JPH06236395A - Word processor with translation conversion function - Google Patents

Abstract

PURPOSE:To easily generate the translation of long sentences by using the translation patterns of the long sentences in a word processor with translation conversion function. CONSTITUTION:In a long sentence pattern collation part 6, a morpheme analyzed result which is the output of a morphemic analyzing part 3 is inputted and expressions which coincide with the long sentence patterns of a source language inside a long sentence pattern dictionary 9 are searched. When such pattern is present, the kind and the position of the pattern are delivered to a long sentence pattern conversion part 7 along with the morpheme result. In the long sentence pattern conversion part 7, the translation patterns corresponding to the long sentence pattern of the source language are taken out from the long sentence pattern dictionary 9 by the delivered kind of the pattern and when plural translation candidates are present, all of them are outputted to a long sentence pattern candicate display part 8. In the long sentence pattern candidate display part 8, the long sentence patterns of input are displayed to a user. At this time, the presentation of conventional translation conversion candidates by the morphemic analyzing part 3, an object word expression generation part 4 and an object word expression stage display part 5 is simultaneously performed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a word processing device having a translation conversion function, and more particularly to a word processing device having a translation conversion function having a long sentence pattern conversion function independent of conventional translation conversion processing.

[0002]

2. Description of the Related Art A conventional word processor having a translation conversion function has a dictionary containing various language information relating to a source language and a target language for an input source language expression, and the application of these language information to the input is performed in order. Or do it all at once.
Therefore, if you input a long sentence using these devices,
Although it is relatively easy to select word candidates and dependencies in a narrow range in single sentences, etc., it is possible to select translation candidates and dependencies in a wide range such as dependency relations for each clause such as compound sentences and compound sentences. Was relatively difficult. For example, Japanese Patent Laid-Open No. 2-24
In the natural language translation device by conversational input described in Japanese Patent No. 4273 "Natural language translation method and device", the grammatical element of the sentence, its range / break, the semantic element,
It holds information about many languages, such as its range / breaks / structure and its role, and has an interactive input means to aid in their application, but the application of these information is sequential. Yes, if you translate according to this sequential input, the correct translation result will be generated in a relatively narrow range, but from the global perspective, the correct translation result has not been made, or the correct translation result will be generated. It has a drawback that it takes a lot of effort to generate.

[0003]

In this conventional word processor having a translation conversion function, it is not possible to present / select translation candidates in a narrow range and to present / select global translation candidates simultaneously in parallel. There's a problem.

An object of the present invention is to provide a word processor having a translation conversion function that solves such a problem.

[0005]

According to the present invention, a source language input section for allowing a user to input a source language expression by an input method similar to that of a word processor, a translation dictionary holding language information for translation, and the translation dictionary. Morphological analysis unit that divides the source language expression input by the source language input unit into morphemes with reference to, and inputs the source language expression of the morphological analysis result from the morpheme analysis unit, and refers to the translation dictionary to An object expression generation unit that generates object expression candidates corresponding to the source language expression;
An object expression stage display unit for presenting to the user the candidate for the object expression output from the object expression generation unit and selecting according to the user's behavior, and outputting the object expression selected in the object expression stage display unit In a word processing device with a translation conversion function having an object output unit, a long sentence pattern dictionary holding a vocabulary pattern and its corresponding translation pattern capable of estimating a dependency pattern of a long sentence such as a compound sentence and a heavy sentence in units of a clause, A long sentence pattern matching unit that matches a long sentence pattern in a long sentence pattern dictionary with an input source language expression subjected to morphological analysis by the morpheme analysis unit and sets a long sentence pattern, and a long sentence pattern certified by the long sentence pattern matching unit. A long sentence pattern conversion unit for converting a sentence into a candidate for a pattern of an object expression using information in the long sentence pattern dictionary Characterized in that it has a long sentence pattern candidate display unit configured to select the behavior of the user presenting the candidates of long sentence translation pattern that has been converted in the long sentence pattern conversion unit to the user.

