GRAPH Explainable ? two ways to detection why we taking them. — structural , Symbolic Reasoning—

i will introduce how we apply the explainable in the industry.

1. Structural

2. Symbolic Reasoning

paper ; CAFE ; Coarse-to-Fine Neural Symbolic Reasoning for Explainable Recommendation.

key contribution are following;

a. highlight important shortcomings of past KG reasoning approaches for explainable recommendation.

b. coarse-to-fine paradigm to approach the problem by explicitly injecting diverse user behavior modeling into the KG reasoning process.

c. novel profile-guided path reasoning algorithm with symbolic reasoning module.

d. specific at e-commerce industry.

core concept

1. Coarse-to-Fine paradim

traditional ; the problem is each existing both recsys performance and path reasoning

→ simultaneously obtaining competitive recommendation performance.

problem definition.