GDS Community Algorithm based on text properties

[bjorge_eikenes]

Related to the field of cyber security I'm looking at graph database modeling and graph algorithms.
I have Requests nodes that typical will contain data related to a HTTP request, in example HTTP Header and HTTP URL information. It is possible to use some sort of word2vec or something directly or as a seed into a GDS graph community algorithm? In short I'm exploring if this can be used to classify botnets and categories of users based on the HTTP header information as well as their HTTP protocol compliance.

At the moment I'm reading https://neo4j.com/labs/apoc/4.2/nlp/gcp/

[alicia_frame1]

bjorge.eikenes:

he field of cyber security I'm looking at graph database modeling and graph algorithms.
I have Requests nodes that typical will contain data related to a HTTP request, in example HTTP Header and HTTP URL information. It is possible to use some sort of word2vec or something directly or as a seed into a GDS graph community algorithm? In short I'm exploring if this can be used to classify botnets and categories of users based on the HTTP header information as well as their HTTP protocol compliance.

You can run word2vec externally to calculate the word embeddings, load those into the graph, and then use KNN to compute similarity: https://neo4j.com/docs/graph-data-science/current/algorithms/knn/

[janezic]

Dear Alicia,

Since your last post some time has passed by, therefore let me put this question again: is there a (new) way to apply Word2Vec on specific properties of nodes within a graph internally?

So, what I mean is, without exporting it to another platform, calculate the the embeddings and re-import them into the graph.