JPH0512250A - Document preparation device - Google Patents

Abstract

PURPOSE:To effectively display a concurrence function and to make a KANA (Japanese syllabary)-KANJI (Chinese character) conversion processing to be efficient by realizing a learning function which always registers appropriate coocurrence relation information. CONSTITUTION:A coocurrence example learning processing part 12 divides a conversion result selected from a conversion candidate from a KANA-KANJI conversion part 5 into plural words consisting of two clauses. A rule extraction processing part 13 refers to a previously prepared modification relation rule table 14a and extracts the plural words applied to a modification relation in the selected conversion result. The concurrence example learning processing part 12 registers the extracted plural words in a cooccurrence information registration table 8.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document preparation apparatus having a co-occurrence function for efficiently implementing Kana-Kanji conversion processing in consideration of the meaning of words.

[0002]

2. Description of the Related Art Conventionally, in a document creating device which is a Japanese word processor, a kana-kanji conversion process is performed for converting a kana character string input from a keyboard into a kanji mixed sentence. In this kana-kanji conversion process, a plurality of conversion candidates corresponding to the same phoneme are generated and displayed. Normally, a required specific conversion candidate is selected from a plurality of generated conversion candidates by operating a specific key of the keyboard, for example, a select / execute key.

By the way, some kana-kanji mixed sentences have a relation (called a co-occurrence relation) in advance that is easily semantically combined with the preceding and following words. For example, if Kana-Kanji conversion processing is performed on the input kana character string "Natsu wa Atsu", conversion candidates such as "summer is thick", "summer is hot", and "summer is hot" are output. Is expected. in this case,
The words "hot" and "thick" in contrast to the previous word "summer" are words that are not normally used. On the contrary,
"Summer is hot" is a sentence that is semantically easy to combine, that is, co-occurring. Besides this, co-occurrence sentences include:
For example, there is “a house is built”.

A co-occurrence information registration table in which co-occurrence relationship information having such a co-occurrence relationship is registered is prepared in advance, and conversion candidates registered in the co-occurrence information registration table are preferentially output in the kana-kanji conversion processing. A document creation device having a co-occurrence function has been developed.

However, the amount of information registered in the co-occurrence information registration table is limited, and the registered information is not always appropriate depending on the field of the created document. Therefore, when a desired word (same word) is selected from the conversion candidates obtained by the kana-kanji conversion process, a method having a learning function of registering the adjacent word together with the adjacent word as co-occurrence relationship information is considered. Be done.

According to this method, the co-occurrence relation information required by the user can be registered. However, if two adjacent words are simply registered as co-occurrence relation information, an inappropriate conversion result in the subsequent Kana-Kanji conversion process. May be output.
Specifically, for example, when a document "The degree of contribution to my fire extinguishing ..." is created, the co-occurrence relationship information "My fire extinguishing" may be registered. However, this co-occurrence relationship information of "my fire extinguishing" is inappropriate, and if such co-occurrence relationship information is registered, the efficiency of the Kana-Kanji conversion processing will rather be lowered.

[0007]

A co-occurrence function can be improved by a learning function of registering co-occurrence relationship information based on a selected word and an adjacent word, but on the other hand, an inappropriate co-occurrence relationship can be obtained. Registering information may cause a decrease in the efficiency of the Kana-Kanji conversion process.

An object of the present invention is to effectively realize a co-occurrence function by realizing a learning function of always registering appropriate co-occurrence relation information, and as a result, to improve efficiency of kana-kanji conversion processing. It is to provide a document creation device capable of

[0009]

SUMMARY OF THE INVENTION According to the present invention, in a document creation apparatus having a co-occurrence function that utilizes co-occurrence relationship information registered in a co-occurrence information registration table, a co-occurrence information registration table is referenced. Kana-Kanji conversion processing means for preferentially outputting conversion candidates for origin relations, dependency relation analysis means for extracting a plurality of words corresponding to dependency relations from the selected conversion result, and dependency relation analysis means for extraction And a registration unit that registers the compound word in the co-occurrence information registration table.

[0010]

According to the present invention, after selecting the conversion result from the kana-kanji conversion processing means, referring to the dependency relationship rule table prepared in advance, a plurality of dependency relationships among the selected conversion results are selected. Extract words. The extracted plural words are registered in the co-occurrence information registration table means and are used as co-occurrence relation information during the kana-kanji conversion processing.

