RECOVERY OF TEXT ANNOTATIONS RELATED TO INFORMATION OBJECTS Russian patent published in 2018 - IPC G06F17/21 

FIELD: data processing.

SUBSTANCE: invention relates to methods for recovering text annotations in information retrieval systems from natural language texts and persistent machine-readable data medium. Method includes obtaining text in natural language, extracting from text in natural language one or more information objects, where each information object is associated with one or more attributes, wherein said retrieval includes determining confidence level value associated with attribute of each information object, verification of attribute values of plurality of information objects, identification in plurality of information objects of such information object, for which at least one attribute of particular information object has not been associated with at least one text annotation, and restoration of text annotation associated with attribute of certain information object, where text annotation is represented by fragment of text in natural language corresponding to value of attribute.

EFFECT: technical result consists in restoring text annotations associated with information objects and attributes of information objects.

16 cl, 25 dwg