A method and apparatus (500) for word segmentation, said method comprising: acquiring a pre-determined word set and a text to be segmented (201), the pre-determined word set being a word set generated in advance on the basis of a pre-determined text set, and words in the pre-determined word set comprising first information and second information; performing word segmentation on said text on the basis of the pre-determined word set, so as to obtain at least one word sequence (202); determining first information and second information about words in a word sequence in at least one word sequence, and determining the probability of the word sequence on the basis of the determined first information and second information (203), the second information about the word in the word sequence being second information determined on the basis of a word adjacent to the word; selecting, from at least one word sequence, a word sequence having the largest probability as a word segmentation result (204). The method and apparatus improve the accuracy of the word segmentation.