Web applications have become an invaluable source of information for many different vertical solutions, but their complex navigation and semistructured format make their information difficult to retrieve. Web Automation and Extraction systems are able to navigate through web links and to fill web forms automatically in order to get information not directly accessible by a URL. In these systems, the main optimization parameter is the time required to navigate through the intermediate pages which lead to the desired final pages. This paper proposes a series of techniques and algorithms that improves this parameter by basically storing historical information from previous queries, and using it to make the browser manager preload an adequate subset of the whole navigational sequence on a specific browser, before the following query is executed. These techniques also handle which sequences are the most common, thus being the ones which are preloaded more often.
展开▼
