Training Series - Create a Knowledge Graph: A Simple ML Approach

This talk will start with unstructured text and end with a knowledge graph in Neo4j using standard Python packages for Natural Language Processing. From there, we will explore what can be done with that knowledge graph using the tools available with the Graph Data Science Library.

Useful links:

[chris]

Just wanna say all these video's your channel releases is really appreciated, as someone trying to learn ml, graph, datascience etc to use in a real word project Its cool to see so much education material, really helps with the learning curve.

[elderspirit]

You guys are extremely generous with your time and vast expertice, I cant send enough of my love your way <33

[dylanmazer]

great video, incredibly useful and interesting

[mohammadhosseinjafari]

it was a really interesting and useful tutorial. Thanks for sharing it

[wu_kenny]

source code 25:00
method1: NLP-only method pipeline formally  starts at 25:29
method2: NLP-lite by using wiki data 1:22:08

[antoniovieiradasilvajunior]

There are some way to automatic specify of P-Values (verbs list) ? Doesn`t make sense doing this manually to each document.

[estieboshoff]

Question, why would we delete duplicates and relationships rather than replace them with one node (that we decide is the primary one for the person, e.g. Barack Obama) and preserve the relationships. After that then check for duplicates throughout the entire graph?

[fanmo]

May I ask, what is the api_key? I want to use my local machine to run this code, not the google colab

[Relative]

I'm guessing Neo4j might be dead by now, I'm having several problems with their site (errors with code you provided and search button doesn't work in two browsers). There were multiple issues with them in your video as well - which was a year ago. But, thank you for this great tutorial, lots to be learned even still, and a really nice Github repo as well. Thanks.

[iiio]

"In order to create a knowledge graph you need a subject matter expert" - sounds like not much progress has been made since AI Winter, not counting hardware.

[hichembouricha]

CALL gds.graph.create(
'all_nodes',
{
AllNodes: {label: 'Node',
properties: {word_vec_embedding: {property: 'word_vec'}}}
},
{
AllRels: {type: '*', orientation: 'UNDIRECTED'}
}
)
YIELD graphName, nodeCount, relationshipCount


There is no procedure with the name `gds.graph.create` registered for this database instance. Please ensure you've spelled the procedure name correctly and that the procedure is properly deployed.

[hichembouricha]

CALL gds.beta.node2vec.stream('all_nodes', {embeddingDimension: 10})
YIELD nodeId, embedding
RETURN gds.util.asNode(nodeId).name as name, embedding

Failed to invoke procedure `gds.beta.node2vec.stream`: Caused by: Graph with name `all_nodes` does not exist on database `neo4j`. It might exist on another database.