Elsevier

Applied Soft Computing

Volume 41, April 2016, Pages 51-65
Applied Soft Computing

A fuzzy framework for efficient user-centric Web service selection

https://doi.org/10.1016/j.asoc.2015.12.011Get rights and content

Highlights

  • This work falls within Web service query optimization and make use of explicit/implicit preference.

  • An appropriate inference mechanism, borrowed from the fuzzy/approximate reasoning field, is used.

  • A priority-based method is introduced to aggregate the elementary similarities.

  • The top-k results are provided with user.

  • A set of experiments is done to show the feasibility and the effectiveness of our proposal.

Abstract

With the development of Web technologies and the increasing usage of Internet, more and more Web Services (WS) are deployed over Internet. Therefore, there will be a large number of candidate services for fulfilling a desired task. In the last decade, several WS selection approaches are proposed to cope with this challenge. In sharp contrast to the existing WS selection approaches that focus only on user-specified preferences, in this paper, we propose a flexible and effective WS selection framework, which gives users an adequate way to express their preferences using linguistic terms, and enhance the WS selection by leveraging their contexts and profiles. The satisfaction of the candidate WS is expressed by an objective score that takes into consideration no only the user-specified preferences, but also additional preferences extracted from both his/her context and profile using fuzzy inference rules, so as to improve the effectiveness of the selection. We then introduce an effective strategy that allows for priority between the two kinds of preferences, for ranking candidate services. Experimental evaluation on a real case study demonstrates the effectiveness of our proposed strategy.

Introduction

Nowadays, vast repositories that contain huge amounts of WS1 on all matters of interest, are available. WS play an increasingly important role in enhancing the user interaction in the Web and enterprise search. Also, the large corporations are now founding their business on an abundant use of WS, the number of publicly available services is then envisioned to be increased in the future in a fast rate [1]. Moreover, the service requestors are often faced with a large mass of competing WS that offer “similar” functionalities but they are associated with “different” constraints, and they are needed to select the best ones with required functionalities and the highest desired quality. Finding optimal Web services – among a collection of services – is known as the Web service query optimization problem in the community of service computing. A great attention has been paid to this problem in the last decade, see for instance [2], [3], [4].

On the other hand, the user preferences play a major role in the customization of the selection process. Service requestors usually have varying preferences for the features depending on their context and their profile they find themselves in, and of course different requestors will also have different preferences. In practice, it is difficult to predict how many services’ properties are available, and additionally the type of preferences that the user must define it. The more the user is able for expressing his/her needs (preferences), the better the expected outcomes quality is.

A good service selection mechanism should not only allow expressing the preferences that the user defines them for each query attribute, but also deducing additional preferences from his/her situation. Augmenting the user's query by such preferences allows for highly improving the result's quality. On the other hand, it is important to help the users to formulate their preferences in human-like language. A more general and crucial approach to model the preferences in this way is based on fuzzy sets theory [5], [6]. Fuzzy sets that are very appropriate for the interpretation of linguistic terms, constitute a convenient way for users to express their preferences. For example, when expressing preferences about the price of a hotel, often the users employ linguistic terms like “rather cheap”, “affordable” and “not too expensive”.

In this paper, the proposed study falls within WS query optimization and resorts to advanced soft computing techniques to provide user with the best WS that really serve his/her complex and diverse needs. First, to better personalize the WS selection we make use of two families of user preferences: (i) explicit preferences and (ii) implicit preferences. The former stands for preferences that are formulated in the initial user query. Such preferences are expressed thanks to gradual predicates modeled by means of fuzzy sets. The latter involves preferences that are inferred from information related to context and profile of the user. An appropriate inference mechanism, borrowed from the fuzzy/approximate reasoning field, is used. Second, to obtain an overall degree of the similarity between the service requestor preferences and the service constraints, a priority-based method is introduced to aggregate the elementary similarities. By this way, one can assign more priority to explicit preferences than implicit preferences. To the best of our knowledge, this is the first time that a WS selection process takes into account both explicit and implicit user preferences.

Our aim is to set up an effective WS selection framework which supports the user in finding the most suitable services and return the top-k WS according to his/her preferences without much effort from him/her, with taking into account both his/her context and profile.

The main challenges involved in selecting the top-k WS are threefold:

  • First, how to model the information related to both the required service and the offered services to select the relevant ones that can contribute to answering the user query.

  • Second, how to efficiently retrieve the most relevant WS that better satisfy both explicit and implicit user's fuzzy preferences.

  • Third, how to rank-order the results services (knowing that the two kind of preferences are not of the same importance w.r.t. end user) and then generate the top-k WS.

The made key contributions in this paper are summarized as follows:

We already tackled the first challenge by proposing in [7] a service model as well as a Contextual Profile (CP) model that contain several dimensions able to describe the most information characterizing required/offered service. In the current work, we have also strengthened the CP model, which has been stored in a hierarchical data structure called: contextual profile ontology with taking into account the gradual nature of the parameters related to CP and preferences expressed in a human-like language.

In this paper, we focus mainly on the second and the third challenges. Thus, the service selection process is accomplished by the integration of three dimensions: user profile, user context and explicit/implicit user preferences into the end user querying process. Explicit Preferences (EP) are provided in the original user query and modeled thanks to the (fuzzy) membership functions. Due to the commensurability property, the elementary satisfaction degrees are aggregated using one of the fuzzy aggregation operators (e.g., min, product operator, etc.). Then, a list of WS satisfying this family of preferences is returned in a discriminated way. Implicit Preferences (IP) are inferred from user's CP and used to expand the original query in order to retrieve most relevant services. To do so, our framework makes the use of an appropriate fuzzy inference mechanism to faithfully capture the IP. Such preferences operate on the list of WS resulting from the processing of EP. Then, an overall score that takes into account not only the EP but also the IP is defined.