[0006]

Embodiments of the present invention will now be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. The word processing device with the translation conversion function of this embodiment is
A source language input unit 1 that allows a user to input a source language expression by an input method similar to a word processor, a translation dictionary 2 that holds language information for translation, and a source language input unit 1 that refers to the translation dictionary 2 A morpheme analysis unit 3 that divides the generated source language expression into morphemes, and a source language expression candidate obtained by inputting the source language expression as a result of the morpheme analysis by the morpheme analysis unit 3 and referring to the translation dictionary 2 An object expression generation unit 4 for generating the object expression, an object expression expression display unit 5 for presenting a candidate of the object expression output from the object expression generation unit 4 to the user, and selecting it according to the user's behavior, and an object expression expression step. Object output unit 1 for outputting the object expression selected on the display unit 5.
0, a long sentence pattern dictionary 9 that holds a vocabulary pattern that can estimate a dependency pattern of a long sentence such as a complex sentence and a heavy sentence and the corresponding translation pattern, a long sentence pattern in the long sentence pattern dictionary 9, and a morphological analysis unit 3 The long sentence pattern matching unit 6 that matches with the input source language expression subjected to morphological analysis to authenticate the long sentence pattern, and the long sentence pattern certified by the long sentence pattern matching unit 6 using the information in the long sentence pattern dictionary as the object word. The long sentence pattern conversion unit 7 that converts the expression pattern candidates and the long sentence pattern candidate display unit 8 that presents the long sentence translation pattern candidates converted by the long sentence pattern conversion unit 7 to the user and selects them according to the user's behavior It

In the long-sentence pattern dictionary 9, not only the dependency relation between words in a narrow range such as the case relation but the global dependency relation in the sentence such as the clause dependency relation, Focusing on the vocabulary that has the function of defining the structure of a compound sentence, the patterns of those vocabularies, their dependency relationships, and translation patterns are retained. As a technique of Japanese sentence analysis using a vocabulary that has a function of defining the structure of this long sentence, paying attention to a functional word such as a connective particle, "strength as a break in the sentence" about the functional word. Lexical context grammar using information such as (Shinichiro Kamei, Kazuto Muraki: “Lex
Proposal of the "ICal Discosure Grammer", Institute of Electronics, Information and Communication Engineers, Language Understanding and Communication Study Group, 456-7, pp. 1-5 (1986)) and a context estimation method using the same (Japanese Patent Laid-Open No. 62-2320)
78), context analysis method (Japanese Patent Laid-Open No. 63-71744).
Issue). FIG. 2 shows a table summarizing information on connection forms between adjectives such as connective particles based on what elements can be included in a clause in a section to which each connective form is attached. In this table, the smaller the level number, the stronger the ability as a break in the sentence.

The long sentence pattern matching unit 6 receives the morpheme analysis result output from the morpheme analysis unit 3 as an input and searches the long sentence pattern dictionary 9 for an expression that matches the long sentence pattern of the source language. If there is such a pattern, the long sentence pattern conversion unit 7 determines the type and position of the pattern together with the morpheme result.
Hand over to. The long sentence pattern conversion unit 7 extracts a translation pattern corresponding to the long sentence pattern of the source language from the long sentence pattern dictionary 9 according to the type of the passed pattern, and if there are a plurality of translation candidates, all of them are displayed in the long sentence pattern candidate display unit. Output to 8. The long sentence pattern candidate display unit 8 presents the input long sentence pattern candidates to the user. At this time, the conventional translation conversion candidates are presented at the same time by the morphological analysis unit 3, the object expression generation unit 4, and the object expression stage display unit 5. The user can make a selection for one or both of these candidates, if desired.

When the user pays attention to the generation of translated sentences in a narrow range, the long sentence pattern candidates may be left as they are displayed. At this time, since the translated sentence in a narrow range is determined, when the number of candidates or the priority of the long sentence pattern is changed, the change is also reflected in the long sentence pattern candidate to be presented.

On the contrary, when the user pays attention to the global translation sentence generation, the candidates in the object representation stage display section 5 by the conventional translation may be left as they are displayed. Also at this time, since the translation candidates for the long sentence pattern have been determined, when the number of translation candidates or the priority in the narrow range is changed, the translation in the narrow range in which the change is presented is presented. It is also reflected in the candidates. For this reason, the user can see the translation that he or she is trying to make, both globally and in detail, and by deciding the translation while paying attention to either one, he / she will not be aware of the other candidate. It becomes possible to generate a more accurate translated sentence more easily by giving information to.

The global translation result thus determined and the translated text in a narrow range are put together, and the target expression is output as the conversion result of the word processor in the object output section.