[0011]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the arrangement of a document creating apparatus according to the embodiment. The present apparatus includes a control / editing unit 1 for executing control of the entire apparatus and document creation / editing processing,
A keyboard 2, an input control unit 3, an input data memory 4, a Kana-Kanji conversion unit 5, a conversion result memory 7, a display 9 and a display control unit 10 are provided.

The keyboard 2 has a selection / execution key 2a and a conversion / next candidate key 2b in addition to various character keys, cursor keys, and various function keys. The input control unit 3 outputs a key code corresponding to a key input from the keyboard 2. The input data memory 4 is a register for temporarily holding a hiragana character string input from the keyboard 2. The kana-kanji conversion unit 5 refers to the kana-kanji conversion dictionary 6 to convert the hiragana character string stored in the input data memory 4 into a kanji mixed sentence. The conversion result memory 7 stores the kana-kanji conversion result (conversion candidate) by the kana-kanji conversion unit 5.

The display controller 10 displays the hiragana character string stored in the input data memory 4 and the kana-kanji conversion result stored in the conversion result memory 7 on the screen of the display 9.

The present apparatus further includes a co-occurrence information registration table 8, a selection result memory 11, a co-occurrence example learning processing unit 12, a rule extraction processing unit 13, and a rule table 14. The co-occurrence information registration table 8 is a table (read / write memory) for registering co-occurrence relation information necessary for the co-occurrence function. The selection result memory 11 is a register for storing the conversion result selected and determined by the operation of the selection / execution key 2a of the keyboard 2 from the conversion candidates of the kana-kanji conversion result.

The co-occurrence example learning processing unit 12 determines the selection result memory 1 based on the determination processing result of the rule extraction processing unit 13.
A process of registering a plurality of words (a plurality of words having a dependency relationship) extracted from the conversion result stored in 1 in the cooccurrence information registration table 8 is performed. The rule extraction processing unit 13 refers to the rule table 14 to determine whether the conversion result extracted by the co-occurrence example learning processing unit 12 is applied to a co-occurrence relation rule (dependency relation rule or the like). judge. The rule table 14 includes a dependency relationship rule table 14a and a modification relationship rule table 14b, which are cooccurrence relationship rules, and is prepared in advance. Next, the operation of the embodiment will be described.

As shown in step S1 of FIG. 2, when a hiragana character string is input from the keyboard 2, the control / editing unit 1 stores the input character string in the input data memory 4 (step S2). The display control unit 10 displays on the screen of the hiragana character string display 9 stored in the input data memory 4.

The kana-kanji conversion unit 5 refers to the kana-kanji conversion dictionary 6 to convert the hiragana character string stored in the input data memory 4 into a kanji mixed sentence (step S3). The control / editing unit 1 stores the Kana-Kanji conversion result of the Kana-Kanji conversion unit 5 in the conversion result memory 7 (step S4). The kana-kanji conversion result consists of a plurality of conversion candidates (same words). The display control unit 10 displays the Kana-Kanji conversion result stored in the conversion result memory 7 on the screen of the display 9.

Here, the select / execute key 2 of the keyboard 2
When a plurality of conversion candidates are selected by the operation of a (YES in step S6), the control / editing unit 1 stores the conversion result selected and determined in the selection result memory 11 (step S7). Specifically, as shown in FIG. 4 (A), a selection result such as “The program is installed in the extended memory.” Is stored in the memory 11 from the kana-kanji conversion result and displayed on the screen of the display 9. Is displayed.

The co-occurrence example learning processing unit 12 is activated in response to the operation of the selection / execution key 2a, and the selection result memory 11 is selected so that the rule determination processing of the rule extraction processing unit 13 becomes possible.
A predetermined processing process is executed on the selection result stored in. Specifically, as shown in FIG. 4A, the selection result is divided from the first phrase to the fourth phrase. Furthermore, as shown in FIG. 4B, the grammatical structure is analyzed for each clause (step S8).

Based on the analyzed grammatical structure, the co-occurrence example learning processing unit 12 extracts all combinations of two phrases as a result of selection (step S).
9). One is extracted from all the extracted combinations and is output to the rule extraction processing unit 13 (step S1).
0, YES in S11). For example, a combination of "program" and adjacent "extension" is extracted.

