Hitachi, Ltd.

Yoshihiko Nitta

Hiroyuki Kaji

Fumiyuki Yamano

Atsushi Okajima

Hirokazu Ihara

Yoshihiro Shima

Toru Hisamitsu

Katsumi Marukawa

Fumon Shimizu

Ryota Kaijima

Hiroaki Shimokawabe

Atsuko Isatsu

A character recognition method is arranged to supplement an erroneously recognized character with a linguistic knowledge. In this method, the extraction of a candidate based on the search of a word dictionary occupies a large part of the operation. To speed up the extraction of a candidate word, therefore, the method is provided to search the dictionary by using a group of candidate characters or a dictionary header for processing an inflected form of a verb. Further, the present method is provided for calculating a word matching cost for improving an efficiency of modifying an erroneous recognition. The word search is done by using a 'hybrid method' arranged of 'candidate-character-driven word extraction' and 'dictionary-driven word extraction'. Moreover, the word-dictionary is arranged to have a header word composed of an inflectional ending of a verb and an auxiliary verb or a particle added to the tail of the inflectional ending. The present method attaches much importance to the difference of the matching cost about a character with a totally high confidence ratio than the difference of the matching cost about a character with a totally low confidence ratio.

Character recognition method and apparatus

A translation method for a machine translation system provided with apparatus for parsing a source language sentence and for forming a target language translation in which a phrase omitted in the source language setence is identified, and a word or phrase to be inserted for the omitted phrase is selected from stored words and phrases. For identifying an omitted phrase, a sentence pattern corresponding to a predicate in the source language sentence is formed so as to include not only cases governed by the predicate but also a semantic feature for each case. By comparing the source language sentence with the sentence pattern, a case which is omitted in the source language sentence but cannot be omitted in the target language translation is identified. For determining a word or phrase to be placed at the position of the omitted phrase, the nouns having appeared in the source language text is stored in a noun, together with the semantic feature, gender, person and number of each noun is searched for a noun having the same semantic feature as the omitted phrase. When a target language translation of the source language sentence is formed, a pronoun having the same gender, person and number as the omitted phrase is used as a target language equivalent for the omitted phrase, and thus a target language translation which is grammatically correct, is obtained.

System for automatic language translation using several dictionary

In machine translation from a first language text such as a Japanese text to a second language text such as an English text, if information necessary for the translation is not directly expressed or is not sufficiently hinted at in the first language text, such information is requested by the translation machine to supplement the first language during inputting of the first language text or pre-editing thereof.

Method and apparatus for supplementing translation information in

An information storage/retrieval system stores information to be retrieved in a storage device having a first memory and a second memory. The first memory stores therein main records each containing an index word data, a derivative pattern code and information pieces while the second memory stores therein auxiliary records each containing a derivative pattern code and supplemental word data. Each derivative pattern code represents a group of words whose primary parts such as word stem parts are commonly combinable with the same remaining parts such as affix parts. Thus, each index word data is qualified by the accompanying derivative pattern code, and the supplemental word data includes data corresponding to the above-mentioned remaining parts. An information retrieval is made by searching in a first memory with respect the primary part of a keyword taken from, for example, its head and then in a second memory with respect to the remaining part of the keyword, thereby identifying the keyword. Relevant information to be outputted is found in at least one of the first and second memories on the basis of a supplmental word data in the second memory.

Method and system for information storing and retrieval using word stems

A method of segmenting a text into words in which a dictionary search is made while using a character string in the text as a search key, and it is checked whether a word retrieved from the dictionary can be grammatically connected to another word adjacent thereto or not. Segmentation processing is carried out using only words registered in a word dictionary, processing for identifying an unknown word is carried out when the segmentation processing comes to a deadlock, and then the segmentation processing is continued for that portion of the text which follows the identified unknown word.

Method for segmenting a text into words

An automatic translation method between natural languages comprises assigning parts of speech to words/idioms of an input text sentence by looking up a lexicon storage. The input text sentence in the form of string of parts of speech is segmented into phrasal elements as minimum units having linguistic meaning to assign parts of speech to the respective elements. The sequence of phrasal parts of speech is converted to a string of syntatic roles to the respective phrasal elements and words/idioms. Patterns representing a sentence, a clause and a quasi-clause are detected from the sequence of syntatic roles to transform the input text sentence to a skeleton pattern represented by a combination of those patterns. The sequence of the simple sentence, clause and quasi-clause inherent to the input language which forms the skeleton pattern is transformed to a sequence inherent to the output language. The sequence of syntatic roles of the simple sentence, clause and quasi-clause inherent to the input language which forms the transformed skeleton pattern is transformed to a sequence inherent to the output language. Finally, target language equivalents are assigned to the transformed string of syntatic roles.

Method for automatic translation between natural languages

In a system wherein a first text in a first natural language is translated into a second text in a second natural language; a text displaying and correcting system comprising a first memory area for storing parts of the first text divided in predetermined units, with identifications assigned to the respective parts, and a second memory area for storing predetermined units of the second text corresponding to the aforementioned units of the first text, with the same identifications assigned thereto, so that the first and second texts are simultaneously displayed on a screen of a display unit, and that the text is revised in each unit with identification assigned.

Displaying and correcting method for machine translation system

An input sentence described by a first natural language is sectioned into individual words. Parts of speech corresponding to the individual words are retrieved from a lexical word storage, whereby the input sentence is described by a corresponding string of the parts-of-speech as retrieved. A translation pattern table is previously prepared which defines correspondence between patterns of strings of parts-of-speech for the first natural language and those for a second natural language by which corresponding output sentence is to be described. By referring to the translation pattern table, the string of the parts-of-speech of the input sentence is transformed into a corresponding string of the parts-of-speech for the second natural language. The output sentence described by the second natural language is generated by sequencing target words in accordance with the sequential order of the parts of speech of the string pattern obtained after the transformation.

Growing community of inventors

Sagamihara, Japan

Yoshihiko Nitta