Invention Application
- Patent Title: SYSTEMS AND METHODS FOR INTELLIGENT LABELING OF INSTANCE DATA CLUSTERS BASED ON KNOWLEDGE GRAPH
-
Application No.: US17033357Application Date: 2020-09-25
-
Publication No.: US20220101151A1Publication Date: 2022-03-31
- Inventor: Jan Portisch , Sandra Bracholdt , Michael Hladik , Oliver Huth
- Applicant: SAP SE
- Applicant Address: DE Walldorf
- Assignee: SAP SE
- Current Assignee: SAP SE
- Current Assignee Address: DE Walldorf
- Main IPC: G06N5/02
- IPC: G06N5/02 ; G06F16/901 ; G06F16/2455 ; G06F16/36 ; G06F16/906 ; G06K9/62

Abstract:
An intelligent labeling method is disclosed herein. The method can receive a plurality of input instances and access a knowledge graph comprising a plurality of nodes and a plurality of edges connecting the plurality of nodes. The plurality of nodes represent objects and the edges represent relationship between the objects. For at least some selected input instances, the method can identify respective matching objects represented by corresponding nodes of the knowledge graph. A selected input instance shares a common label with the corresponding matching object. For identified matching objects, the method can identify a common class object represented by a corresponding node of the knowledge graph. A percentage of identified matching objects are descendants of the common class object, and the percentage is greater than a predefined threshold. The method can associate a label of the common class object with the plurality of input instances.
Public/Granted literature
- US11954605B2 Systems and methods for intelligent labeling of instance data clusters based on knowledge graph Public/Granted day:2024-04-09
Information query