Intelligent Web Service

In this article, we show that, in order to improve customer satisfaction, we need to incorporate communication modes (e.g., speech act) in the current standards of web services specifications. We show that with the communication modes, we can estimate various affects on service consumers during their interactions with web services. With this information, a web-service management system can automatically prevent and compensate potential negative affects, and even take advantage of positive affects.

We discuss an important factor with regard to creating a successful web services user experience that has been largely ignored. We propose that, in order to counter (detect, prevent, and resolve) negative affects that can be caused by web services, we need to reconsider some of web standards (e.g., WSDL) to include information about communication modes of interfaces (e.g., speech acts).

Communication modes are important information for checking whether or not web services are well-behaved. A web service violating the well-behavedness can cause negative affects on its users. For instance, when a directive operation is invoked by a user (e.g., ‘add item A to my shopping cart’), it is a common knowledge that the user is expecting an acknowledgement within a certain time frame. A violation of this common knowledge can cause negative affects on the users.

Despite great deal of efforts on semantic web, this factor has been largely ignored. Thus, currently a large amount of development time is spent on making sure that web services behave as intended. For example, in most e-commerce environments, the following problems frequently arise:

1. Customers often feel ignored or uncertain because prospective events, such as delivery notices, are not informed properly.

2. For customer services personals, it is hard to feel how customers might be affected by the overall processes. Therefore, it is difficult for them to provide more adaptive and reasonable services, and thus differences in customers’ situations are often ignored.

3. Promises are often not full-filled. E-commerce web-sites often promise customers for certain behaviors of their services in their web site, such as promotions, but actually certain behaviors do not meet such information.

We propose that web services can be guaranteed to meet certain acceptable quality by incorporating affective computing, and thus avoiding these common problems.

Affective computing is not a new concept. In designing web applications (e.g., e-commerce applications) that carry out a certain set of goals for human users, such as purchase orders, the importance of the affects of such applications has already been put forward to system designers’ attention. Therefore, there has been much research into evaluating human emotions, expressing emotions, and the effectiveness of such approaches to improving human-computer interactions. Business communities also have been aware of the significance of customer satisfaction in measuring business performances (e.g., American Consumer Satisfaction Index) because for most companies, repeat customers are major contributors to profit. Preventing negative affects on computer users is also one of the primary goals of HCI communities.

However, there are still no well defined languages to represent or account for affects of web services on human users. Current modelling approaches in web service design primarily focus on domain modelling, performance, and reliability of applications ignoring various affects of services on human users. As a result, current web service design approaches do not have means for representing and estimating affects on users. Thus, it is impossible for web service management systems to prevent possible negative affects or to take advantage of positive affects.