JPH09325960A - Document processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To extract a non-language document or a link destination document as well by extracting a section matched with the specified pattern of format designation information to extract a character string and generating the output document by arranging the extracted section or arranging the post-processed extracted section. SOLUTION: A machine translation system is connected through an input/ output means 2 to a network 1, and an input document 3 is defined as a document with tag inputted from the network 1 through the input/output means 2 or inputted from a user through the input/output means 2. A tag identifying means 4 identifies tag information contained in the input document 3 and extracts a translation object expression containing the tag information, and a translation object tag information storage means 5 stores the translation object tag information to be referred to when the tag identifying means 4 identifies the tag information. A translating means 6 translates the document with tag into a document in a target language, and an output document generating means 7 generates the document with tag in the target language from the output result of the translating means 6.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document processing system, for example, a machine translation for translating a document including tag information for designating a format when the document is displayed and printed out by using a computer system. It is suitable for application to a system.

[0002]

2. Description of the Related Art Recently, computer network systems have been enhanced and electronic documents have become popular. Computer network systems are spread all over the world, and the documents that are distributed include not only their own language (Japanese) but also foreign language documents such as English, which makes the language difference for users who are not familiar with foreign languages. Is a major obstacle to receiving or transmitting information. In this way, users of computer network systems that mix different languages will need to translate the target documents when receiving or transmitting information from overseas, but the cost of translation Is by no means cheap.

Therefore, when receiving or transmitting information from abroad, it is important to reduce the cost of translation by translating the information while limiting the information to the necessary minimum. To limit the information in a document to the minimum necessary, a technique for summarizing the document is necessary.

Conventionally, as an apparatus of this kind, Japanese Patent Laid-Open No. 6-
There is one disclosed in Japanese Patent No. 149876. This device aims to support the creation of an easy-to-understand document by graphically representing the structure of the document, and to make it possible to instantly understand the contents of the existing document. In order to achieve this purpose, (1) a document cutout unit that divides the input document into sentence units, and (2) an input sentence is received from the cutout unit for each sentence, and a keyword is referred to by referring to a predetermined keyword dictionary. (3) A keyword extraction unit that separates the input sentence into word units while extracting the input sentence, and (3) Receives the input sentence from the keyword extraction unit in word units, and standardizes and stores the sentence structure. And extract the structure of the input sentence,
Relationship extraction means that creates an intermediate result consisting of a predetermined relationship symbol that represents the structure of the extracted sentence and words that should be displayed in the figure among the words that make up the input sentence, and (4) the relationship included in the intermediate result. In addition to associating a predetermined graphic with a symbol, determine the size and layout of the graphic and the word to be displayed, and create keyword information in a predetermined format that represents the result of the determination, and (5) receive the graphic information. And a display unit for displaying the input document as a graphic according to the content of the received graphic information.

[0005]

By the way, some electronic documents include tag information (hereinafter referred to as a tagged document) that specifies the format in which the document was printed. , Documents that are composed only of conventional character set codes are distinguished. By indicating the relationship with other documents (hereinafter referred to as links) in the tag information,
There are things that define the connection between documents, emphasize any part in a document, and indicate that any part is a figure or table.

In the case of summarizing and translating this tagged document, in the conventional apparatus having the above-mentioned structure, (1) the constituent elements of the document include not only text but also non-language information such as figures and tables. The information is missing. (2) Since it is not assumed that multiple documents are related by a link, the problem of missing the information of the linked document occurs, and the summarized result is the intention of the entire document. I can't figure out exactly. That is, the conventional apparatus having the above configuration has a problem in that the content of the original document cannot be translated so that it can be accurately understood.

[0007] In the case of simply creating a summary document without executing the translation process, the same problem that the information of the non-language document or the linked document is lost even when the process other than translation is performed on the summary document. Has occurred.

[0008]

In order to solve such a problem, according to the present invention, in a document processing system for processing a document accompanied by format designation information for designating a format at the time of display and print output, (1) An extraction target specific information storage unit that stores a specific pattern of format specification information for extracting a character string of a predetermined type in the input document; and (2) an input target specific information storage unit that stores the extraction target specific information storage unit in the input document. Output document with format specification information identification and extraction means that extracts the part that matches the specified pattern of the format specification information, and (3) prepare the extracted part, or prepare the extracted part that has been processed thereafter. And an output document generating means for generating.

