This description provides tools and techniques for automatically extracting data from semi-structured documents. A computer-readable storage medium may contain computer-executable instructions that, when executed by a computer, cause the computer to receive a request for data representing an inferred structure of an input document. For the request, the computer may determine whether a repository containing mined information includes the requested data. If the repository contains the requested data, the computer may return the data representing the inferred structure of the input document in response to the request.