uclnlp / cqd

Continuous Query Decomposition for Complex Query Answering in Incomplete Knowledge Graphs
MIT License
95 stars 11 forks source link

The procedure of the t-norms and neural link prediction #6

Closed Joyrocky closed 3 years ago

Joyrocky commented 3 years ago

I sorry to say that I could't really understand the procedure of the t-norms and neural link prediction when I studying the codes of this module๐Ÿ˜” ,could you give some pseudo codes, math formulas or illustrations about this module. thx!๐Ÿ˜˜ Looking forward for your reply๐Ÿ˜Š

pminervini commented 3 years ago

Hi @Joyrocky, think of a t-norm as a generalisation of the logical AND!

It takes two scalar inputs x, y โˆˆ [0, 1], and returns a value z โˆˆ [0, 1]. In case x, y โˆˆ {0, 1}, then the t-norm will behave like the logical AND (and return z โˆˆ {0, 1}). In case x โˆˆ (0, 1) or y โˆˆ (0, 1), the t-norm may also return a value z โˆˆ (0, 1).

There are many examples of t-norms -- check this out: https://en.wikipedia.org/wiki/T-norm#Prominent_examples