Analysis and classification of algorithms for relation extraction from text data

Kobyshev Kirill Serveevich, Molodyakov Sergey Aleksandrovich

Now, searching for information in unstructured data presented in the form of text is a non-trivial task. Text data may be casted to structured data with automatic relation extraction algorithms. Structured text representation allows to use structured data advantages: explainability of found facts, simplicity of fact search, possibility of data access acceleration mechanisms using. In this paper the following relation extraction approaches were considered and compared: semi-automatic approach, weakly supervised learning, supervised learning, distantly supervised learning, unsupervised learning.

Keywords:relation extraction, structured data, computational linguistics, knowledge extraction, text data processing.