Similarity measures for incomplete database instances

The problem of comparing database instances with incompleteness is prevalent in applications such as analyzing how a dataset has evolved over time (e.g., data versioning), evaluating data cleaning solutions (e.g., compare an instance produced by a data repair algorithm against a gold standard), or comparing solutions generated by data exchange systems (e.g., universal vs core solutions). In this work, we propose a framework for computing similarity of instances with labeled nulls, even of those without primary keys. As a side-effect, the similarity score computation returns a mapping between the instances’ tuples, which explains the score. We demonstrate that computing the similarity of two incomplete instances is NP-hard in the instance size in general. To be able to compare instances of realistic size, we present an approximate PTIME algorithm for instance comparison. Experimental results demonstrate that the approximate algorithm is up to three orders of magnitude faster than an exact algorithm for the computation of the similarity score, while the difference between approximate and exact scores is always smaller than 1%.