The fuzzy objects recognition in scientific and technical papers by means of natural languages processing technologies.

The article describes an approach to build a search index for fuzzy objects recognition in scientific and technical papers. A fuzzy object is explained as an entity described by some physical parameters. A physical parameter consists of two parts: a value, represented in various forms of digits, and a physical unit, represented as text with some digits. The approach presented in the article allows users to search not by ordinary substrings, but by physical parameters with range, less than or greater than operators. The current realisation of the approach is a web service that has functionality to perform all stages for extraction of physical parameters of a fuzzy object. These stages are the extraction of sentences, the extraction of physical parameters and the normalisation of physical parameters values. The web service can be exploited in the synchronous and asynchronous modes. The synchronous mode helps to troubleshoot the software. The asynchronous mode is designed to process papers using queues in the distributed computer systems. The current work is aimed to develop a search engine which will address the challenge of structuring information of a continuously growing number of scientific and technical papers. The search index will help to perform robust navigation among a huge number of documents for scientific workers and engineers. [ABSTRACT FROM AUTHOR]