Next, the operation of this device will be described using an example sentence. First, consider a situation in which the user inputs "I asked while laughing because he asked, and I answered." In this sentence, the subject of "asked" is "he is" and the subject of "laughing" is "I". Such a correspondence between the subject and the predicate can be determined by the strength of the connecting particles "node" and "while" in the sentence. The connective particles “node” and “while” are shown in FIG.
It is presumed that "he ga" is not suitable as the subject of "laughing" because "do" has a stronger ability as a sentence break. Therefore, in the translation conversion according to the present invention, it is presumed that the subject is input after "laughing" and the predicate sharing the subject by "while" is expressed by (it did), and "Because he
... ed, (it did) while ... ing ”is presented. Also, as a conventional translation conversion executed in parallel, as a candidate for "ask", "ask, inquire, questio
“N” and the like are presented as “laugh, smile” and the like as candidates for “laugh”. If the user feels comfortable with this suggestion, then enter "I answered."
he asked, I replied while laughing. "

In another situation, the user is conscious of the sentence structure in English and "laughs because he asked."
It is assumed that "he was" was intended as the subject of "laughing". In this case as well, the input character string is the same as in the above scene, so the first candidate is also the same as in the above example. However, in the above example, the subject of "laughing" is not "he is", so the user selects the next candidate. Then, the candidate whose subject of "laughing" is "he is", "Because he ... ed while
... ing, ... "candidates are presented. The user has this subject predicate relationship,
Select this candidate. After that, when I typed "I answered," finally, "Because he asked while laughing,
I replied. " At this time, since the user has already selected the next candidate, it was found that the subject "I answered" was not the subject of the predicate before that, and only the subject predicate relation of "answered" was found. It can be easily estimated to have, and analysis in a narrow range can be performed more accurately.

[0014]

As described above, in the word processor having the translation conversion function according to the present invention, by using the translation pattern of a long sentence, translation generation of a long sentence is facilitated, and
By performing global translation by long sentence pattern and translation in a narrow range independently and presenting those candidates to the user,
This has the effect of allowing the user to select one or both of them as necessary and performing each analysis more accurately.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a table summarizing information that classifies the connection forms between adjectives, which are held in the long sentence pattern dictionary, based on what kind of element can be included inside the clause in the clause to which each connection form is attached. It is a table.

[Explanation of symbols]

 1 source language input unit 2 translation dictionary 3 morphological analysis unit 4 object expression generation unit 5 object expression stage display unit 6 long sentence pattern matching unit 7 long sentence pattern conversion unit 8 long sentence pattern candidate display unit 9 long sentence pattern dictionary 10 object sentence output unit

Claims (3)

[Claims] 1. A source language input unit for allowing a user to input a source language expression by an input method similar to a word processor, a translation dictionary holding language information for translation, and the source language input with reference to the translation dictionary. A morpheme analysis unit that divides the source language expression input by the unit into morphemes, and the source language expression of the morphological analysis result is input from the morpheme analysis unit, and the object word corresponding to the source language expression by referring to the translation dictionary An object expression generation unit for generating expression candidates, an object expression stage display unit for presenting to the user candidates for object expression output from the object expression generation unit, and selecting according to the user's behavior; In a word processing device with a translation conversion function that has an object output unit that outputs the selected object expression in the expression display unit, it is possible to estimate the dependency pattern of a long sentence such as a compound sentence or a compound sentence in units of clauses. A long sentence pattern dictionary holding a vocabulary pattern and its corresponding translation pattern, and a long sentence pattern in the long sentence pattern dictionary are matched with an input source language expression morphologically analyzed by the morphological analysis unit to set a long sentence pattern. A long sentence pattern matching unit, a long sentence pattern converting unit that converts the long sentence pattern recognized by the long sentence pattern matching unit into a candidate of a pattern of an object expression using information in the long sentence pattern dictionary, and the long sentence pattern converting unit. A word processing device with a translation conversion function, comprising: a long sentence pattern candidate display unit that presents converted long sentence translation pattern candidates to a user and selects the long sentence pattern candidate according to the user's behavior. 2. The object word output unit selects one or both of a translation conversion candidate presented in the object word expression stage display unit and a long sentence pattern candidate presented in the long sentence pattern candidate display unit, from a user. The word processing device with a translation conversion function according to claim 1, wherein the output is made by selection. 3. A vocabulary having a function of defining a structure within a sentence of a compound sentence or a compound sentence having a global dependency relation such as a clause dependency relation in the long sentence pattern dictionary. The word processing device with a translation conversion function according to claim 1 or 2, wherein the vocabulary patterns, their dependency relations, and translation patterns are retained.