Translation processing method and storage medium
A translation processing method executed by a computer, the translation processing method includes calculating a first translation probability from each of first phonemes included in a first document described in a first language into each of second phonemes included in a second document, whose contents substantially equivalent to those of the first document, described in a second language and a second translation probability from each of the second phonemes into each of the first phonemes; extracting a phoneme pair in which the first translation probability and the second translation probability are equal to or higher than a threshold value; and generating translation phrases in the first document and the second document based on the extracted phoneme pair.
1. A translation processing method executed by a computer, the translation processing method comprising:
calculating a first translation probability from each of first morphemes included in a first document described in a first language into each of second morphemes included in a second document, whose contents are substantially equivalent to those of the first document, described in a second language and a second translation probability from each of the second morphemes into each of the first morphemes;
extracting a morpheme pair in which the first translation probability and the second translation probability are equal to or higher than a threshold value; and
generating translation phrases in the first document and the second document based on the extracted morpheme pair,
wherein the extracting process includes specifying a plurality of morpheme pairs in each of which translation probabilities in both directions of the first translation probability and the second translation probability are equal to or higher than the threshold value, and wherein the generating process includes joining the plurality of morpheme pairs into one word to generate the translation phrases,
wherein the translation processing method further includes:
generating a deemed compound word so that the plurality of morpheme pairs are deemed as one word;
identify a third translation probability and a fourth translation probability while the deemed compound word is taken as one morpheme; and
generating translation phrases having the deemed compound word as one word when the third translation probability and the fourth translation probability are higher than the first translation probability and the second translation probability.
2. The translation processing method according to claim 1 , wherein the extracting process includes determining a maximum range deemed as the deemed compound word using translation probabilities in both directions of adjacent ones of the first morphemes and translation probabilities in both directions of adjacent ones of the second morphemes.
3. The translation processing method according to claim 1 , wherein in the extracting process, irrespective of each calculated translation probability, a first translation probability or the second translation probability of a morpheme that is applicable as a particle designated in advance the morpheme is treated to be lower than the threshold value or a first translation probability or a second translation probability of a morpheme that is applicable as a known translation phrase is treated to be equal to or higher than the threshold value.
4. A non-transitory computer-readable storage medium storing a program that causes a computer to execute a process, the process comprising:
calculating a first translation probability from each of first morphemes included in a first document described in a first language into each of second morphemes included in a second document included in a second document in which contents substantially equivalent to those of the first document are described in a second language and a second translation probability from each of the second morphemes into each of the first morphemes;
extracting a morpheme pair in which the first translation probability and the second translation probability are equal to or higher than a threshold value and which is a set of the first language and the second language; and
generating translation phrases in the first document and the second document based on the extracted morpheme pair,
wherein the extracting process includes specifying a plurality of morpheme pairs in each of which translation probabilities in both directions of the first translation probability and the second translation probability are equal to or higher than the threshold value, and wherein the generating process includes joining the plurality of morpheme pairs into one word to generate the translation phrases,
wherein the process further includes:
generating a deemed compound word so that the plurality of morpheme pairs are deemed as one word;
identify a third translation probability and a fourth translation probability while the deemed compound word is taken as one morpheme; and
generating translation phrases having the deemed compound word as one word when the third translation probability and the fourth translation probability are higher than the first translation probability and the second translation probability.