As a result, depending on the storage contents of the extraction target specific information storage means, not only the predetermined portion of the input document but
A non-language document or a linked document can also be taken out, and an output document that reflects such special document information can be formed.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION

(A) First Embodiment Hereinafter, a first embodiment in which the document processing system according to the present invention is applied to a machine translation system for tagged documents will be described in detail with reference to the drawings.

In the machine translation system of the first embodiment, it is considered that the tag information of the tagged document has some intention of the author of the document, and is limited to that part. By performing the abstract, the minimum necessary information of the input document is extracted and translated.

(A-1) Configuration of the First Embodiment The machine translation system of the first embodiment includes, for example, an input device, a processing device, a storage device (including an auxiliary storage device), an output device, and a communication device. Although it is constructed on an information processing apparatus such as a workstation or a personal computer provided, it functionally has the configuration shown in FIG.

In FIG. 1, the machine translation system according to the first embodiment has an input / output unit 2, an input document (buffer) 3,
Tag identification means 4, translation target tag information storage means 5, translation means 6, output document generation means 7 and output document (buffer) 8
This machine translation system is connected to the network 1 via the input / output means 2.

The network 1 is, for example, a network that is connected all over the world, and is capable of exchanging a tagged document between each device (including the machine translation system of the embodiment).

The input / output unit 2 serves as an interface between the machine translation system of the embodiment and the network 1 and an interface with the user, that is,
Input / output of documents with tags is performed. The input document (buffer) 3 is a tagged document input from the network 1 via the input / output means (communication configuration) 2 or a tag input from the user by the input / output means 2 (input configuration). It is an attached document. The tag identifying means 4 identifies tag information included in the input document 2 and extracts a translation target expression including the tag information. The translation target tag information storage unit 5 stores the translation target tag information that the tag identifying unit 4 refers to when identifying the tag information. The translation means 6 translates the tagged document into a document in the target language. The output document generation means 7 generates a tagged document in the target language from the output result of the translation means 6. The output document (buffer) 8 is a target language tagged document generated by the output document generating means 7.

FIG. 2 is an explanatory diagram showing the structure of the translation target tag information stored in the translation target tag information storage means 5.

In the case of a document with a tag, the character string to which the tag information is added is a title, and since it is important, an underline addition, an emphasized font output, etc. are instructed. Is a characteristic character string of. Therefore, if only such a character string part is extracted from the input document,
You can get a summary of tagged documents. This first
In the case of the above embodiment, the user does not necessarily need the information of the entire input document, and obtains a translated document of such a summary document.

The translation target tag information stored in the translation target tag information storage means 5 is information capable of extracting a character string portion that can be an element of the summary document.

In FIG. 2, translation target tag information 200
Is composed of a plurality of (nine in FIG. 2 are specifically shown) tag patterns 201 to 209, .... Each tag pattern defines an expression pattern including tag information that may appear in the input document 3. Each tag pattern has a "."
Alternatively, a regular expression such as “+” indicating one or more repetitions is also included as appropriate, and is a definition of a character string to be collated with the character string that appears in the input document 3.

For example, in the tag pattern 201 "<I>. + </ I>", "<I>" and "</ I>"
The ". +" Sandwiched between "" means one or more repetitions of any one character, and a character string such as "parsing" is matched. In addition, “<H [0-9]>. +
In the tag pattern 202 "</ H [0-9]>", since one arbitrary numeral is collated with "[0-9]", for example, "<H1> Recent Tre"
nd of Research And Develo
The character string "pment </ H1>" is matched. Similarly, each of the other tag patterns 203, ..., 209
There are character strings that match

In FIG. 2, nine tag patterns 201 to 209 are described as the translation target tag information 200, but the translation target tag information 200 includes the input / output means 2.
It is preferable that the user can appropriately add and correct the tag pattern via the. That is, HTML (HyperText Markup Language) that appears in the input document 3
It is preferable that the description content can be arbitrarily changed according to the type of the tagged document such as. This embodiment is
It is assumed that the tag pattern described in the translation target tag information 200 can be edited. Therefore, the tag pattern can be increased or decreased according to the type of the tagged document to be defined.

