wrong schema depiction regarding the containerOf edge

[scampi]

Based on this dataset It seems the schema has errors regarding the containerOf edge:

• wrong arrow direction: The schema shows a connection from a Post to a Forum but it should be the opposite, since the Post has a column ContainerForumId. The schema wrongly indicates that the Forum has the foreign key to a Post.
• missing connection: The data includes the column ParentPostId but there is no connection between a Comment to a Post. It seems there should be a link from a Comment to a Post that is not displayed in the diagram.

In both cases, the label of the edge could be labelled hasContainer, instead of containerOf.

See https://github.com/ldbc/ldbc_snb_docs/blob/dd444ead15c3ebad5bcaf426a0d123809e2d2736/figures/schema-comfortable.pdf

[szarnyasg]

Hi @scampi, thanks for opening this issue. Regarding your two points:

containerOf

I understand your point but I think it stems from applying the relational model on the schema diagram. This schema diagram tries to capture a graph schema where

See the table on page 22:

Name	Source	Destination	Type	Description
containerOf	Forum[1]	Post[0..*]	D	A

This should be read and thought of as Source, does something to, Destination, which in this case means "Forum (is the) containerOf (a) Post".

Now even if we wanted to change this, renaming this edge would be a massive amount of work – we have tens of terabytes of pre-generated data stored along with about a dozen official implementations of the benchmark. So I don't think it's realistic to change it in any case.

Message edges

The dataset indeed encodes the relationship between Message instances with the ParentPostId and the ParentCommentId fields, and the meaning of these is not clear from the specification. This can be amended from the documentation quite simply and I will do so soon.