| CPC G06V 10/26 (2022.01) [G06F 40/295 (2020.01); G06V 10/82 (2022.01); G10L 15/16 (2013.01); G10L 15/1815 (2013.01)] | 20 Claims |
|
1. One or more non-transitory computer-readable media having computer-executable instructions stored thereon that, when executed by one or more processors of a semantic network platform, cause the semantic network platform to perform operations comprising:
receiving, by a resource processor of the semantic network platform, a resource comprising a digital document;
segmenting, by the resource processor, the resource into a set of content chunks based on dynamically identified boundaries, wherein the segmenting comprises applying a chunking algorithm, and wherein the chunking algorithm includes one or more of: content-based chunking using resource section headings, content-based chunking using named entity recognition, page-based chunking, or fixed-size chunking;
using the set of content chunks, generating, by the resource processor, a first concept, a second concept, and a relationship, wherein the relationship is between the first concept and the second concept and is determined by performing one or both of: (a) performing proximity analysis on a first token that corresponds to the first concept and a second token that corresponds to the second concept, or (b) analyzing resource metadata;
using the first concept, the second concept, and the relationship, generating, by a node generator, computer-executable code to generate a node triple in a graph structure, the node triple comprising a first node encoding the first concept, a second node encoding the second concept, and a relationship node;
executing, by the semantic network platform, the computer-executable code against the graph structure;
generating, by a node indexer, a first set of embeddings that corresponds to at least a portion of the first node, a second set of embeddings that corresponds to at least a portion of the second node, and a third set of embeddings that corresponds to at least a portion of the relationship node, wherein the first set of embeddings, the second set of embeddings, and the third set of embeddings are indexed relationally to a first node identifier, and wherein the first node identifier correlates to the first concept;
receiving, by an interface module of the semantic network platform, a query that pertains to the resource, wherein at least a portion of the query identifies the first concept, the second concept, or the relationship; and
using the at least a portion of the query to query, by the interface module of the semantic network platform, one or more of the first set of embeddings, the second set of embeddings, and the third set of embeddings to identify embeddings associated with the first node identifier that correlates to the at least a portion of the query.
|