Relationships in Structured Text Retrieval

SYNONYM None DEFINITION In structured text retrieval, the relationship between text components may be used in ranking components relative to a given query. MAIN TEXT In a structured text document, there exists a relationship between the document components. In the context of XML retrieval, the relationships between elements are provided by the logical structure of the XML markup. An element, unless it is the root element (the document itself), has a parent element, which itself may have a parent element. Similarly, non-leaf elements have children elements, and so on. Considering relationships between elements appears to be beneficial for XML retrieval. For instance, in a collection of scientific articles, it is reasonable to assume that the " abstract " of an article is a better indicator of what the article is about than a " future work " section in the same article. The challenge in XML retrieval is what types of relationship should be considered, and how this information can be used to score elements according to how relevant they are for a given query. In the contextualization approach [1], considering the " root element – element " relationship to rank an element in addition to the element own content has shown to improve retrieval effectiveness [2,3].