AWS Neptune - Causation Relationship Use Case

Question

I have a use case where i want to create a database of related entities . I am familiar with graph database ability to manage static relationships such as friend of friend and being able to traverse the database but this a different case. I want to manage a numeric value and the factor of those relationships. For example if i had 3 entities A , B and C . A and B store a value while C stores the calculation of the value of A and B. If A is 10 and B is 5 and i know that A impacts C by 50% and B impacts C by 25% then C = 10x.5 + 5x.25 = 6.25. I want to expand this to 1000s of chained nodes but be able to use this causation analysis to find the paths of least or most impact (optimization) . In more concrete terms if C = Sales and i have the option of increasing investment A vs investment B - A would be the better choice since is impacts C to a larger degree. I am interested any ideas on alternate databases choices if Neptune is not the recommended database type. Performance has to be near real-time to support scenario analysis . My ultimate goal was to have the Neptune DB as a source for Bedrock to analyze and recommend optimization strategies

AWS-KRL
vor einem Monat
Are you thinking along the lines of precomputing the impacts in real time too, or could they be stored as edge weights between the nodes? A bit like (A)-[:IMPACT {weight:0.5}]->(C)

Answer

We have discussed several times having a pluggable belief calculus.  This could achieve what you want in a general framework.  This might operate by tracking the identity of each tuple supporting a query, then going back over those tuples and applying weights using a customer specified formula to compute the aggregate score for each solution.   #neptune-interest if you want to discuss use cases.

You can use things such as shortest paths with Neptune Analytics to find a minimum cost path.  If your graph is just edges and weights, some flavor of graph algorithm might give you want you want with an appropriate semi-ring.  Again, #neptune-interest to discuss.

AWS Neptune - Causation Relationship Use Case

Relevanter Inhalt