| 摘要 |
<p>PURPOSE: To allocate a speech part to the word of a message by optimizing the product of the probability of the vocabulary of individual words and the probability of the contexts of standardized three words. CONSTITUTION: A stage for generating the probability of the vocabulary expresses a specified word, spaces in front and at the back of a sentence and the sign of punctuation marks as the words provided with an empirically decided generation frequency in the case of the recording of the non-word of the message in a speech part tag attaching part 31 and smoothes at least the frequency of the speech part relating to the specified word. Then, the stage for selecting the probability of the context maximizes the probability of the context relating to the speech part of the close word provided with the next word at least in a noun phrase dissection part 32. Thus, the speech part is identified at the probability of a high level.</p> |
