Semantic-based approach to improve the description and the discovery of Linguistic Web Services
Introduction
Nowadays, the Web services have become one of the most relevant research topics in the software engineering field. As the number of Web Services increases, the issue of selecting a desired service(s) becomes a challenging research topic. Initially, Web services were described using WSDL (Web Service Description Language). However, the lack of semantics in WSDL prevents an automatic discovery of Web services (Papazoglou et al., 2007). In order to enhance the description of the Web services, several languages and approaches, such as OWL-S (W3C, 2004), WSMO (W3C, 2004), and SAWSDL (W3C, 2007), have been proposed using semantic models (e.g., ontology).
The Linguistic Web Services (LingWS for short) are a kind of Web Services related to the linguistic information system (e.g., Part of speech Tagger, Tokenizer, Morphological analyzer). Such services are used to compose other LingWS corresponding to well-known applications such as Text Summarization, Machine Translation, and Information Retrieval (Bramantoro, March 2011).
Considering the richness of the linguistic knowledge, researchers in the Natural Language Processing (NLP for short) field have proposed many attempts to improve the description of LingWS using semantic approaches (Klein and Potter, 2004) or semantic wrappers (Ishida, 2006). Nevertheless, they did not offer the possibility to represent all the linguistic features. In fact, the LingWS׳s are characterized by several features called nonfunctional linguistic properties which are already discussed in previous works (Baklouti et al., 2012a, Baklouti et al., 2012b). The LingWS description should cover all these properties and their inter-relationships like the treatment type (analysis and/or generation), the used formalism (e.g., contextual grammar, unification grammar), the processing level (e.g., morphological, syntactic, semantic), and so on. Unfortunately, the existing semantic approaches are unable to represent this kind of properties and their relations.
As far as the LingWS discovery is concerned, Bramantoro and Ishida (2011) proposed a new technique to measure the semantic similarity between LingWS descriptions through specified concepts in an ontology. They used a domain ontology previously proposed by Hayashi (2007). However, their proposal considers only the LingWS Inputs/Outputs (I/O for short) data type which is not the unique aspect that fully characterizes a LingWS (Hayashi and Narawa, 2012). Concerning the semantic matchmakers, there are many developed tools which ensure the matching of semantic Web services but there are no tools for discovering LingWS. Therefore, it is worth suggesting an appropriate matchmaker that mainly considers the nonfunctional linguistic properties.
In this paper, we aimed to enhance the LingWS description by integrating the nonfunctional linguistic properties and their relationships while annotating both functional and nonfunctional properties using a linguistic domain ontology.
Furthermore, we sought to propose an appropriate matching algorithm in order to improve the LingWS discovery. The description and discovery of LingWS would be consolidated by implementing some appropriate tools.
As a result, the integration of the nonfunctional linguistic properties and their relationships inside the description of LingWS helped the lingware system developer to explore the required properties, notably the linguistic ones. Moreover, the evaluation of the proposed matching algorithm using the SME2 environment has improved the discovery of LingWS. The obtained discovery results showed that the proposed approach can be applied for industrial application where the LingWS are numerous and various. In this context, the highly expressive descriptions offering more selecting features ensure the efficiency of the discovery process.
The remaining of this paper is organized as follows: Section 2 shows the nonfunctional linguistic properties. A comparative study between semantic approaches is provided in Section 3. Section 4, however, presents the proposed solution to enhance the description and discovery of LingWS. In Section 5, we depict the proposed solution to integrate and annotate the description elements. Then, Section 6 focuses on the LingWS discovery. Details on the implementation of the description and matching tools are discussed in Section 7. Section 8 presents the comparison experiments and analysis of the results. In Section 9, we briefly comment on some related works before drawing our conclusion in Section 10.
Section snippets
Nonfunctional linguistic properties
In general, knowledge items could be considered to describe a lingware system. In the context of LingWS, these knowledge items cover not only the service name and the functional properties but also some nonfunctional properties which are very useful to both describe and discover LingWS. These nonfunctional properties can be classified according to the processing level in the following way:
- •
The Lexical Level: It is characterized by the use of lexical linguistic resources, approaches, formalisms,
Semantic approaches
To deal with the description of the Web Service issues, the software engineering domain provides several approaches that use semantic models (e.g., ontologies) to describe services. Among the well-known approaches, we can mention OWL-S (W3C, 2004), WSMO (W3C, 2004), and SAWSDL (W3C, 2007).