FIG. 3 is a diagram showing an example of the input document 3 (301) used in the description of the operation described later. The target document in this embodiment is, for example, the tagged document 301 described in HTML. This tagged document 301 is linked to the figure (image) information 302 (<IMG
SRC = "...">) and other HTML description documents 3
Tag indicating the link to 03 (<A HREF
= "...">. . . </A>) and other link tags are included.

FIG. 4 is a diagram showing a display example of the input document 3 (301). When the above-mentioned tagged document 301 is displayed via the input / output unit 2, the display screen 4 shown in FIG.
00. In the display area 401, “<TITLE> Recent Trend ... D of the input document 301 is displayed.
This corresponds to the "environment </ TITLE>" portion. The display area 402 corresponds to the description of the part sandwiched between “<BODY>” and “</ BODY>” of the input document 301. Note that the inserted drawing in FIG. 4 is the same drawing (image) information as that in FIG. 3, and “<IMG SRC =” TRANSFE of the input document 301.
R. It is displayed corresponding to "gif">". Further, in FIG. 4, as a display corresponding to the document 303 of FIG. 3, “<A HREF =” ambiguit of the input document 301 is displayed.
y. html ”> ambiguity problem
"/ A>" corresponding to the underlined "ambiguity problem" portion of the display area 402, and the entity of the document 303 that is the link destination is not displayed. As described above, in the tagged document, the diagram may be embedded in the document by the link tag and displayed, or conversely, only the link information may be displayed, and the linked document may not be hidden and displayed.

The input document 301 (3) of this example is
An output document 8 (900) processed by the machine translation system of the first embodiment is shown in FIG. 9, and the output document 8 (90) is shown.
The display screen of 0) is shown in FIG.

(A-2) Operation of the First Embodiment Next, the operation of the machine translation system of the first embodiment for translating a tagged document as described above will be described.

FIG. 5 is a flow chart showing the operation of the entire machine translation system of the first embodiment.

As described above, the machine translation system of the first embodiment considers that the portion of the tagged document to which the tag information is added has some intention of the author of the document, and only that portion is considered. By performing the abstraction limited to, the minimum necessary information of the input document 3 is extracted and translated. Therefore, the tag information included in the input document 3 is focused and the tag information is added. The above expression is used as a translation target and translated, and the output document 8 is generated and displayed.

First, the tag identifying means 4 inputs the input document 3 with a tag from the network 1 via the input / output means 2 (step 501). Since the input document 3 includes the tag information, the tag identifying means 4 obtains the translation target tag information 200 from the translation target tag information storage means 5, identifies the translation target of the input document 3, and identifies it. It is passed to the translation means 6 (step 502). The translation unit 6 translates the translation target obtained from the tag identification unit 4, and passes the result to the output document generation unit 7 (step 503). The output document generation means 7
The translation result obtained from the translation means 6 is shaped with reference to the input document 3 and stored as the output document 8 (step 50).
4).

The operation of these steps 501 to 504 is as follows.
The tag identification means 4 processes all of the input document 3, the output document generation means 7 receives all the translation results from the translation means 6, and the process is repeated until the shaping of the output document 8 is completed (step 505).

When the input document 3 is all processed and the shaping of the output document 8 is completed, the output document generating means 7 displays the output document 8 via the input / output means 2 (step 50).
6) The machine translation system ends the operation (step 507).

Next, the operation of the tag identifying means 4 will be described with reference to the drawings. Here, FIG. 6 is a flowchart showing the operation of the tag identifying means 4.

The tag identifying means 4 first obtains the input document 3 from the network 1 via the input / output means 2 (step 501).

Next, in order to extract the expression related to the tag information of the obtained input document 3, 1 of the translation target tag information 200 stored by the translation target tag information storage means 5 is extracted.
The individual tag patterns are obtained (step 601). Next, the input document 3 is searched for a character string that matches the obtained tag pattern (step 602). If a character string matching the tag pattern is present in the input document 3 (step 603), the matched character string is transferred to the translation means 6 as a translation target (step 604). Step 603
If the matching character string does not exist in the input document 3, the operation of step 604 is omitted.

Thereafter, the translation target tag information storage means 5
It is checked whether or not the input document 3 has been searched for in all the tag patterns of the translation target tag information 200 stored by, and if there is a tag pattern that has not been searched yet (step 605), the above-mentioned is performed. Step 601-
The processing of 604 is repeated. On the other hand, if the search by all the tag patterns is completed (step 605), the tag identifying means 4 ends the operation (step 606).

