A DOM-based Anchor-Hop-T Method for Web Application Information Extraction

摘要

In order to implement the information fusion of electronic products, the widely adopted approach is to extract information from HTML structure of business Website with deeply data processing. However, modeling Web application is hard to be solved that the data in HTML is semi-formal which displayed as DOM (Document Object Model) tree when using XML schema to data analysis. How to understand and extract information is first to be researched. The general model Anchor-Hop considering the text property and label property is simple to handle this problem. Therefore, it has low effectiveness. This model is sensitive to the data of HTML structure, that if the website structure is slightly changed the issue of extraction accuracy is encountered. As a result, the extraction rules should be redefined because of the changed structure. In order to improve extraction efficiency, this paper proposed a DOM-based dynamic model Anchor-Hop-T information extraction model. The HTML tags including table, ol and ul can be searched and processed using XPath so that it is convenience to extract corresponding Anchor data block. Furthermore, the location of Hop point is considered as invariant, by which our new model based on Anchor and Hop point introduces more concepts for extracting information, such as Anchor data block, Anchor locating library and AH relevance value. Finally, we try to give out an experiment to demonstrate the applicability of our approach.