A deep web query interface discovery method

For the purpose of obtaining deep web query interface from forms accurately, this paper proposes a framework of automatic deep web discovery, which includes procedures of collecting web pages, extracting forms and features, filtering forms, and identifying forms. A heuristic rule-based k-nearest neighbor algorithm for identifying the query interfaces is introduced. In the experiments, a number of query interfaces and non-query interfaces from different domains are selected for classifying the query interfaces. Experimental results demonstrate that the presented algorithm can significantly improve the accuracy of deep web query interface discovery.