search

vadasg / githubarchive-parser

10 stars 3 forks source link

add makeType with unidirectional() #10

Open vadasg opened 11 years ago

vadasg commented 11 years ago

should make for a cleaner graph for faunus

okram commented 11 years ago

Make sure for only edges you don't want bidirectional trvaersal. It is usually the case you want bidirectional traversal. Unidirectional is usually used for for ontological mappings. On Nov 8, 2012 6:52 PM, "Vadas Gintautas" notifications@github.com wrote:

should make for a cleaner graph for faunus

— Reply to this email directly or view it on GitHubhttps://github.com/vadasg/githubarchive-parser/issues/10.

vadasg commented 11 years ago

Ok. I've reviewed the schema and I believe all of the edges should be unidirectional for GitHubArchive, since they are all generated by unilateral events (as far as the vertices are concerned). That is, a user -- pushes -->commit -- to --> repository but a repository never 'to' a commit or a commit never 'pushes' a user. Worse yet, if User1 -- follows -- User2 that is not the same thing as User2 -- follows --> User1.

okram commented 11 years ago

Let's talk about this... You are not thinking right.. its not the semantics of the link, but the requirement of the travesal. Unidirwctional is all about high entroy in the information sense.... On skype ping me.

http://markorodriguez.com On Nov 10, 2012 6:14 PM, "Vadas Gintautas" notifications@github.com wrote:

Ok. I've reviewed the schema and I believe all of the edges should be unidirectional for GitHubArchive, since they are all generated by unilateral events (as far as the vertices are concerned). That is, a user -- pushes -->commit -- to --> repository but a repository never 'to' a commit or a commit never 'pushes' a user. Worse yet, if User1 -- follows -- User2 that is not the same thing as User2 -- follows --> User1.

— Reply to this email directly or view it on GitHubhttps://github.com/vadasg/githubarchive-parser/issues/10#issuecomment-10262082.