Granted Patent B2

US 12,748,763 · App. 18/537,429 · Granted Sep 29, 2026

Inventors: Matthias Theodor Middendorf (Reichenau, DE); Jochen Matthias van den Bercken (Munich, DE)

Assignee: Open Text SA ULC

G06F16/24578G06N20/00

Abstract

Claims (76)

1 . A computer-implemented method for automatic filing of documents, the method comprising:

maintaining, by a content management system, workspace type definitions defining a plurality of workspace types, wherein each workspace type defined by the workspace type definitions includes one or more document types supported by workspaces of that workspace type;

maintaining, by the content management system, a workspace data store of workspace data that comprises attribute values describing properties of entities and relationships for a plurality of workspaces in the content management system, wherein the workspaces represent the entities;

maintaining, by the content management system, a specification of strong indicators of the entities, the strong indicators of the entities associated with the workspace types;

receiving an electronically captured document for filing by the content management system, the document comprising unstructured text;

processing the document to automatically file the document to a selected workspace from the plurality of workspaces according to an entity-linking, comprising:

extracting the unstructured text from the document to generate extracted text;

processing the document with a machine learning classifier to assign a document type to the document, wherein processing the document with the machine learning classifier comprises processing the extracted text; and

based on the document type assigned to the document, selecting, by the content management system, as selected strong indicators to detect in the document, strong indicators associated with the workspace types that include the document type assigned to the document;

analyzing, by the content management system, the document using the selected strong indicators to detect a strong indicator from the extracted text, the detected strong indicator corresponding to a workspace attribute value;

querying, by the content management system, the workspace data for workspaces with the workspace attribute value and determining a result set of candidate records based on the querying, each candidate record in the result set of candidate records corresponding to a corresponding workspace that has the workspace attribute value corresponding to the detected strong indicator and including a set of attribute values from the corresponding workspace;

generating, by the content management system, scores for the result set of candidate records, wherein generating the scores for the result set of candidate records comprises, for each candidate record in the result set of candidate records:

detecting mentions in the document that match the attribute values from the set of attribute values in the candidate record; and

generating a score for the candidate record based on the mentions detected in the document that match the attribute values from the set of attribute values in the candidate record;

linking the document to a selected workspace based on the scores for the result set of candidate records; and

automatically storing the document to the selected workspace based on the linking.

2 . The computer-implemented method of claim 1 , wherein the detected strong indicator comprises a list of attribute values.

3 . The computer-implemented method of claim 2 , further comprising:

generating an acyclic graph representing the list of attribute values, wherein detecting the strong indicator from the document comprises evaluating the document using the acyclic graph.

4 . The computer-implemented method of claim 1 , wherein the detected strong indicator comprises a regular expression.

5 . The computer-implemented method of claim 1 , further comprising:

generating a knowledge graph from the workspace data, the knowledge graph comprising attributed nodes representing the workspaces in the plurality of workspaces and edges representing the relationships between the workspaces, wherein querying the workspace data for the workspaces with the workspace attribute value comprises querying the knowledge graph.

6 . The computer-implemented method of claim 1 , wherein generating the score for the candidate record based on the mentions detected in the candidate record comprises weighting mentions of attribute values detected in the extracted text based on the document type.

7 . The computer-implemented method of claim 1 , further comprising automatically filing the document in a folder of the selected workspace based on the document type.

8 . A computer program product comprising a non-transitory, computer readable medium embodying thereon a set of computer executable instructions, the set of computer executable instructions including instructions for:

maintaining, by a content management system, workspace type definitions for a plurality of workspace types, wherein each workspace type defined by the workspace type definitions includes one or more document types supported by workspaces of that workspace type;

accessing, by the content management system, workspace data comprising attribute values describing properties of entities and relationships for a plurality of workspaces in the content management system, wherein the workspaces represent the entities;

maintaining a specification of strong indicators of the entities, the strong indicators of the entities associated with the workspace types;

receiving an electronically captured document for filing by the content management system, the document comprising unstructured text;

processing the document to automatically file the document to a selected workspace from the plurality of workspaces according to an entity-linking, comprising:

extracting the unstructured text from the document;

processing the document with a machine learning classifier to assign a document type to the document, wherein processing the document with the machine learning classifier comprises processing the extracted text; and

identifying workspaces of workspace types that include the document type assigned to the document;

selecting, by the content management system, as selected strong indicators to detect in the document, strong indicators associated with the workspace types that include the document type assigned to the document;

analyzing, by the content management system, the document using the selected strong indicators to detect a strong indicator from the extracted text, the detected strong indicator corresponding to a workspace attribute value;

