<XML><RECORDS><RECORD><REFERENCE_TYPE>0</REFERENCE_TYPE><REFNUM>9029</REFNUM><AUTHORS><AUTHOR>Ashoori,E.</AUTHOR><AUTHOR>Tsikrika,T.</AUTHOR><AUTHOR>Lalmas,M.</AUTHOR></AUTHORS><YEAR>2007</YEAR><TITLE>Examining Topic Shifts in Content-Oriented XML Retrieval</TITLE><PLACE_PUBLISHED>International Journal on Digital Libraries, 8(1)</PLACE_PUBLISHED><PUBLISHER>Springer</PUBLISHER><PAGES>39-60</PAGES><LABEL>Ashoori:2007:9029</LABEL><KEYWORDS><KEYWORD>content-oriented XML retrieval</KEYWORD></KEYWORDS<ABSTRACT>Content-oriented XML retrieval systems support access to XML repositories by retrieving, in response to user queries, XML document components (XML elements) instead of whole documents. The retrieved XML elements should not only contain information relevant to the query, but also provide the right level of granularity. In INEX, the INitiative for the Evaluation of XML Retrieval, a relevant element is de?ned to be at the right level of granularity if it is exhaustive and speci?c to the query. Speci?city was speci?cally introduced to capture how focused an element is on the query (i.e., discusses no other irrelevant topics). To score XML elements according to how exhaustive and speci?c they are given a query, the content and logical structure of XML documents have been widely used. One source of evidence that has led to promising results with respect to retrieval e?ectiveness is element length. This work aims at examining a new source of evidence deriving from the semantic decomposition of XML documents. We consider that XML documents can be semantically decomposed through the application of a topic segmentation algorithm. Using the semantic decomposition and the logical structure of XML documents, we propose a new source of evidence, the number of topic shifts in an element, to re?ect its relevance and more particularly its speci?city. This paper has three research objectives. Firstly, we investigate the characteristics of XML elements re?ected by their number of topic shifts. Secondly, we compare topic shifts to element length, by incorporating each of them as a feature in a retrieval setting and examining their e?ects in estimating the relevance of XML elements given a query. Finally, we use the number of topic shifts as evidence for capturing speci?city to provide a focused access to XML repositories.</ABSTRACT></RECORD></RECORDS></XML>