Modern Methods of Extracting Key Information From Regulatory Documents

Abstract

This article is an attempt to comprehend the difficulties and propose approaches to eliminate them when analyzing legal documents in the framework of economic and interdisciplinary research. The utmost goal is to seek incorporating advances in computational linguistics and natural language analysis into the discourse of the digital economy in order to develop methods involved in decision-making and strategy development based on the analysis of textual information. In conditions when the amount of information is too large, is constantly updated and / or the area of study is new, the most expedient at the first stage is to obtain the general structure of the entire collection of documents, some kind of semantic compression of information. The practical part contains the development of an approach for the analysis of regulations governing food and nutrition issues, in particular, related to the prevention of the development of iron deficiency anemia (IDA). The approach includes the extraction of key information of voluminous texts (keywords and key sentences) based on the TextRank graph algorithm. An important link contributing to cognition is also the visualization of semantic relationships between words within documents. In our opinion, it is the combination of semantic compression and visualization of information as a “close-up” of text documents, as well as the possibility of further detailing by linear reading and analysis, which are the most relevant approach in conditions of information overload and attention deficit. The active introduction of text analytics methods for systems that are not involved in attention markets, which lag significantly behind in the convenience of extracting meaningful information, is especially important. Approaches to improve the understanding of large volumes of regulations will be of significant value to researchers in economic, legal or multidisciplinary research.