Research On Processing Null Values In Probabilistic Database And Probabilistic Interval Model

With the development of computer and information science, it is believed that the amount of information is increasing sharply. But the situation is different as sometime. The information is uncertain and incomplete at the same time. In 1996, Dev and Sarkalar proposed a new probabilistic data schema and Probabilistic database model to handle such data, but these schema and model have many shortcomings. Thus, we must extend there probabilistic data schema and make the probabilistic database thoery perfect, so that handle the uncertain data and describe the real world.Probabilistic Relational model applies the probability theory to the classic relation model, and has a much complete algebraic system structures. The reality is not always the relational data, so the probability theory also applies to other types data. At present, scholars are most interested in is the probability of semi-structured data model, Dekhtyar et al. proposed a probabilistic semi-structured data management approach, which supports a wealth of algebraic queries based on relation database technology.There uses null value to describe the incomplete data in the probabilistic database, the traditional method of processing null value is shielding it. In this thesis,, the probability interval can translates to the point using the compromise algorithm, value type data can calculate mathematical expectation use one moments operation by the probability distribution. On this basis, this thesis proposes interval probabilistic XML model (IPXML), the model uses the interval probabilities instead of point probabilities, can describe the probabilistic data and incomplete data more preferably, and the answers which satisfy the query will be more abundant.This thesis first introduces the uncertain data, possible world model, probabilistic theory basic, null value theory basic and probabilistic relation model, and described the its operation. Null value in the probabilistic database, it is more complicated, and two kinds of different interpretations by the null probability role, one of which generate probability interval. This thesis address such problems that basic algebra shield the null value in the relational model, and present a compromise algorithm to probability interval, so that the probability interval can be replaced by point probability. Finally, on this basis, this thesis proposed instead of the probability interval for point probability of each node point, establish the probability interval XML model based on the weak instance, declare their semantics, and prove the correctness of the semantics, propose query algorithm and the corresponding return, and do experiment and analysis.