Next, the operation of the translation means 6 will be described with reference to the drawings. Here, FIG. 7 is a flowchart showing the operation of the translation means 6. It should be noted that the translation means 6 can naturally translate a tagged document, and, for example, the one described in the following document can be applied.

Reference “Naoto Ishikawa, Masayuki Hiyama”, “English-Japanese Machine Translation Support System for Tagged Documents”, CALS Japan '9
4, S2-1 ”First, in the translating means 6, the tag identifying means 4 performs step 604.
The translation target transferred in (1) is obtained (step 701), and the translation corresponding to the tag information is performed (step 702). The translation corresponding to the tag information is realized by hiding the tag portion, translating a portion other than the tag portion by an ordinary machine translation method, and restoring the concealed tag portion as a result. Finally, the translation unit 6 transfers the translation result including the tag to the output document generation unit 7 (step 703) and ends the operation (step 704).

Next, the operation of the output document generating means 7 will be described with reference to the drawings. Here, FIG. 8 is a flowchart showing the operation of the output document generation means 7.

First, the output document generation means 7 is the translation means 6
Receives the translation result transferred in step 703 (step 801). Next, the received translation result is stored in the output document 8 with reference to the format of the input document 3 (step 802). Step 801 and step 8 above
The operation of 02 is repeated until the tag identifying means 4 processes all of the input document 3, the output document generating means 7 receives all the translation results from the translating means 6, and the shaping of the output document 8 is completed (step 505). ).

When all the input documents 3 are processed and the shaping of the output document 8 is completed (step 505), the output document generation means 7 displays the output document 8 via the input / output means 2 (step 506). , End the operation (step 80)
3).

In the following, the tagged document 30 shown in FIG.
1 as an example of the input document 3 and the translation target tag information storage unit 5 storing the translation target tag information 200 shown in FIG. 2, the operation of the machine translation system of the first embodiment will be described in detail. Explained.

First, the tag identifying means 4 obtains the input document 301 (step 501). Next, the tag identifying means 4 obtains the first tag pattern 201 of the translation target tag information 200 stored in the translation target tag information storage means 5 (step 6).
01). The obtained tag pattern is "<I>. + </ I
> ”, A character string matching this is input document 30
Search from 1 (step 602). As a result, "<I
> Parsing </ I>, “<I> transfer
"ring </ I>", "<I>generation"
</ I> ”and“ <I> transfer method ”
Since "d </ I>" is extracted as a character string that matches the tag pattern 201 (step 603), the tag identification means 4 transfers these character strings to the translation means 6 as a translation target (step 604).

The translation means 6 obtains these four translation objects (step 701), translates them, and then respectively translates them into "<I> analysis </ I>", "<I> transformation </ I>",
“<I> Generation </ I>”, “<I> Transfer method <
/ I> ”is obtained (step 702). The translation means 6 outputs these four translation results to the output document generation means 7
(Step 703).

The output document generation means 7 stores the four translation results obtained from the translation means 6 in a predetermined position (see FIG. 9) of the output document 900 with reference to the format of the input document 301 (step 802). .

After that, the tag identification means 4 obtains the second tag pattern 202 (step 601) because the tag pattern next to the translation target tag information 200 remains (step 605). The obtained tag pattern is "<H
[0-9]>. + </ H [0-9]> ”, a character string matching this is retrieved from the input document 301 (step 602). As a result, “<H1> Recent
Trend of Research And Dev
"Epment </ H1>" and "<H2> The
Latest Technological Tren
Since "d </ H2>" is extracted as a character string that matches the tag pattern 202 (step 603), the tag identification means 4 transfers these character strings to the translation means 6 as a translation target (step 604).

The translation means 6 obtains these two objects to be translated (step 701), translates them and translates them into "<H1> Recent research and development trends </ H1>" and "<H2
A translation result "> latest technological trend </ H2>" is obtained (step 702). The translation means 6 transfers these two translation results to the output document generation means 7 (step 703).

The output document generation means 7 stores the four translation results obtained from the translation means 6 in a predetermined position (see FIG. 9) of the output document 900 with reference to the format of the input document 301 (step 802). .

.., 209 of the input document 301 are sequentially translated, stored in the output document 900, and finally displayed in the output document display screen 1000 shown in FIG. Is displayed via the input / output unit 2 (step 506).

