The maintainability plays an important role in controlling quality of web services. Maintaining a software project depends on the degree of complexity, which can be estimated by metrics. Web Service Description Language (WSDL) documents are used for describing Web services. The complexity of web services can be estimated by analysing its WSDL document. The WSDL provide the description of the web service to the service requestors. However, the WSDL does not contain implementation details information of a web service, hence, one can only estimate the data complexity of a web service. The data complexity can be characterised by an effort required to understand the structures of the messages that are responsible for exchanging and conveying the data. Furthermore, the complexity of web service can be computed by the metrics which analyse the structures of the messages.
In this work we present a metric to compute the complexity of web services data weight (DW) of the WSDL. DW is defined as the sum of the data complexity of each input and output messages. Further, the data complexity of a message can be computed by analyzing the message structures with the arguments that the operations of a Web service take. To prove the value of metric, it is evaluated and validated both theoretically and empirically. The theoretical evaluation is performed by the Weyuker's properties based on the measurement theory. The practical utility of the metric is evaluated by the Kaner's framework, which consists of several questions to evaluate the practical usefulness and scientific base of the metric. The most important validation of this metric is the empirical validation. The data weight of WSDL is empirically validated by applying it on more than 50 WSDL real files available on the web. The metric is also compared with similar metrics to prove its value. The empirical, theoretical and practical validation and comparative study proved that this data weight metric is a very good indicator for estimating the quality of web services. The experimentations proved that if the data weight metric's value increases, the quality of web services will decrease, because increasing value of data weight implies inefficient use of memory and time.