Automatic discovery of significant events from databases

The advent of the internet has caused enormous amounts of data available online causing many significant facts to be hidden within this data. Searching for a significant fact within these large datasets is a query intensive process involving large amounts of queries which needs to be executed hence slowing the process of finding the significant facts from a large dataset. In this thesis, a novel approach has been designed exploiting the mathematical characteristics of the data present in the dataset to reduce the number of queries on the dataset. A two phased approach is considered for making fact finding more efficient. The approach consists of design and implementation of the prediction and the decision making algorithm. The prediction algorithm predicts the time frame for a significant event to happen and the decision algorithm uses the results from the prediction algorithm to decide whether to check for a significant event or not. We compare our results obtained after the implementation of the designed algorithms and found that queries are executed lesser number of times compared to the other existing solutions to this problem.