Literature Review on Information Extraction by Partitioning
Information extraction systems are implemented traditionally as a pipeline of special-purpose processing modules that target at the extraction of a specific type of information. Such an approach has major drawback that whenever a module is improved or a new extraction goal emerges, extraction need to be applied to all parts even if only a small part is affected. Here, the authors describe an approach for information extraction in which they partition the dimensions (attributes), i.e., a higher dimension of large data set can be transformed into relatively smaller subsets of data in certain numbers that might be processed easily. Here they use heterogeneous database as large data set.