User의 PoI historical data로부터 spatial, temporal, preference 세 측면의 새로운 PoI exploration을 수행함Exploit user historical data(local view) + Explore new PoIs(g

A page is important if it is pointed to by other important pagesConnection to random walkWhat is the long-term distribution of the surfers?PageRank is

Goal: embedding space에서 가깝에 위치하게 embedding하기 위함$$similarity(u,v) \\approx \\bold{z}{v}^{T} \\bold{z}{u}$$Encoder is just an ebmbedding-lookupEach node

Graph Kernels: Measure similarity between graphsGoal: Design graph feature vector $\\phi (G)$Key Idea: Bag of WordsKey Idea: Count the number of diffe

ddd