Relation extraction from text documents

Business intelligence became a very significant part of many business systems by enabling better insight into corporate data and quicker decision making. Combination of business intelligence with information extraction methods can significantly improve data set used for analysis with the result of enlarged company knowledge base. Information extraction focuses on unstructured data (e.g. text documents). Some of the most significant factors in unstructured source are relations among specific entities. Relation extraction is information extraction method for partial analyses of natural language. This kind of information system is exploring text in order to detect previously defined relations. Machine learning is probably the optimal approach for relation extraction and this paper will give the overview and discuss this type of methods. Special emphasis will be given to weakly supervised techniques, or more specifically, to co-learning method.