In this paper, we presented an approach that can learn transformation rules from examples. These rules are then used to transform the remaining data. In order to tackle the large grammar space, a subgrammar space is introduced to reduce the search space and a search algorithm is adopted to search in these subgrammar spaces for a sequence of rules that is consistent with the examples. Our preliminary results show this approach is very promising. Future work will focus on accelerating the search algorithm and expanding the coverage.
展开▼