Identification of Deep Web Entries by Using Neural Network

摘要: Deep web is the fastest-growing new resource on Internet. The establishment of its data integration system has become a research focus. deep entries, with automatic identification as basis integration, usually appears in HTML forms. Owing to subjectivity form design, lack unified construction standards makes it difficult judge whether or not entry by heuristics and manually specified rules. Based global schema notion machine learning, this paper proposes an approach identify entries using neural network. Through statistic abundant forms data, provides 14 features distinguish query interface from non-query interface. Experiments 12 sets show higher accuracy our proposed use thus recommended.