TypeDB Lecture: Why We Need a Polymorphic Database

TypeDB
1

Seeing the lecture https://www.youtube.com/watch?v=2S0zPXQCy0U&t=1541s on Why We Need a Polymorphic Database, does this meen that TypeDB will support graphs without a schema?

2

No it will not! TypeDB is a polymorphic database. One of the essential components of a polymorphic database is the schema, which provides the context necessary to enforce semantic integrity and resolve declarative polymorphic queries. If you’re looking for a schemaless graph database, then there are many great solutions out there.

3

What I am looking is for a schema less hyper graph database with an inference engine that can take rules like TypeDB Can you recommend me one?

4

You could just have a trivial schema.

define 
id sub attribute value string;
label sub attribute, value string;

vertex sub attribute, own id @key, plays edge:from, edge:to;
edge sub relation, relates from, relates to, owns label; # or just to

insert 
$a isa vertex, has id "here";
$b isa vertex, has id "there";
(from:$a,  to:$b) isa edge, has label "road";

match 
(from:$a, to: $b) has label "road"; (from: $b, to: $c) has label "rail";

If we let type-inference handle things, we can play some golf in the match statements:

define 
label sub attribute, value string;
vertex sub attribute, value string, plays edge:endpoint;
edge sub relation, relates endpoint, has label ;

match 
($a,$b) has "road"; ($b, $c) has "rail";
5

I don’t know of any off the top of my head. That’s quite a unique intersection of features, and I think you’ll struggle to find something that does those things without a schema.

6

I like the type-inference but the data are stored as strings. Is there a way to store the data in their correct data-type regardless of how they were inserted. Also is there a way for the database to infer the input is a label, vertex or edge? What about unique keys. Can Typedb infer those? I am looking for the absolute generalist schema where the Database inference engine does as most of the work as possible.

7

I’m not sure what you mean.
Attributes are stored as whatever value they’re declared to be.

If you defined attributes

define
  name sub attribute, value string;
  age sub attribute, value long;

You would only be able to insert strings for the attribute name and longs for the attribute age. Trying to insert a string for an age would throw an error.

8

define
label sub attribute, value string;
vertex sub attribute, value string, plays edge:endpoint;
edge sub relation, relates endpoint, has label ;

I am referring to the above schema you’ve provided. It defines everything as string.