LINE Second-Order Loss

Problem

Compute the second-order LINE loss for one context node: the negative log softmax probability of the true neighbor.

scores is the list of context scores and true_index marks the observed neighbor. Apply a numerically stable softmax and return -log p as a float.

Hint

Second-order LINE treats every other node as negative context, weighted by softmax.

case 1

in: [[1,2,0.5],1]

expected: 0.46436878410794485

case 2

in: [[0,0],0]

expected: 0.6931471805599453

case 3

in: [[3,1],0]

expected: 0.12692801104297258

Starter codePython

import math
def line_second_order_loss(scores, true_index):
    # Your code here
    pass

3 test cases · run your code against them in the editor