Systems and methods for generating and implementing knowledge graphs for knowledge representation and analysis
View Patent ↗This disclosure relates to knowledge generation and implementation. A knowledge graph system comprises at least one processor, at least one database communicatively connected to the at least one processor, and a memory storing executable instructions. When executed, the instructions cause the at least one processor to aggregate, from the at least one database, entity data for a plurality of homes. Attribute information identifying geographic locations of the plurality of homes and relationships between pairs of the plurality of homes is extracted from the aggregated data. Knowledge graph data structures are populated with the extracted attribute information. A home knowledge graph is built, having nodes corresponding to the plurality of homes and edges corresponding to the identified relationships. A hierarchical cluster tree structure of the plurality of homes is outputted, wherein levels of the hierarchical cluster tree correspond to clusters of homes determined based in part on the knowledge graph edges.
1. A knowledge graph computer system, comprising: at least one processor;
at least one database communicatively connected to the at least one processor; and
a memory storing executable instructions which, when executed, cause the at least one processor to perform operations including:
aggregating, from the at least one database, entity data for a plurality of homes;
extracting, from the aggregated data, attribute information identifying geographic locations of the plurality of homes, and relationships between pairs of the plurality of homes;
populating knowledge graph data structures with the extracted attribute information;
building a home knowledge graph having nodes corresponding to the plurality of homes and edges corresponding to the identified relationships;
determining the edges between the plurality of homes and corresponding to the identified relationships by:
identifying at least one comp node in the attribute information;
identifying one or more attributes of the home and the at least one comp node in the attribute information;
determining attribute adjustments of the identified one or more attribute adjustments;
scoring the determined attribute adjustments; and calculating an edge weight for each of the edges based on the scored attribute adjustments; and
outputting a hierarchical cluster tree structure of the plurality of homes, wherein levels of the hierarchical cluster tree correspond to clusters of homes determined based in part on the knowledge graph edges.
2. The knowledge graph computer system of claim 1 , the operations further including:
identifying the clusters of homes in the hierarchical cluster tree, using a machine learning algorithm trained using the home geographic locations and a level of the hierarchical cluster tree.
3. The knowledge graph computer system of claim 2 , wherein the at least one processor is configured to iteratively execute the machine learning algorithm to identify the clusters of homes for multiple levels of the hierarchical cluster tree.
4. The knowledge graph computer system of claim 1 , wherein the calculated edge weight is directly proportional to a degree of similarity between attributes of the home and a respective comp node.
5. The knowledge graph computer system of claim 1 , the operations further including:
receiving geographic location information for an unknown home; and
determining, using the hierarchical cluster tree structure and based on the received geographic location information, a cluster of homes to which the unknown home belongs.
6. The knowledge graph computer system of claim 1 , wherein the at least one processor is part of a distributed computing system.
7. A non-transitory computer readable medium storing instructions which, when executed, cause at least one processor to perform operations comprising:
aggregating, from at least one database communicatively connected to the at least one processor, entity data for a plurality of homes;
extracting, from the aggregated data, attribute information identifying geographic locations of the plurality of homes, and relationships between pairs of the plurality of homes;
populating knowledge graph data structures with the extracted attribute information;
building a home knowledge graph having nodes corresponding to the plurality of homes and edges corresponding to the identified relationships;
determining edges between the plurality of homes and corresponding to the identified relationships by:
identifying at least one comp node in the attribute information; identifying one or more attributes of the home and the at least one comp node in the attribute information;
determining attribute adjustments of the identified one or more attribute adjustments;
scoring the determined attribute adjustments; and
calculating an edge weight for each of the edges based on the scored attribute adjustments; and
outputting a hierarchical cluster tree structure of the plurality of homes, wherein levels of the hierarchical cluster tree correspond to clusters of homes determined based in part on the knowledge graph edges.
8. The non-transitory computer readable medium of claim 7 , the operations further including:
identifying the clusters of homes in the hierarchical cluster tree, using a machine learning algorithm trained using the home geographic locations and a level of the hierarchical cluster tree.
9. The non-transitory computer readable medium of claim 8 , wherein the at least one processor is configured to iteratively execute the machine learning algorithm to identify the clusters of homes for multiple levels of the hierarchical cluster tree.
10. The non-transitory computer readable medium of claim 7 , wherein the calculated edge weight is directly proportional to a degree of similarity between attributes of the home and a respective comp node.
11. The non-transitory computer readable medium of claim 7 , the operations further including:
receiving geographic location information for an unknown home; and
determining, using the hierarchical cluster tree structure and based on the received geographic location information, a cluster of homes to which the unknown home belongs.
12. The non-transitory computer readable medium of claim 7 , wherein the at least one processor is part of a distributed computing system.
13. A knowledge graph computer system, comprising:
at least one processor;
at least one database communicatively connected to the at least one processor; and
a memory storing executable instructions which, when executed, cause the at least one processor to perform operations including:
receiving, from the at least one database, knowledge graph data structures with node data and edge data, the node data corresponding to a plurality of homes, and the edge data corresponding to identified relationships between the plurality of homes;
building a home knowledge graph having nodes and edges based on the node data and the edge data, the home knowledge graph having a hierarchical cluster tree structure of the plurality of homes, wherein levels of the hierarchical cluster tree correspond to clusters of homes determined based in part on the knowledge graph edges;
identifying a first cluster of homes associated with a first level of the hierarchical cluster tree structure;
determining, for each home in the first cluster, at least one vector proportionate to a first edge weight associated with the respective home the first edge weight being directly proportional to a degree of similarity between attributes of the home and a respective comp node;
normalizing the determined vectors;
identifying, based on the normalized vectors, a number of homes of the first cluster that are no longer associated with the first cluster; and
responsive to a determination that the identified number of homes is below a predetermined threshold, assigning the identified homes to a second cluster, and adding the second cluster to the hierarchical cluster tree structure.
14. The knowledge graph computer system of claim 13 , the operations further including:
receiving geographic location information for an unknown home; and determining, using the hierarchical cluster tree structure and based on the received geographic location information, a cluster of homes to which the unknown home belongs.
15. The knowledge graph computer system of claim 13 , wherein the at least one processor is part of a distributed computing system.
16. The knowledge graph computer system of claim 13 , wherein the number of homes of the first cluster that are no longer associated with the first cluster are identified by:
associating a first cluster membership label to the homes of the first cluster;
assigning a second cluster membership label to the homes based on values of the normalized vectors; and
determining whether the first cluster membership label differs from the second cluster membership label.
17. The knowledge graph computer system of claim 13 , wherein the node data includes attribute information identifying geographic locations of the plurality of homes.