Neural networks as possible hyphenation technique for Afrikaans

摘要

In Afrikaans compound words are written as one word. New words are therefore created by simply joining words. Word hyphenation during typesetting by computer is often a problem, because the source of reference changes all the time. A neural network (feedforward backpropagation) was trained with about 5 000 Afrikaans words with correct syllabification. The neural network classified 97,56% of possible points in 5 000 randomly chosen words correctly as either valid or invalid hyphenation points. In a test with 510 words from an Afrikaans magazine the neural network classified 98,75% of possible positions correctly. We came to the conclusion that neural networks can be used successfully as hyphenation technique for Afrikaans.