querying, by the content management system, the workspace data for workspaces with the workspace attribute value and determining a result set of candidate records based on the querying, each candidate record in the result set of candidate records corresponding to a corresponding workspace that has the workspace attribute value corresponding to the detected strong indicator and including a set of attribute values from the corresponding workspace;

generating, by the content management system, scores for the result set of candidate records, wherein generating the scores for the result set of candidate records comprises, for each candidate record in the result set of candidate records:

detecting mentions in the document that match the attribute values from the set of attribute values in the candidate record; and

generating a score for the candidate record based on the mentions detected in the document that match the attribute values from the set of attribute values in the candidate record;

linking, by the content management system, the document to a selected workspace based on the scores for the result set of candidate records; and

automatically storing, by the content management system, the document to the selected workspace based on the linking.

9 . The computer program product of claim 8 , wherein the detected strong indicator comprises a list of attribute values.

10 . The computer program product of claim 9 , wherein the set of computer executable instructions includes instructions for:

generating an acyclic graph representing the list of attribute values, wherein detecting the strong indicator from the document, comprises evaluating the text of the document using the acyclic graph.

11 . The computer program product of claim 8 , wherein the detected strong indicator comprises a regular expression.

12 . The computer program product of claim 8 , wherein the set of computer executable instructions includes instructions for:

providing a data store storing the workspace data; and

generating a knowledge graph from the workspace data, the knowledge graph comprising attributed nodes representing the workspaces in the plurality of workspaces and edges representing relationships between the workspaces, wherein querying the workspace data for the workspaces with the workspace attribute value comprises querying the knowledge graph.

13 . The computer program product of claim 8 , wherein generating the score for the candidate record based on the mentions detected in the candidate record comprises weighting mentions of the attributed values detected in the extracted text based on the document type.

14 . The computer program product of claim 8 , wherein the set of computer executable instructions includes instructions for automatically filing the document in a folder of the selected workspace based on the document type.

15 . A computer system comprising:

an enterprise management system comprising a set of business objects modelling entities in a business process;

a machine learning classifier trained to classify documents by document type;

a content management system comprising a plurality of workspaces, the content management system further comprising:

a memory storing:

workspace type definitions defining a plurality of workspace types, wherein each workspace type defined by the workspace type definitions includes one or more document types supported by workspaces of that workspace type; and

workspace data for a plurality of workspaces, the workspace data comprising attribute values describing properties entities and relationships for the plurality of workspaces, the attribute values representing properties of the entities;

a processor;

a non-transitory, computer-readable medium embodying thereon a set of computer executable instructions executable by the processor, the set of computer executable instructions including instructions for:

specifying strong indicators of entities workspaces, the strong indicators of the entities associated with the workspace types;

receiving an electronically captured document for filing in the content management system;

processing the document to automatically file the document to a selected workspace from the plurality of workspaces according to an entity-linking, comprising:

extracting unstructured text from the document to generate extracted text;

processing the document with the machine learning classifier to assign a document type to the document, wherein processing the document with the machine learning classifier comprises processing the extracted text;

identifying workspaces of workspace types that include the document type assigned to the document;

selecting as selected strong indicators to detect in the document, strong indicators associated with the workspaces of the workspace types that include the document type assigned to the document;

analyzing the document using the selected strong indicators to detect a strong indicator from the extracted text, the detected strong indicator corresponding to a workspace attribute value;

querying the workspace data for workspaces with the workspace attribute value and determining a result set of candidate records based on the querying, each candidate record in the result set of candidate records corresponding to a corresponding workspace that has the workspace attribute value corresponding to the detected strong indicator and including a set of attribute values from the corresponding workspace;

generating scores for the result set of candidate records, wherein generating the scores for the result set of candidate records comprises, for each candidate record in the result set of candidate records:

 detecting mentions in the extracted text that match the attribute values from the set of attribute values in the candidate record; and

 generating a score for the candidate record based on the mentions detected in the text that match the attribute values from the set of attribute values in the candidate record;

linking the document to a selected workspace based on the scores for the result set of candidate record; and

automatically storing the document to the selected workspace based on the linking.

16 . The computer system of claim 15 , wherein the selected workspace includes a set of folders, and wherein the set of computer executable instructions includes instructions for automatically filing the document to a folder in the set of folders based on the document type.

17 . The computer system of claim 15 , wherein the detected strong indicator comprises a list of attribute values.

18 . The computer system of claim 15 , wherein the detected strong indicator comprises a regular expression.

Assignments (1)

Continuity (2)

References Cited (19)