what does the item_bias mean?

[kiminh]

hi, I'm confused with the parameter item_bias. I just read the original paper("BPR").
self.item_bias = np.zeros(self.num_items)
Would you please figure out the related formula in the original paper?

And does the "bias" like in LR(logistic regression): wx+b？ In MF(matrix factorization), I'm not clear about the bias.