Islam Elgedawy

A Conceptual Framework for Web Services Semantic Discovery

Current service discovery techniques mainly depend on keyword matching approaches. That could lead to a high service retrieval recall whereas it leads to a low service retrieval precision. To overcome this problem, a service matchmaking process should use the semantics of both the service and the user. Hence, we propose the G + conceptual model that explicitly captures the service/user objectives, scenarios and contexts. Scenarios capture the required snapshots of the service/user behaviour. Based on the G + model, we introduce the PILLAR conceptual model that captures multiple semantic descriptions of the service. The PILLAR concept is also used to semantically model the user request. Different domain ontologies could be used to provide a service/user PILLAR models with the basic vocabulary. Consequently the ontology mapping process becomes fundamental during the matching process. Hence, we propose an ontology meta-structure to facilitate the ontology mapping process. PILLAR explicitly captures the semantics of both the service and the user creating a framework for more precise service semantic discovery process.

A high-level functional matching for semantic web services

Existing service matching techniques such as keyword-based and ontology-based, do not guarantee the correctness of the matching results (i.e. do not guarantee fulfilling user goals). This paper deals with this problem by capturing the high-level functional aspects (namely goals, contexts, and expected external behaviors) for both web services and users in a machine-processable format, then matching these aspects using the proposed functional substitutability matching scheme (FSMS). Based on FSMS, this paper describes a direct matching technique in which a user request is examined against one service description at a time, such that web services match users’ requests when they have substitutable goals, contexts and expected external behaviors. The substitutability semantics between the elements of application domains are captured via the proposed substitutability graphs, which are used during the matching process to mediate between users’ requests and web services’ descriptions. Simulation results show that the proposed matching approach succeeds in retrieving only the correct answers, while keyword-based and ontology-based retrieval techniques could not eliminate the appearance of false negatives and false positives.

Exploring queriability of encrypted and compressed XML data

When large XML documents are shared, documents compression and encryption become simultaneously important for efficient and secure access. Existing approaches for compressing and encrypting large XML documents provide a non-queriable intermediate document representation, creating a need to decompress and decrypt the whole document before any access, which simply wastes computing resources (such as memory, time, and power) especially in mobile environment. To overcome this limitation, this paper proposes a new approach for compressing and encrypting large XML documents while maintaining queriability over the intermediate document representation. The proposed approach separates document structure from its contents using a variation of the Ctree XML indexing approach (known as the Ctree+), then applies context-free lossless encryption and compression techniques over the resulting Ctree+ intermediate representations. Experiments results show that the proposed approach dramatically enhances queries response times when compared with existing approaches. However, the proposed approach provides lesser compression ratios when compared with context-based compression techniques.

A dual-layered model for web services representation and composition

Nowadays more and more companies and organizations implement their business services in the Internet due to the tremendous progress made recently in the field of Web services. It becomes possible to publish, locate and invoke applications across the Web. Thus, the ability to select efficiently and integrate at runtime services located in different sites on the Web is an important issue. In some situations, if no single Web service can satisfy the request of the user, there should be a possibility to combine existing services together in order to meet the user’s request. This paper provides a dual-layered model for web services, where the first model layer captures the high-level functional specifications (namely goals, achievement contexts, and external behaviours), and the second model layer captures the low-level functional specifications (namely interfaces). This model allows the service composition process to be performed on both high-level and low-level specifications. We also introduce the composition operators (both high-level and low-level) to allow composition of virtual services.

Scenario matching using functional substitutability in web services

Existing web services discovery techniques mainly match the services’ descriptions using keyword-based techniques. This leads to a low service discovery precision as the semantics of services and their domains are ignored. To overcome this problem, we propose a matching approach that matches web services based on their goals as well as their expected external behavior. Scenario-networks are used to partially describe the external behavior of web services. A many-to-many semantic-based matching approach for scenario networks is proposed adopting the concept of functional substitutability.

Automatic generation for web services conversations adapters

Web services interact with each other over the Web to accomplish different business objectives, such interactions are known as services conversations, which are specified via the exchanged messages containing the operations to be invoked. Service consumers search the Web for services that fulfil their requirements. Unfortunately, due to services heterogeneities, services consumers usually cannot find the perfect match for their requests, hence they manually create adapters to mediate between chosen services and their requests. Creating such adapters manually becomes an obstacle for achieving high business agility, and definitely increases development efforts and costs. To overcome this problem, this paper proposes an approach for automatically generating conversations adapters. The proposed approach is ontology-based, where the adopted ontology provides the conversion semantics between application domain concepts via a graph data structure known as the concepts substitutability graph. The proposed approach uses such conversion semantics to determine the mappings between different conversation messages, and to generate the suitable concepts convertors, which are later used to build the required adapter. We believe the proposed approach helps in improving business agility and responsiveness, and of course minimizes efforts needed for developing SOA applications.

Rapid Identification Approach for Reusable SOA Assets Using Component Business Maps

Substantial savings from asset reuse result when the right assets are identified in the very early stages of a client engagement.Unfortunately, advanced identification approaches (known by having high precision and recall, such as behavior-based approaches)cannot be adopted in these early stages, because at these early stages, there is no many details nor much understanding about the client functional requirements. On the other hand, unstructured keyword-based identification approaches are known of having low precision and recall. To overcome this problem, we argue that assets descriptions should have explicit information about the business activities realized by the assets. To be able to capture this information in a machine understandable format, this paper proposes a model for describing reusable assets functional scopes using Component Business Maps (CBMs), in which the asset scope is represented as a hierarchy of CBM elements. Adopting this scope model, the paper proposes a rapid identification approach for reusable assets that retrieves assets based on their CBM projections. We believe the proposed approach provides better precision and recall when compared to unstructured keyword-based approaches.

JAMEJAM: A Framework for Automating the Service Discovery Process

The service discovery problem is not trivial, as it requires solutions for many complex sub-problems such as service semantic description, service identification, service composition, service selection, service evaluation, service adaptation and presentation. Currently, companies manually construct their discovery processes in an ad-hoc tightly-coupled manner using different platform-services that separately handle the identification, composition, selection, evaluation, adaptation and presentation sub-problems. However, when users’ requirements change, the already constructed discovery process needs to be manually reconstructed and reevaluated again. This creates a need for an automated approach that allows different users to dynamically construct their discovery processes on the fly. Therefore, we propose JAMEJAM, a framework for service discovery automation. It enables users to create their customizable discovery processes on demand as an executable BPEL process that describes the required matching aspects, matching schemes and matching policies. JAMEJAM realizes such process by dynamically searching for the suitable platform-services in a context-sensitive manner using different types of knowledge (e.g., aspects, services, and matching schemes knowledge), captured via different software ontologies. Experimental results show that JAMEJAM increases the accuracy and the adaptability of the service discovery process.

CRESCENT: A Reliable Framework for Durable Composite Web Services Management

ion to improve fault tolerance. ACM Trans. Comput. Syst.,

A Context-Sensitive Approach for Ontology Mapping Using Concepts Substitution Semantics

Existing approaches for ontology mapping are known of having limited accuracy, as they are basically based on generic schematic relations (such as Isa and Part-of), and ignore the usage context as well as the logic of the involved operation. To overcome this limitation, this paper proposes a novel approach for ontology mapping that uses concepts substitution semantics to resolve concepts incompatibilities in a context-sensitive manner. Experiments results show that the proposed approach is more accurate than existing generic ontology mapping approaches.