(A-3) Effects of First Embodiment As described above, according to the machine translation system of the first embodiment, a translation target tag that specifies a portion of the input document 3 that can be a translation target. Since the information is described and stored, and only the part of the input document 3 corresponding to this translation target tag information is translated, the output document 8 which is the result of translating the input document 3 is obtained. You can

In this way, by describing the information defining the figure or table in the translation target tag information, the translation result can include non-language information such as the figure or table, and the translation target tag information can be included. By describing the information designating the document of the link destination, the document information of the link destination can be included in the translation result. That is, it is possible to obtain the translation result summarized into the minimum necessary information without impairing the intention of the author of the input document 3.

Now, the input document 301 shown in FIG. 3 and FIG.
Comparing the output document 900 shown in FIG.
Since the part limited to only the expressions related to the tag information of is extracted and translated, it is possible that the translation summarized into the minimum necessary information is performed without impairing the intention of the author of the input document 301. I understand. This is further apparent by referring to the display screen 1000 of the output document 900 shown in FIG.

Note that the display area 1001 of FIG. 10 shows the “<TITLE> Recent research and development trend </ T> of the output document 900.
This corresponds to the part of "ITLE>". Display area 10
02 is "<BODY>" and "</ B
It is a display area corresponding to the description of the part sandwiched between ODY> ”. Note that FIG. 302 in FIG. 10 is the same diagram (image) information as that of FIG.
MG SRC = “TRANSFER.gif”> ”is displayed. Further, in FIG. 10, an output document 900 is displayed as a display corresponding to the link destination document 303 in FIG.
“<A HREF =” ambiguity. htm
The underlined "ambiguity problem" part corresponding to "l"> ambiguity problem </A>"is also displayed.

(B) Second Embodiment Next, a second embodiment in which the document processing system according to the present invention is applied to a machine translation system for tagged documents will be described in detail with reference to the drawings.

As with the first embodiment, the machine translation system of the second embodiment is adapted to obtain a translation result summarizing a tagged document. In addition to this,
The translations of the words existing in the figures and tables included in the abstract are also included in the translation result.

(B-1) Configuration of the Second Embodiment FIG. 11 is a block diagram of the machine translation system of the second embodiment, which is the same as FIG. 1 according to the first embodiment described above. Corresponding parts are designated by the same reference numerals.

As is clear from comparison with FIG. 11 and FIG. 1, the machine translation system of the second embodiment has the link object acquisition means 9 and the code in addition to the configuration of the machine translation system of the first embodiment. It is provided with a conversion means 10.

The link object acquisition means 9 uses the image tag extracted by the tag identification means 4 to input / output means 2
The entity referred to by the image tag (hereinafter referred to as a link object) is acquired from the network 1 via the. The encoding means 10 recognizes character information from the image of the link object obtained by the link object acquisition means 9. As the encoding means 10, it is possible to apply an existing character recognition device for recognizing a character contained in the dot pattern image information.

The tag identifying means 4 of the second embodiment
Not only identifies the tag information included in the input document 2 and extracts the translation target expression including the tag information, but also the input document 2
Tag information (hereinafter referred to as an image tag) for referring to non-language information (figure or image information) included in is also extracted.

In the machine translation system of the second embodiment, by adding the link object acquisition means 9 and the encoding means 10 described above to the system configuration of the first embodiment, the non-language information of the input document 3 ( (Figures and image information) can be converted into language information and translated.

The input document 301 (3) shown in FIG.
FIG. 16 shows an output document 8 (1600) processed by the machine translation system of the second embodiment, and FIG. 17 shows a display screen of the output document 8 (1600). By comparing these FIGS. 16 and 17 with FIGS. 8 and 9 described above, according to the machine translation system of the second embodiment, the input document 3 of the input document 3 which could not be executed in the first embodiment is displayed. It can be seen that non-linguistic information (figure and image information) can be converted into linguistic information and translated.

(B-2) Operation of the Second Embodiment Next, the operation of the machine translation system of the second embodiment for translating a tagged document as described above will be described.

FIG. 12 is a flow chart showing the operation of the entire machine translation system of the second embodiment.

