Probability based sampling

[YolandaLyj]

Nebula does not yet have the ability to sample based on probability.

I am often faced with the need that in classical graph neural networks, the acquisition of samples is usually based on probabilistic sampling. For example, in the graphsage model, the positive sample of a node is a probability sampling of the neighbor node based on the weight value of the neighbor node.

I want to add a query phrase (e.g. SAMPLING) to sample the output based on probability. The syntax is ...｜ SAMPLING <expression> <sample_number>。

For example, go from 1 yield id($$) as id, properties($$).weight as weight | sampling weight 5。

Probability sampling is based on the value of the weight column, and the larger the weight value, the greater the probability of being sampled.