Evaluating Queries on Structure with eXtended Access Support Relations

There are a three ommon design de isions taken by todays sear h engines. First, they do not repli ate the data found on the Web. Se ond, they rely on full-text indexes instead. Third, they do not support the querying of do ument stru ture. The main reason for the latter is that HTML's ability to express semanti s with synta ti stru ture is very limited. This is di erent for XML sin e it allows for self-des ribing data. Due to its exibility by inventing arbitrary new element and attribute names, XML allows to en ode semanti s within syntax. The onsequen e is that sear h engines for XML should support the querying of stru ture. In our urrent work on sear h engines for XML data on the Web, we want to keep the rst two design de isions of traditional sear h engines but modify the last one a ording to the new requirements implied by the ne essity to query stru ture. Sin e our sear h engine a epts queries with stru tural information, a full-text index does not sufe any longer. What is needed is a s alable index stru ture that allows to answer queries over the stru ture of XML do uments. One possible index stru ture alled eXtended A ess Support Relation (XASR) is introdu ed. Further, we report on a sear h engine for XML data alled Mumpits. Due to its prototypi al hara ter, we intentionally kept the design and implementation of Mumpits very simple. Its design is entered around a single XASR and its implementation heavily builds on a ommer ial relational database management system.