| Research Abstract |
1. In usual databases, it is assumed that the stored data must be in a fixed format for the underlying database schema that is designed in advance. In semi-structured data, on the other hand, the data (that would have a common concept) may not have the same structure, so that it is difficult to design a fixed database schema in advance. In this project, we focus on types of properties, introduce a constraint that can expresses a type information including ambiguity of semi-structured data, and aim to solve a number of basic problems on such a constraint.
2. We extend the constraint called SC (specialization constraint) that express a type of a property so as to be able to express a constraint which a semi-structured data should satisfy. By the extension, an SC may have a set of path expressions.
3. For the inference problem on SCs (which decides if an SC σ is a logical consequence of a set of SCs Σ), we obtain a set of inference rules that is sound and complete.
4. We obtain a sufficient condition such that the inference problem is partially decidable. Similarly, for the co-problem (that is, negation) of the inference problem, we obtain a sufficient condition such that the co-problem is partially decidable.
5. By the result above, if the set of path expressions in an SC is context-free, then the inference problem is undecidable. If the set of path expressions in an SC is regular, then the inference problem is decidable.
6. As for the navigation problem of a path expression (which decides if there is a database that contains a path which we can navigate along the given path expression), we obtain a similar result as above.
|
