XSeek A Semantic XML Search Engine Using Keywords ABSTRACT.pdfVIP

XSeek: A Semantic XML Search Engine Using Keywords∗ Ziyang Liu Jeffrey Walker Yi Chen Arizona State University Arizona State University Arizona State University ziyang.liu@ jeffrey.walker@ yi@ ABSTRACT There are two baseline approaches for determining return We present XSeek, a keyword search engine that enables nodes adopted in the existing works. One is to return the users to easily access XML data without the need of learn- subtrees rooted at VLCA nodes [3, 9], named as Subtree ing XPath or XQuery and studying possibly complex data Return. Alternatively, we can return the paths in the XML tree from each VLCA node to its descendants that match an schemas. XSeek addresses a challenge in XML keyword search that has been neglected in the literature: how to input keyword, as described in [1, 4], named as Path Return. determine the desired return information, analogous to in- However, neither approach is effective in identifying return ferring a “return” clause in XQuery. To infer the search information as shown in the following examples. semantics, XSeek recognizes possible entities and attributes Let us look at the sample queries listed in Figure 2 on in the data, differentiates search predicates and return spec- XML data in Figure 1(a). For Q1 (Rockets), it is likely that a user is interested in the information about Rockets. Both ifications in the keywords, and generates meaningful sea