On the Effectiveness of Flexible Querying Heuristics for XML Data

The ability to perform effective XML data retrieval in the absence of schema knowledge has recently received considerable attention. The majority of relevant proposals employs heuristics that identify groups of meaningfully related nodes using information extracted from the input data. These heuristics are employed to effectively prune the search space of all possible node combinations and their popularity is evident by the large number of such heuristics and the systems that use them. However, a comprehensive study detailing the relative merits of these heuristics has not been performed thus far. One of the challenges in performing this study is the fact that these techniques have been proposed within different and not directly comparable contexts. In this paper, we attempt to fill this gap. In particular, we first abstract the common selection problem that is tackled by the relatedness heuristics and show how each heuristic addresses this problem. We then identify data categories where the assumptions made by each heuristic are valid and draw insights on their possible effectiveness. Our findings can help systems implementors understand the strengths and weaknesses of each heuristic and provide simple guidelines for the applicability of each one.