The rule extraction processing unit 13 refers to the rule table 14, and determines whether the combination (plural words) extracted by the co-occurrence example learning processing unit 12 corresponds to the dependency relation rule or the modification relation rule. Is determined (step S12). Here, as shown in FIG. 5A, the dependency-related rule table 14a prepares rule information that may be dependent on the grammatical structure. Further, as shown in FIG. 5B, the modification relation rule table 14b prepares rule information that may have a modification relation between adjacent clauses in the grammatical structure.

The rule extraction processing unit 13 extracts combinations corresponding to the rules of the dependency relation rule table 14a from the respective combinations shown in FIG. 4C and outputs them to the co-occurrence example learning processing unit 12. Co-occurrence example learning processing unit 12
Registers the extracted combination in the co-occurrence information registration table 8 (step S14).

Concretely, for example, a combination of "incorporated (incorporated) program" corresponding to rule 5 shown in FIG. 5A is stored in the cooccurrence information registration table 8 as shown in FIG. register. Other than this, for example, "to extended memory", "embedded in memory (embedded)"
Each combination of is registered in the co-occurrence information registration table 8. Here, the extracted combination does not correspond to the rule of the modification relation rule table 14b.

By the way, before the rule determination processing by the rule extraction processing unit 13, there are cases where there is a contradiction or an inappropriate one in the combination when the rules of the dependency relation are applied. Therefore, step S15 of FIG. As shown in, the co-occurrence example learning processing unit 12 may perform a check process of whether or not there is a contradiction (inappropriate). If it exists (YES in step S16), the combination is deleted and excluded from the target of the rule determination processing by the rule extraction processing unit 13 (step S17).

As a concrete example, it is assumed that the selection result memory 11 stores the selection result of "I watched the situation with a life-threatening feeling." In this example, there are intersecting dependency relationships such as "I shrink" and "In the thought of life." Further, for example, in the selection result of "the destination of her is unknown.", There is a clause having no dependency relation such as "her". Furthermore, in the selection result of "I went to the park because the cherry blossoms have bloomed", an inappropriate dependency relationship such as "I bloomed" is extracted.

In this way, with respect to the conversion result selected from the kana-kanji conversion candidates, by referring to the dependency relationship rule table 14a prepared in advance, whether or not there is any that applies to the dependency relationship rule. The determination process is performed. A plurality of words (combinations) applied to the dependency relationship rule are registered in the co-occurrence information registration table 8 as co-occurrence relationship information. By referring to the co-occurrence information registration table 8, the conversion candidates corresponding to the registered co-occurrence relationship information can be preferentially output in the next Kana-Kanji conversion process. In other words, a learning function of registering appropriate co-occurrence relation information can be realized, and the co-occurrence function can be improved.

Further, the dependency relation rule table 14a
Besides, by preparing the modification relationship rule table 14b, not only the dependency relationship but also the co-occurrence relationship information of the modification relationship can be registered.

[0029]

As described in detail above, according to the present invention, it is possible to realize a learning function of registering appropriate co-occurrence relation information by utilizing the kana-kanji conversion result. Therefore, by utilizing the registered co-occurrence relation information obtained by the learning function, the co-occurrence function can be effectively exhibited, and as a result, the efficiency of the kana-kanji conversion processing can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a document creation device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the embodiment.

FIG. 3 is a flowchart showing the operation of the embodiment.

FIG. 4 is a conceptual diagram for explaining the operation of the embodiment.

FIG. 5 is a conceptual diagram for explaining the operation of the embodiment.

FIG. 6 is a conceptual diagram for explaining the operation of the embodiment.

[Explanation of symbols]

1 ... Control / editing unit, 5 ... Kana-Kanji conversion unit, 8 ... Co-occurrence information registration table, 12 ... Co-occurrence example learning processing unit, 13 ... Rule extraction processing unit, 14 ... Rule table.

Claims (1)

Claims: 1. A co-occurrence information registration table means for registering co-occurrence relation information, and a kana-kanji conversion of an input hiragana character string. With reference to the kana-kanji conversion processing means that preferentially outputs the conversion candidates of the origin relationship, and the dependency relationship rule table prepared in advance,
Dependency relationship analysis means for extracting a plurality of words corresponding to the modification relationship from the conversion results selected from the conversion candidates output by the kana-kanji conversion processing means, and the modification relationship analysis means And a registration unit that registers the plurality of words in the co-occurrence information registration table unit.