Combining statistical methods with a knowledge graph
Certain aspects of the present disclosure provide techniques for node matching with accuracy by combining statistical methods with a knowledge graph to assist in responding (e.g., providing content) to a user query in a user support system. In order to provide content, a keyword matching algorithm, statistical method (e.g., a trained BERT model), and data retrieval are each implemented to identify node(s) in a knowledge graph with encoded content relevant to the user's query. The implementation of the keyword matching algorithm, statistical method, and data retrieval results in a matching metric score, semantic score, and graph metric data, respectively. Each score associated with a node is combined to generate an overall score that can be used to rank nodes. Once the nodes are ranked, the top ranking nodes are displayed to the user for selection. Based on the selection, content encoded in the node is displayed to the user.
1. A method, comprising:
receiving, via a conversational user interface, a query;
executing, based on the query, a keyword matching algorithm on a knowledge graph comprising a set of nodes, wherein each node includes content and a descriptive label;
generating a matching metric score for each node of the knowledge graph to the query based on the keyword matching algorithm;
identifying, based on the matching metric score, a subset of nodes with matching metric scores that meet a threshold;
calculating a semantic score for each pairing of the query and a node in the subset of nodes by:
extracting a vector representation of each query and node pairing with a neural network model;
retrieving a graph metric data for each node in the subset of nodes;
generating an overall score for each node in the subset of nodes based on:
the matching metric score,
the semantic score, and
the graph metric data;
ranking each node in the subset of nodes according to the overall score of the node; and
presenting the descriptive label of each node in the subset of nodes according to the ranking in the conversational user interface.
2. The method of claim 1 , wherein the neural network model is a Bidirectional Encoder Representations from Transformers (BERT) model.
3. The method of claim 1 , further comprising:
receiving, via the conversational user interface, a selection of a node in the ranking based on the corresponding descriptive label; and
presenting the content associated with the selected node in the conversational user interface.
4. The method of claim 3 , wherein the selection of the node is feedback in continuing training of the neural network model.
5. The method of claim 1 , wherein the query is an audio query.
6. The method of claim 1 , wherein the keyword matching algorithm is a trigram matching algorithm.
7. The method of claim 1 , wherein the graph metric includes a position of the node in the knowledge graph or a relevance metric.
8. A system, comprising:
a processor; and
a memory storing instructions which when executed by the processor perform a method comprising:
receiving, via a conversational user interface, a query;
executing, based on the query, a keyword matching algorithm on a knowledge graph comprising a set of nodes, wherein each node includes content and a descriptive label;
generating a matching metric score for each node of the knowledge graph to the query based on the keyword matching algorithm;
identifying, based on the matching metric score, a subset of nodes with matching metric scores that meet a threshold;
calculating a semantic score for each pairing of the query and a node in the subset of nodes by:
extracting a vector representation of each query and node pairing with a neural network model;
retrieving a graph metric data for each node in the subset of nodes;
generating an overall score for each node in the subset of nodes based on:
the matching metric score,
the semantic score, and
the graph metric data;
ranking each node in the subset of nodes according to the overall score of the node; and
presenting the descriptive label of each node in the subset of nodes according to the ranking in the conversational user interface.
9. The system of claim 8 , wherein the neural network model is a Bidirectional Encoder Representations from Transformers (BERT) model.
10. The system of claim 8 , further comprising:
receiving, via the conversational user interface, a selection of a node in the ranking based on the corresponding descriptive label; and
presenting the content associated with the selected node in the conversational user interface.
11. The system of claim 10 , wherein the selection of the node is feedback in continuing training of the neural network model.
12. The system of claim 8 , wherein the query is an audio query.
13. The system of claim 8 , wherein the keyword matching algorithm is a trigram matching algorithm.
14. The system of claim 8 , wherein the graph metric includes a position of the node in the knowledge graph or a relevance metric.
15. A non-transitory computer-readable storage medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to perform a method, the method comprising:
receiving, via a conversational user interface, a query;
executing, based on the query, a keyword matching algorithm on a knowledge graph comprising a set of nodes, wherein each node includes content and a descriptive label;
generating a matching metric score for each node of the knowledge graph to the query based on the keyword matching algorithm;
identifying, based on the matching metric score, a subset of nodes with matching metric scores that meet a threshold;
calculating a semantic score for each pairing of the query and a node in the subset of nodes by:
extracting a vector representation of each query and node pairing with a neural network model;
retrieving a graph metric data for each node in the subset of nodes;
generating an overall score for each node in the subset of nodes based on:
the matching metric score,
the semantic score, and
the graph metric data;
ranking each node in the subset of nodes according to the overall score of the node; and
presenting the descriptive label of each node in the subset of nodes according to the ranking in the conversational user interface.
16. The non-transitory computer-readable storage medium of claim 15 , wherein the neural network model is a Bidirectional Encoder Representations from Transformers (BERT) model.
17. The non-transitory computer-readable storage medium of claim 15 , further comprising:
receiving, via the conversational user interface, a selection of a node in the ranking based on the corresponding descriptive label; and
presenting the content associated with the selected node in the conversational user interface.
18. The non-transitory computer-readable storage medium of claim 17 , wherein the selection of the node is feedback in continuing training of the neural network model.
19. The non-transitory computer-readable storage medium of claim 15 , wherein the query is an audio query.
20. The non-transitory computer-readable storage medium of claim 15 , wherein the keyword matching algorithm is a trigram matching algorithm.