Also in the machine translation system of the second embodiment, as described above, it is considered that some part of the tagged document to which the tag information is added has some intention of the author of the document, and only that part is considered. By performing the summary only in the above, the minimum necessary information of the input document 3 is extracted and translated. Therefore, the tag information included in the input document 3 is focused and the tag information is added. The expression is used as a translation target and is translated to generate and display the output document 8. In addition, it recognizes linguistic information from the non-linguistic information (figure or image information) of the input document 3 and translates it, thereby more accurately transmitting the author's intention.

First, the tag identifying means 4 inputs the input document 3 with a tag from the network 1 via the input / output means 2 (step 501). Since the input document 3 contains the tag information, the tag identifying means 4 obtains the translation target tag information 200 from the translation target tag information storage means 5 and identifies the translation target part of the input document 3 (step 120
1).

If the identified tag is an image tag that refers to non-language information (table or image information) (step 1202), the link object acquisition means 9
Acquires the link object corresponding to the link tag from the network 1 via the input / output unit 2 (step 1203). Thereafter, the encoding means 10 recognizes the language information, that is, the character string, from the link object (step 1204). The tag identification means 4 obtains the character string recognized by the encoding means 10 via the link object acquisition means 9, and passes the character string of the link tag and the recognized character string to the translation means 6 as a translation target (step 120).
5).

On the other hand, in the judgment at step 1202 described above,
If the tag information identified by the tag identifying means 4 is not an image tag, the tag identifying means 4 passes it to the translating means 6 (step 1206).

The translation means 6 translates the translation object obtained from the tag identification means 4 and transfers the result to the output document generation means 7 (step 503). The output document generation means 7 shapes the translation result obtained from the translation means 6 while referring to the input document 3 and stores it in the output document 8 (step 504).

In the operations of the above steps 501 to 504, the tag identifying means 4 processes all of the input document 3,
The output document generation means 7 receives all the translation results from the translation means 6 and repeats until the shaping of the output document 8 is completed (step 505).

When all the input documents 3 are processed and the shaping of the output document 8 is completed, the output document generating means 7 displays the output document 8 via the input / output means 2 (step 50).
6), the machine translation system of the second embodiment ends the operation (step 1207).

Next, the operation of the tag identifying means 4 will be described with reference to the drawings. Here, FIG. 13 shows the tag identifying means 4
6 is a flowchart showing the operation of the first embodiment.

The tag identifying means 4 first obtains the input document 3 from the network 1 via the input / output means 2 (step 501).

Next, in order to extract the expression related to the tag information of the obtained input document 3, 1 of the translation target tag information 200 stored by the translation target tag information storage means 5 is extracted.
The individual tag patterns are obtained (step 601). Next, the input document 3 is searched for a character string that matches the obtained tag pattern (step 602). If a character string matching the tag pattern is present in the input document 3 (step 603), it is further checked whether the tag information included in the matching character string is an image tag (step 1).
202).

If the tag information is an image tag, the character string is transferred to the link object acquisition means 9 (step 1301). Next, the character string recognized through the operations of the link object acquisition means 9 and the encoding means 10 is received from the link object acquisition means 9 (step 1302). Then, the character string of the image tag and the received recognized character string are transferred to the translation means 6 as translation targets (step 1205).

On the other hand, if it is determined in step 1202 that the tag information identified by the tag identifying means 4 is not an image tag, the tag identifying means 4 passes it to the translating means 6 (step 1206).

If there is no matching character string in the input document 3 in the above step 603, step 1
The operations of 206 and steps 1301 to 1205 are omitted.

After that, the translation target tag information storage means 5
It is checked whether or not the input document 3 is searched for in all the tag patterns of the translation target tag information 200 stored by (step 605), and if there is a tag pattern that has not been searched yet, steps 601 to 601 are executed. Step 1
205, or steps 601-1206 are repeated.

When the search using all the tag patterns is completed (Yes in step 605), the tag identifying means 4 ends the operation (step 1303).

Next, the operation of the link object acquisition means 9 will be described with reference to the drawings. Here, FIG.
7 is a flowchart showing the operation of the link object acquisition means 9.

The link object acquisition means 9 first
The tag identifying means 4 receives the character string (link information) indicating the reference destination of the object transferred in step 1301 (step 1401). Next, the link object corresponding to the link information is obtained from the network 1 via the input / output unit 2 (step 1203). And
The obtained link object is transferred to the encoding means 10 (step 1402), and the recognition character string recognized by the encoding means 10 from the link object is obtained from the encoding means 10 (step 1403). Finally, the link object acquisition means 9 returns the recognized character string to the tag identification means 4 (step 1404) and ends the operation (step 1405).

