| Deep Web not only has features of frequently visiting, growing fast, covering a wide range of areas, but also contains a large number of worthy information. On the other hand, most of the results returned by dynamic pages are unstructured or semi-structured. It’s worth to study how to efficiently and accurately get the information hidden in the Deep Web, and extract them in a structured form.Deep Web data integration is mainly composed of three modules: query interface discovery and management module, query interface integration module and query result processing module. This paper primarily studies the third module, which is about how to get the query results pages, extract and process the query results. On the basis of investigating the existing approaches about the Deep Web query result processing, it focuses on recognizing the data regions, making use of the visual features, domain-specific features and query keywords to identity the data regions. Also, it studies the query results extracting and attribute labeling, and solves the problem of multiple attribute data nesting.At the end of this paper, the experiments have been done to validate the functionality and the performance of the query result processing system. The experimental results show that the system can dynamically return the query results which have been process in real time, and the designed system has a good query performance and higher precision. |
