| CPC G06F 40/295 (2020.01) [G06F 18/22 (2023.01); G06F 18/24 (2023.01); G06F 40/211 (2020.01); G06F 40/216 (2020.01); G06F 40/284 (2020.01); G06F 40/30 (2020.01); G06N 3/042 (2023.01); G06N 3/044 (2023.01); G06N 5/01 (2023.01); G06N 5/02 (2013.01); G06N 5/022 (2013.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01)] | 18 Claims |
|
1. A knowledge platform comprising:
a memory storing a knowledge engine with a dictionary of one or more named entities and one or more topics, and a relationship graph of nodes that correspond to at least some of the named entities and at least some of the topics, and edges that correspond to relationship scores between the entities, wherein the relationship graph defines paths of the nodes and the edges between different entities and topics;
at least one processor coupled to the memory programmed with executable instructions, the instructions including an interface to receive a request identifying a topic of the one or more topics, the request from a requestor;
wherein the at least one processor is configured to:
receive electronic communication data;
continuously update the knowledge engine by processing the communication data using an entity recognition unit comprising a syntactic dependency parser, a named entity recognizer, and a probabilistic latent semantic analysis model, wherein the syntactic dependency parser parses words within the communication data to recognize the one or more named entities and the one or more topics in the words within the communication data using relationships between the words in the communication data, wherein the named entity recognizer detects and classifies the one or more named entities in the words in the communication data by entity type, the named entity recognizer trained on one or more domain models, and wherein the probabilistic latent semantic analysis model identifies the one or more topics in the communication data and computes relationships between the one or more topics and the named entities, wherein the probabilistic latent semantic analysis model is trained using a vector space and a topic model;
classify, using an expert classifier, an expert entity from the one or more named entities as an expert in the topic of the request based on a threshold metric that relates to expert classification, wherein the processor uses one or more models to convert words to vector representations of the words, and computes one or more metrics using the vector representations to classify the expert entity as the expert in the topic of the request based on the threshold metric, wherein the one or more models are trained on a corpus of text using the one or more named entities and the one or more topics;
generate or update, by processing the communication data using a relationship model, the relationship graph of the nodes that correspond to at least some of the named entities and at least some of the topics in the knowledge engine, and the edges that correspond to the relationship scores between the entities and the topics, wherein each of the one or more relationship scores comprise a sentiment score, a formality score, and a duration score generated from relationship data of the communication data, wherein the relationship graph has a node corresponding to the requestor and a node corresponding to the expert entity;
compute a relationship path between the requestor and the expert entity using one or more edges connecting the node corresponding to the requestor and the node corresponding to the expert entity of the relationship graph, wherein the relationship path is selected from a plurality of different paths connecting the node corresponding to the requestor and the node corresponding to the expert entity of the relationship graph;
generate, using the relationship model and the relationship path between the requester and the expert entity, a relationship score for the requestor and the expert entity indicating strength of a relationship between the requestor and the expert entity, the relationship score generated based on the one or more relationship scores corresponding to the one or more edges between the node corresponding to the requestor and the node corresponding to the expert entity of the relationship graph;
a presentation server configured to generate visual effects representing the one or more named entities and the one or more topics, the relationship path between the requestor and the expert entity classified as the expert in the topic of the request, and the relationship score for the requestor and the expert entity indicating the strength of the relationship between the requestor and the expert entity, wherein the interface transmits the visual effects to a device for display.
|