A Kana-Kanji Translation System for Non-Segmented Input Sentences Based on Syntactic and Semantic Analysis

A Kana-Kanji Translation System for Non-Segmented Input Sentences Based on Syntactic and Semantic Analysis Masahiro ABE, Yoshimitsu OOSHIMA, Katsuhiko YUURA and Nobuyuki TAKEICHI Central Research Laboratory, Hitachi,.Ltd. Kokubunji, Tokyo, Japan Abstract This paper presents a disambiguation approach for translating non-segmented-Kana into Kanji. The method consists of two steps. In the first step, an input sentence is analyzed morphologically and ambiguous morphemes are stored in a network form. In the second step, the best path, which is a string of morphemes, is selected by syntactic and semantic analysis based on case grammar. In order to avoid the combinatorial explosion of possible paths, the following heuristic search method is adopted. First, a path that contains the smallest number of weighted-morphemes is chosen as the quasi-best path by a best-first-search technique. Next, the restricted range of morphemes near the quasi-best path is extracted from the morpheme network to construct preferential paths. An experimental system incorporating large dictionaries has been developed and evaluated. A translation accracy of 90.5% was obtained. This can be improved to about 95% by optimizing the dictionaries. 1. INTRODUCTION Ordinary Japanese sentences are written using a combination of Kana, which are Japanese phonogramic characters, and Kanji, which are ideographic Chinese characters. Nouns, verbs and other independent words are generally written in Kanji. On the other hand, dependent words such as postpositons, and auxiliary verbs, etc., are written in Kana. While there are about fifty Kana, there are several thousand Kanji, thus making it difficult to input Japanese sentences into a computer system. Extensive research has been carried out on methods of inputting Kanji in an attempt to .realize rapid and easy input. Among the methods proposed, Kana-Kanji translation appears to be the most promising. In this method, input sentences are entered in Kana using a conventional typewriter keyboard, and those parts of the sentences which should be written in Kanji are translated into Kanji automatically. In this process a non-segmented input form is desirable for users because there is no custom of segmentation in writing Japanese sentences. Therefore, the ultimate goal of a Kana-Kanji translation scheme should be to achieve error-free translation from non-segmented Kana input sentences. This paper describes a system for achieving high accuracy in the Kana-Kanji translation of non-segmented input kana sentences. 1.1 Disambiguation Approaches in Kana-Kanji Translation If ambiguity were not a problem in non-segmented input Kana sentences, a perfect Kana-Kanji translation could be easily made using simple transliteration techniques. The fact is that the input Kana sentences are highly ambiguous. The ambiguity of non-segmented input Kana sentences can be categorized into following two types. (a) The ambiguity of segmentation of a sentence into morphemes. example (Input Kana sentence) (Output sentence) [kokodeha iru] (It's necessary here.) [kokode hairu] (Enter here.) (b) The ambiguity of homonyms. example (Kana) (Homonyms) ~ ~- ~ @ i~i~ (atrain) [kisha] (2) N ~J: (your company) @ ~gN (a pressman) @ ~ ~J: (return to office) @ ~ (donate) Makino and Kizawa [1] proposed an automatic Kana-Kanji translation system in which these …

Content truncated to 500 words