dumas, J. Kalagnanam, and H. Chang. QoS-aware middleware for web services composition. 2004
方法2：Ardagna and Pernici model service selection as a mixed integer linear programming problem and global constraints where both local are specified.
----D. Ardagna and B. Pernici. Global and local QoS constraints guarantee in web service selection
方法3：Yu and Lin propose a model where web service selection is mapped to the multi-choice multi-dimension 0-1 knapsack problem.
In this approach, the practice of offering different QoS levels is taken into account.
-----T. Yu and K.J. Lin. Service selection algorithms for composing complex services with multiple QoS constraints. 2005
方法4：Gao et al. apply integer programming to dynamic web service selection in the presence of inter service dependencies and conficts
上述方法的缺点：
The above solutions depend strongly on weights assigned to each QoS parameter by prospective clients.However, there is no clear mechanism for a client to set up these weights in order to obtain a desired result. Several approaches try to avoid a client involvement into service selection procedure.
改进方法：
方法1：Hu et al. [66] model service selection as the multiple-attribute decision making problem. Four models for determining relative weights of QoS attributes are proposed: subjective, single, objective and subjective-objective.
方法2：Claro et al. [36] extends the QoS-driven selection algorithm proposed by Zeng et al. [163]. The first extension concerns the concept of reputation which is ranked based on the client's knowledge about the ser vice domain. As the second extension, the authors apply a multi-objective genetic algorithm to select the optimal composition without giving any weight to any QoS criterion.
方法3：Canfora et al. [30] also propose to use genetic programming for QoS-aware workflow replanning.
其它：An interesting approach to dynamic web service selection was studied by Martin-Diaz et al. [104]. The approach applies constraint programming for procuring web services with temporal-aware demands and offers. Bonatti and Festa [21] consider three instances of service selection problem which aim at optimizing the overall quality of mapping between multiple requests and multiple offers. Mou et al. [110] experiment with expressing client's QoS preferences in a fuzzy way and specify the corresponding choice-making process.
主要存在的问题：
Firstly, service selection models are oversimplified.
most of the service selection algorithms choose one service for one task while it is recognized that redundant service compositions exhibit better fault-tolerance,
Secondly, existing QoS management and service selection techniques do not suffciently consider functional relations among conflicting QoS characteristics and simply optimize their linear combination.
Thirdly, customers often are asked to rate QoS parameters being unaware of how their preferences are treated.
Finally, there are no service selection algorithms which would consider web services as long-living components of SOSs and optimize direct or indirect profit from multiple composition executions (e.g., total number of successfully executed requests) rather than every single execution.
