System and method for global identification in a collection of documents

Patent metadata

Field	Value
Publication number	US-2016275347-A1
Application number	US-201514714556-A
Country	US
Kind code	A1
Filing date	May 18, 2015
Priority date	Mar 19, 2015
Publication date	Sep 22, 2016
Grant date	—

Abstract

Techniques for machine-based identification of objects extracted from text documents in natural language are disclosed. Text documents with extracted objects are presented in a form of Resource Description Framework (RDF) graphs with the nodes correspondent to the objects and arcs correspondent to the relations between objects. Identification of objects is implemented using specific combinations of patterns which define features of the objects.

First claim

What is claimed is:

1 . A method of machine-based identification of one or more information objects in a document and in a document storage, the information objects in the document and information objects in the document storage corresponding to the same real world object, the method comprising:
(a) searching for global identification patterns and for a combination of the global identification patterns in the document; (b) searching for the same global identification patterns and their combinations in the document storage; (c) finding matching pairs of the information objects, one information object from the document and at least one information object from the document storage for the same combination of the patterns; (d) ascertaining consistency of the matching pairs and determining which said one or more information objects in the document are suitable for merging into the document storage; and (e) adding the one or more information objects from the document to the document storage.

2 . The method of claim 1 , wherein step (e) further comprises adding one or more features of said one or more information objects in the document to the document storage if said one or more feature is absent from the document storage and if said one or more information objects in the document and in the document storage correspond to the same real world object.

3 . The method of claim 1 , wherein step (e) further comprises adding one or more information objects from the document to the document storage as new information objects if said one or more information objects in the document storage do not have one or more information objects in the document storage corresponding to the same real world object.

4 . The method of claim 1 , wherein the global identification patterns correspond to features of real world objects.

5 . The method of claim 1 , wherein the document and the document storage are presented in a form of an RDF graph.

6 . The method of claim 1 , further comprising using computer language SPARQL for communicating with the document storage.

7 . The method of claim 1 , wherein ascertaining consistency of the matching pairs further comprises ascertaining consistency of features of said one or more information objects with the ontology.

8 . The method of claim 1 , wherein the global identification pattern comprise a combination of pattern.

9 . The method of claim 8 , further comprising automatically generating the combination of patterns.

10 . The method of claim 9 , further comprising computing weights of each pattern of the combination and training the classifier.

11 . The method of claim 8 , further comprising manually generating the combination of pattern.

12 . The method of claim 11 , further comprising computing weights of each pattern of the combination and generating a unicity parameter and a specialty parameter for the combination.

13 . A platform for machine-based identification of one or more information objects in a document and in a document storage, the information objects in the document and in the document storage corresponding to the same real world object, the platform comprising:
at least one local, remote, distributed or web-based computing device; and a memory locally or remotely coupled to the computing device and storing instructions which, responsive to execution on the computing device, cause the computing device to perform: (a) searching for global identification patterns and for a combination of the global identification patterns in the document; (b) searching for the same global identification patterns and their combinations in the document storage; (c) finding matching pairs of the information objects from the document and the document storage for the same combination of the patterns; (d) ascertaining consistency of the matching pairs and determining which said one or more information objects in the document are suitable for merging into the document storage; and (e) adding the one or more information objects from the document to the document storage.

14 . The platform of claim 13 , wherein the platform is configured to use computer language SPARQL for communicating with the storage.

15 . The platform of claim 13 , wherein the document and the document storage are in the form of an RDF graph.

16 . The platform of claim 13 , wherein step (e) further comprises adding one or more features of said one or more information objects from the document to the document storage if said one or more feature is absent from the document storage and if said one or more information objects in the document and in the document storage correspond to the same real world object.

17 . The platform of claim 13 , wherein step (e) further comprises adding one or more information objects from the document to the document storage as a new information object if said one or more information objects in the document storage do not have one or more information objects in the document storage corresponding to the same real world object.

18 . A computer-readable medium storing processor-readable instructions for machine-based identification of one or more information objects in a document and in a document storage, the information objects in the document and in the document storage corresponding to the same real world object, the instructions which, responsive to execution in a computing device, cause the computing device to perform:
(a) searching for global identification patterns and for a combination of the global identification patterns in the document; (b) searching for the same global identification patterns and their combinations in the document storage; (c) finding matching pairs of the information objects from the document and the document storage for the same combination of the patterns; (d) ascertaining consistency of the matching pairs and determining which said one or more information objects in the document are suitable for merging into the document storage; and (e) adding the one or more information objects in the document to the document storage.

19 . The medium of claim 18 , wherein the instructions further cause:
using computer language SPARQL for communicating with the storage.

20 . The medium of claim 18 , wherein the instructions further cause step (e) to further comprise adding one or more feature of said one or more information objects from the document to the document storage if said one or more feature is absent from the document storage and if said one or more information objects in the document and in the document storage correspond to the same real world object.

21 . The medium of claim 19 , wherein the instructions further cause step (e) to further comprises adding one or more information objects from the document to the document storage as a new information object if said one or more information objects in the document storage do not have one or more information objects in the document storage corresponding to the same real world object.

Assignees

• Abbyy Infopoisk Llc

Inventors

• Sukhodolov Dmitry
• Matskevich Stepan
• Starostin Anatoly

Classifications

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06K9/6215

  Physics · mapped topic

• G06K9/00456

  Physics · mapped topic

• G06K9/00469 Primary

  Physics · mapped topic

• G06V30/416 Primary

  Extracting the logical structure, e.g. chapters, sections or page numbers; Identifying elements of the document, e.g. authors · CPC title