Abstract
Without any doubt XML is currently a de facto standard for data representation. Its popularity is given by the fact that it is well-defined, easy-to-use, and, at the same time, sufficiently expressive.
However, most of XML documents are still created without the respective description of their structure, i.e. an XML schema. Since the knowledge of XML schema is crucial for various data processing approaches as well as a standard optimization tool, approaches to (semi-)automatic inference of XML schema become an interesting research problem.
Recently, there have been introduced several approaches that improve particular steps of the inference process. The problem is how to compare such approaches, how to find the optimal ones and how to join them to get the best possible result.
In this paper, we introduce jInfer, a general framework for XML schema inference. It represents an easily extensible tool that enables one to implement, test and compare new modules of the inference process.
Since the compulsory parts of the process, such as parsing of XML data, visualization of automata, transformation of automata to XML schema languages, etc. are implemented, the user can focus purely on the research and the improved aspect of the inference process. We describe not only the framework, but the area of schema inference in general, including related work and open problems.