Our proposal allows selecting optimal WS in a qualitative (the k WS with the overall scores greater than a given user threshold) or a quantitative (the top-k WS) way. An efficient WS selection algorithm which returns the best k services according to the one of the above points of view, is provided. Note that the overall score is based on the degree of satisfaction of the requester's EP as first priority, and the one of the IP in a second priority. Experimental results, designed to evaluate our WS query optimization method and ranking mechanism, demonstrate significant performance improvements over the traditional optimization approach.

Outline. The remainder of this paper is structured as follows. In Section 2, we provide first a review of existing proposals for WS selection problem and, second we give some basic notions about ontologies and fuzzy ontologies. Section 3 presents a case-study, which is used as a running example throughout the paper. In Section 4, the modeling aspect related to WS, contextual profiles and preferences is addressed in a detailed way. Section 5 discusses the preference query processing in a real case study related to the field of restaurant business and the ranking mechanism as well. Section 6 describes an experimental study to show the effectiveness of our proposal. Finally, Section 7 concludes the paper and provides some research lines for future work.

Section snippets

Related work and background

In this section, we first give an overview of the most closely related work that exist in the literature, then we present some background on ontologies and fuzzy ontologies.

Problem description and illustrative example

Nowadays, most of restaurants own booking WS (called TableOnlineService in the following) whose descriptions are published in services registry6 in the Web. Such services constitute a practical tool to users to make their

Modeling

In this section, we first discuss service modeling, then contextual profile modeling and finally preference modeling.

Contextual profile – aware preference query processing

This section gives an overview of the proposed selection framework and describes the query processing in an explicit way.

Experimental evaluation

In this section, we report a set of experiments conducted to evaluate the outcomes of our approach. The main purpose of this evaluation is to (i) compare the effectiveness of our proposed selection framework (referred to as IPr for Inference Process) with the traditional frameworks that do not use the inference process (referred to as TR for TRaditional), and (ii) subtract the objective values of θ and α, to free users from assigning these parameters.

Note that since there is not any public test

Conclusion and future work

Web services are emerging to provide framework for systematic and extensible business-to-business and business-to-consumer interactions. Improving the capabilities and functionalities of current research engines on the Web, through effective research and selection techniques of services, is becoming increasingly important. Making the WS selection more customized is a critical issue in present day services search. In this paper, we discussed a step to cope with this issue in an efficient way by

References (64)

  • Q. Yu et al.

    Framework for web service query algebra and optimization

    ACM Trans. Web

    (2008)
  • K. Benouaret

    Advanced Techniques for Web Service Query Optimization

    (2012)
  • D. Dubois et al.

    Fundamentals of Fuzzy Sets

    (2000)
  • A. Hadjali et al.

    Database preferences queries – a possibilistic logic approach with symbolic priorities

    J. Ann. Math. Artif. Intell.

    (2012)
  • Z. Chouiref et al.

    Multi matchmaking approach for semantic web services selection based on fuzzy inference

  • H.Q. Yu et al.

    A method for automated web service selection

  • M. Comerio et al.

    WSMOD: a methodology for QOS-based web services design

    Int. J. Web Serv. Res.

    (2007)
  • A.E. Walsh

    UDDI, SOAP, and WSDL: The Web Services Specification Reference Book

    (2002)
  • A. Tsalgatidou et al.

    An overview of standards and related technology in web services

    Distrib. Parallel Databases

    (2002)
  • J. Cardoso

    Semantic Web Services: Theory, Tools, and Applications

    (2007)
  • D. Bianchini et al.

    Flexible semantic-based service matchmaking and discovery

    World Wide Web

    (2008)
  • D. Kourtesis et al.

    Combining SAWSDL, OWL-DL and UDDI for semantically enhanced web service discovery

    Semant. Web: Res. Appl.

    (2008)
  • B. Benatallah et al.

    Request rewriting-based web service discovery

    The Semantic Web – ISWC 2003

    (2003)
  • M. Paolucci et al.

    Semantic matching of web services capabilities

    The Semantic Web ISWC 2002

    (2002)
  • A.N. Ahmed et al.

    Selection of web services by using diversified service rank

    Int. J. Softw. Eng. Appl.

    (2014)
  • Q. Yu et al.

    Multi-attribute optimization in service selection

    World Wide Web

    (2012)
  • Y. Liu et al.

    QOS computation and policing in dynamic web service selection

  • B. Pernici et al.

    A fuzzy service adaptation based on QOS satisfaction

    Advanced Information Systems Engineering

    (2011)
  • D. Mobedpour et al.

    User-centered design of a QOS-based web service selection system

    Serv. Oriented Comput. Appl.

    (2013)
  • H.Q. Yu et al.

    Automated context-aware service selection for collaborative systems

    Advanced Information Systems Engineering

    (2009)
  • B.T. Kumara et al.

    Context aware post-filtering for web service clustering

  • D.-K. Chen et al.

    A location-based context-aware service discovery approach for cycling experience

    Int. J. Ad Hoc Ubiquitous Comput.

    (2014)
  • Cited by (31)

    • A formal approach for matching and ranking trustworthy context-dependent services

      2018, Applied Soft Computing Journal
      Citation Excerpt :

      They failed to rank the suitable services in relation to context information. In [15] authors use fuzzy logic to rank web services. The suggested method allows users to express their preferences and by using fuzzy inference rules some preferences derived from users’ contexts and profiles to improve web services ranking.

    • A Model Driven Approach for Web Service Composition

      2022, ICAASE 2022 - 5th Edition of the International Conference on Advanced Aspects of Software Engineering, Proceedings
    View all citing articles on Scopus
    View full text