Here, the tag identification means 4 and the link object acquisition means 9 have steps 1301 and 14 respectively.
01, and step 1404 and step 1302 operate in synchronization with each other.

Next, the operation of the encoding means 10 will be described with reference to the drawings. Here, FIG. 15 shows the encoding means 10.
6 is a flowchart showing the operation of the first embodiment.

In the encoding means 10, first, the link object acquisition means 9 receives the link object transferred in step 1402 (step 1501). next,
Recognize the character string in the link object (step 1
204). The recognition of the character string is realized by any existing character recognition method. Finally, the encoding means 10
Returns the recognized character string to the link object acquisition means 9 (step 1502) and ends the operation (step 1503).

The link object acquisition means 9 and the encoding means 10 operate in synchronization with each other at step 1402 and step 1501, and at step 1502 and step 1403.

The translation means 6 in the second embodiment
Operates in the same manner as the translation means 6 in the first embodiment according to the flowchart shown in FIG. The output document generation means 7 in the second embodiment also operates in the same manner as the output document generation means 7 in the first embodiment according to the flowchart shown in FIG.

In the following, the tagged document 30 shown in FIG.
1 as an example of the input document 3 and the translation target tag information storage unit 5 storing the translation target tag information 200 shown in FIG. 2, the operation of the machine translation system of the second embodiment will be described in detail. Explained.

First, the tag identifying means 4 obtains the input document 301 (step 501). Next, the tag identifying means 4 obtains the first tag pattern 201 of the translation target tag information 200 stored in the translation target tag information storage means 5 (step 6).
01). The obtained tag pattern is "<I>. + </ I
> ”, A character string matching this is input document 30
Search from 1 (step 602). As a result, "<I
> Parsing </ I>, “<I> transfer
"ring </ I>", "<I>generation"
</ I> ”and“ <I> transfer method ”
d </ I> ”is extracted as a character string that matches the tag pattern 201 (step 603), and since they are not image tags, the tag identifying means 4 translates these character strings as translation targets. Transfer to the means 6 (step 1206).

The translation means 6 obtains these four translation objects (step 701), translates them, and outputs "<
“I> analysis </ I>”, “<I> conversion </ I>”, “<I
> Generation </ I>, “<I> Transfer method </ I>
> ”Is obtained (step 702). The translation means 6 transfers these four translation results to the output document generation means 7 (step 703).

The output document generation means 7 stores the four translation results obtained from the translation means 6 in a predetermined position (see FIG. 16) of the output document 1600 while referring to the format of the input document 301 (step 802). .

Thereafter, since the tag pattern next to the translation target tag information 200 is left in the tag identifying means 4 (step 605), the input document 30 is similarly processed.
, 207 of 1 are sequentially translated and stored in the output document 1600.

Next, the tag identifying means 4 determines the eighth of the translation target tag information 200 stored in the translation target tag information storage means 5.
The tag pattern 208 is obtained (step 601). As a result, "<center><IMG SRC =" TRA
NSFER. gif "></center>" is extracted as a character string that matches the tag pattern 208 (step 603). Here, since the tag included in the character string is an image tag (<IMG ...>) (step 1202), the character string is transferred to the link object acquisition means 9 (step 1301).

The link object acquisition means 9 displays "<c
enter><IMG SRC = “TRANSFER.
A link object 302 conforming to gif "></center>" is obtained from the network 1 via the input / output unit 2 (step 1203), and further, the encoding unit 1
0 is the character string “Orig from the link object 302.
"internal Sentence", "Parsing",
..., "Transfer Method" is obtained (step 1204). The tag identifying means 4 recognizes the six recognized character strings thus obtained and the character string “<ce which includes the link tag.
inter><IMG SRC = “TRANSFER.g
If "></center>" is transferred to the translation means 6 (step 1205).

At this time, a tag (in HTML, <PRE in HTML is used to instruct the translation means 6 to perform non-translation on the recognition character string.
>. . . </ PRE>) may be added to make the relationship between the recognized character string and the corresponding image clearer. As a result, “<center><IMG SR
C = “TRANSFER.gif”></ center
> ”,“ <PRE> [Original Senten
ce: </ PRE> Original Senten
ce <PRE>] </ PRE> ”,“ <PRE> [Pa
rsing: </ PRE> Parsing <PRE
>] </ PRE> ”, ...,“ <PRE> [Transf
er Method: </ PRE> Transfer
Method <PRE>] </ PRE> ”is passed to the translation means 6.

