Research On A Deep Web Uncertainty Probability Model

Deep Web contains vast amounts of hith-quality information. It’s strong in the field ofinformation and related to the fields are very extensive. The amounts of information of DeepWeb grow rapid every year. It’s becoming primary means that obtain information of theinterest of network information through the Deep Web. Deep Web contains a wealth of data,but the characheristrics of wide range distribution, heterogeneous, autonomous, and so onaffect the access to information. Inorder to utilize the Deep Web information effectively that ithas become the main method to use Deep Web through integrated a unified query to obtaininteresting information of Deep Web.Uncertainty of Deep Web occurred in several stages of the integration process. First ofall,the data of Deep Web maybe uncertainty in Deep Web integration, with the generation of avariety of emerging applications, the data integration faced is no longer the traditional sensedata. The data of new applications (such as sensor data), and the data extracted from theunstructured data themselve may be inaccurate or uncertain; Second, the mappings betweenthe mediated schema of the integrated system and data sources are also uncertainty; Finally,the query response of the system is uncertainty too. Sometimes the queries are uncertaintycoming from users submitted, such as keyword queries and the unclear description.This paper presents a Deep Web uncertainty probability model (DUPM)that make a validdescription of the problem of uncertainty in the integration peocess. Deep Web involved in thevarious fields. Deep Web data sources need to be classified by fields to provide the systemwith the “homogeneous” data sources. When the query interfaces integrated, the first thing isto make a screening of tag attributes in the query interfaces, integrating those “important”attribute, then calculate the similarity between the properties according to the similarity rules.The mediated schema is build by constructing attribute cluster collections and uncertaintyattributes to cluster a collection. And the probability will assign to the mediate schema withthe present of the consistency and cooccurrence properties. Finally, the specific process of thequery processing with uncertainty is shown.