3.1 Extracting and parsing

Every Web service description includes the definition of the programming interfaces to be invoked remotely. Figure 3 shows the abstract service interface

Figure 2. Clustering process of Web services.

definition; from this, the elements extracted for similarity calculation are the name

Description Approach Advantage

Corpus-based approach

Taxonomic-based approach

It is a hybrid approach that combines corpusbased statistical methods with knowledge-based taxonomic information

It is a hybrid approach that combines taxonomic-based approach with corpus-based approach

Information content measure (corpus-

Information content measure that uses lexical chains as a context

This is an information content measure that adds topical to local context using a statistical classifier

based)

It is not a syntactic technique, and it is not dependent on global information

It is a syntacticsemantic technique

It is a semantic technique

Based on the tests reported, this combined approach outperforms other computational models

It is a universally applicable similarity measure, independent of domain or form of knowledge representation

It is a semantic and context-based technique

It is a semantic and context-based technique

Address word sense disambiguation by counting overlaps between dictionary definitions

Bio-Inspired Hybrid Algorithm for Web Services Clustering

DOI: http://dx.doi.org/10.5772/intechopen.85200

Path length to the root node from the least common super-concept of two

taxonomy, based on the notion of

It combines a lexical taxonomy structure with corpus statistical

information needed to state the commonality between the two concepts and the information needed

concepts

information

Lin [15] This measure uses the amount of

to describe these terms

This measure states that two lexicalized concepts are semantically close if their synonyms are connected by a path that is not too long, and it is not changing its direction frequently

This measure finds the shortest path length between two concepts, and scales that value by the maximum path length in the is-A hierarchy in

Resnik [13] Evaluate the semantic similarity in a

information content

Measuring the similarity between two concepts is not a new topic. Throughout the last decades, many measures of similarity have been reported using different perspectives: syntactic, semantic, contextual, etc. In this work, we use a set of semantic similarity measurements based on WordNet.<sup>1</sup> Computing similarity

of the operations and their associated input and output parameters.

4. Semantic similarity measures

Summary of similarity measurements.

https://wordnet.princeton.edu/

which they occur

Semantic similarity measure

Lesk [10]; Banerjee and Pedersen [11]

Wu and Palmer [12]

Jian and Conrath [14]

Hirst Onge [16]

Leacock and Chodorow [17]

Table 2.

1

15

Figure 3. Web service interface definition.


#### Table 2.

3. Clustering process

different clusters [6].

3.1 Extracting and parsing

Clustering process of Web services.

Figure 2.

Figure 3.

14

Web service interface definition.

is described in the following subsections (Figure 2).

Advanced Analytics and Artificial Intelligence Applications

Clustering of Web services consist of partitioning the set of Web services in the collection into an appropriate number of clusters based on a similarity measure. Therefore, services in the same cluster are more similar than the services in the

In this section the clustering approach implemented is described. This process has as input a collection of Web services formatted according to Web Service Description Language (WSDL). This collection of Web services is processed utilizing specific parsers to extract the most important data of the service description, which are the method names and input and output parameters. The detailed process

Every Web service description includes the definition of the programming interfaces to be invoked remotely. Figure 3 shows the abstract service interface

Summary of similarity measurements.

definition; from this, the elements extracted for similarity calculation are the name of the operations and their associated input and output parameters.
