What is CRF?

CRF or Conditional Random Field is a graphical model based on probability that is used to predict by taking sequential data or patterns into account. In NER, CRF can take labels of words in a sentence in a sequential manner to predict the entities. Though CRF cannot handle the features of the forward labels in a sequence, it is very less used and the algorithm seems to be very typical among other statistical models.

The above representation shows how the algorithm in CRF works.