- •
OWL-S: The OWL for Services proposes an ontology of services. It provides three essential types of knowledge about a Web Service: The first is the Profile which is used to advertise the service. The service
Approach overview
The proposed approach is shown in Fig. 1. It consists of two layers: the Description Layer that manages the description of LingWS by adding the nonfunctional linguistic properties inside the OWL-S ontology and using the linguistic domain ontology to semantically annotate both the functional and nonfunctional properties. The second is the Discovery Layer that provides an extension of the best OWL-S matchmaker to consider the nonfunctional linguistic properties in its matching algorithm. The
OWL-S extension
The aim of this section is to extend the OWL-S ontology using nonfunctional linguistic properties such as approaches, phenomena, resources, and formalisms that are relevant features in the linguistic field. It is known that the OWL-S can represent this kind of properties but it cannot: (I) classify the nonfunctional linguistic properties by category. For example, using the OWL-S model and for one LingWS, we cannot get both the processing level and its linguistic properties, such as approaches,
The OWLS-MX matchmaker
We have chosen the OWLS-MX matchmaker for many reasons. Indeed, it is available as an open source from the portal semwebcentral.org.7 In addition, it has been successfully used in many fields, such as mobile e-health systems for emergency medical assistance and repatriation planning, namely the Health-SCALLOPS system8 and the CASCOM system9 (Klusch et al., 2009). This matchmaker offers a
Implementation details
At this point, we present some implementation details about the developed tools to describe and discover the LingWS.
Evaluation environment
The Semantic Web Service Matchmaker Evaluation Environment (SME2 for short) (Klusch, 2009) is widely used to evaluate matchmakers for Semantic Web services over given test collections in terms of standard performance evaluation measures (Klusch et al., 2010). The performance of our proposal (i.e., OWL-LingS-MX) is evaluated using this environment. We have implemented an SME2 plug-in in order to evaluate it within an SME2 benchmark. To perform this evaluation, we have used our OWL-LingS editor (
Related work
To enhance the LingWS description, Klein and Potter (2004) used OWL-S approach to describe LingWS. However, this contribution proposed an annotation of I/O and ignored the nonfunctional linguistic properties (e.g., processing level, approach, and phenomenon) which are mandatory to know how the LingWS operates.
Later, Ishida (2006) proposed a wrapper around LingWS that represents the LingWS profile containing the LingWS name, type, a textual description, LingWS status, and so on. However, this
Conclusion and future work
This paper provided a solution to the problems related to the lack of semantic knowledge within the LingWS description. Indeed, we proposed an OWL-S extension to integrate the nonfunctional linguistic properties and relations between them. Besides, we developed a linguistic domain ontology using ISO standards. To ensure the LingWS description, we implemented a tool, called OWL-LingS Editor, that supports the proposed extension of OWL-S. This editor is available as an open source software at //www.redcad.org/members/nabil.baklouti/OWL-LingS-Editor
References (21)
- et al.
Owls-mx: a hybrid semantic web service matchmaker for owl-s services
J. Web Seman.
(2009) - Baklouti, N., Bouaziz, S., Gargouri, B., Aloulou, C., Jmaiel, M., 2010. Towards the reuse of lingware systems: a...
- Baklouti, N., Gargouri, B., Jmaiel, M., 2012a. Enhancing linguistic web service description with non-functional nlp...
- Baklouti, N., Gargouri, B., Jmaiel, M., 2012b. An ontology-based approach for linguistic web service description. In:...
- Baklouti, N., Fakhfakh, F., Gargouri, B., Jmaiel, M., 2013. Owl-lings editor: a tool for semantic description of...
- Bramantoro, A., 2011. Composing and Organizing Language Services (Ph.D. thesis). Ishida & Matsubara Laboratory, Kyoto...
- Bramantoro, A., Ishida, T., 2011. Cultural language service: a discovery, composition and organization. In: IEEE SCC,...
- Hayashi, Y., 2007. Conceptual framework of an upper ontology for describing linguistic services. In: IWIC, pp....
- Hayashi, Y., 2011. Prospects for an ontology grounded language service infrastructure. In: Proceedings of Workshop on...
- Hayashi, Y., Narawa, C., 2012. Classifying standard linguistic processing functionalities based on fundamental data...
Cited by (10)
A semantic matching approach addressing multidimensional representations for web service discovery
2022, Expert Systems with ApplicationsCitation Excerpt :However, inevitable keyword-related issues such as the keyword polysemy problem and insufficient keyword provision within syntactic discovery are crucial factors that restrict and degrade the accuracy of service discovery. Semantic discovery approaches attempt to overcome these issues by incorporating the power of ontologies that provide semantics for obtaining formalized service descriptions (Baklouti et al., 2015). However, as the number of web services continues to grow, many services will provide the same functionality, making it extremely difficult to capture accurate user functional requirements for semantic discovery approaches (Zhang et al., 2018).
A Semantic approach for constructing valid composition scenarios of linguistic Web Services
2018, Procedia Computer ScienceSemantic web services for ROS: a Robot as a Service approach
2022, Automated Software EngineeringA context-based service matching approach towards functional reliability for industrial systems
2019, Enterprise Information Systems