ANDES: An approach to embed search services on the Web browser

This article addresses search tasks, which are the most frequent type of tasks users do nowadays on the Web. We suggest that the customization of search services would improve the user's searching experience, and it also might make search tasks more effective and efficient. We analyze user tasks for performing searches and types of search services. Then, we distinguish between primary search tasks (related to the initial user goal) and ancillary search tasks (those that are performed to find additional information necessary to understand how to achieve a primary search task). We also explain different types of search services, including global search services (which has an extensive search scope, possibly the whole Web) and in-site search services (which has a narrow scope of search focused on a specific Web site or data set). We also discuss features embedded into Web browsers to support search tasks and how Web browsers can support extensions allowing users to integrate customized search services. We propose an approach called ANDES that allows users to integrate into the Web browser new search services that are created by customizing search services offered by existing Web sites. By means of Web augmentation techniques, we demonstrate how users can customize the browser to trigger search services to any Web site and integrate the Web page results. We report on two evaluations showing that end-users can specify and use search services following our approach and that the proposed interaction is more convenient in contrast to traditional mechanisms. Beyond the tool used to demonstrate the feasibility of the ANDES approach, this paper aims to discuss new strategies for end-user programming that allow users to customize search services over the Web.