The translating means 6 obtains these seven translation objects (step 701), translates them, and outputs "<center><IMG SRC =" TRANSF.
ER. gif ”></center>”, “[Orig
internal Sentence: original text, ”“ [Pars
ing: Analysis] ”, ...,“ [Transfer Met
"hod: transfer method]" is obtained (step 702). The translation means 6 transfers these seven translation results to the output document generation means 7 (step 703).

The output document generation means 7 refers to the seven translation results obtained from the translation means 6 with reference to the format of the input document 301, and outputs the output document 1600 at a predetermined position (reference numeral 16 in FIG. 16).
No. 01) (step 802).

Similarly, the character strings conforming to the tag pattern 209 of the input document 301 are sequentially translated and stored in the output document 1600. Finally, the output document display screen 1700 shown in FIG. Is displayed via (step 506).

(B-3) Effect of Second Embodiment The machine translation system of the second embodiment also has the technical idea of the first embodiment as it is.
It is possible to achieve the effect that the embodiment described above has.

In addition to this, according to the second embodiment,
It is possible to obtain the output document 8 in which the translated content of the character string information of the image portion (see reference numeral 1601 in FIG. 16) is added to the result of the translation of the input document 3 in a summarized manner, and the intention of the author is further clarified. Can be

Comparing the input document 301 shown in FIG. 3 with the output document 1600 shown in FIG. 16, even for the image portion 302 in which the character string is displayed in the original language (English),
Since the character string part is translated and output in Japanese,
It can be seen that the author's intention is clearer. Also,
This is further apparent by referring to the display screen 1700 of the output document 1600 shown in FIG.

The display area 302 shown in FIG. 17 is the same as that shown in FIG.
Figure (image) information similar to that of
A display area 1601 in which a character string portion of this display area 302 (FIG.) Is translated is inserted.

(C) Third Embodiment Next, a third embodiment in which the document processing system according to the present invention is applied to a machine translation system for tagged documents will be described in detail with reference to the drawings.

The machine translation system of the third embodiment has the same function as that of the second embodiment, and in addition to the function of the second embodiment, when displaying the translation result of the input document, the link destination document specified in the input document is displayed. It also has a function of displaying the translation result together.

(C-1) Configuration of Third Embodiment FIG. 18 is a block diagram of a machine translation system of the third embodiment, and FIG. 11 according to the second embodiment described above.
The same and corresponding parts as those shown in FIG.

As is clear from comparison with FIG. 18 and FIG. 11, the machine translation system of the third embodiment has an object storage means 11 in addition to the configuration of the machine translation system of the second embodiment. There is.

The object storage means 11 stores the link object obtained by the link object acquisition means 9 in order to make it a new input document.

The tag identifying means 4 of the third embodiment is used.
Identifies not only the tag included in the input document 2 and the translation target expression including the tag and the image tag information included in the input document 2, but also tag information indicating a link to another document (hereinafter referred to as a link tag). Also called).

The machine translation system according to the third embodiment has the same structure as that of the second embodiment except that the object storage means 11 is used.
Is added, the document referred to by the input document 3 (link destination document) can be translated as a new input document.

Here, the input document 301 shown in FIG.
An output document 8 (2200) obtained by processing (3) in the machine translation system of the third embodiment is shown in FIG. 22, and a display screen of the output document 8 (2200) is shown in FIG. By comparing these FIG. 22 and FIG. 23 with the above-mentioned FIG. 16 and FIG. 17, the machine translation system of this third embodiment has a link relationship that could not be executed in the second embodiment. It can be seen that the translation results of multiple documents can be displayed simultaneously.

(C-2) Operation of the Third Embodiment Next, the operation of the machine translation system of the third embodiment for translating a tagged document as described above will be described.

FIG. 19 is a flow chart showing the operation of the entire machine translation system of the third embodiment.

The machine translation system of the third embodiment not only operates so as to realize the functions of the machine translation system of the second embodiment, but also refers to the link tag of the input document 3. A linked document is acquired and translated as a new input document, and the summary translation result of the link destination document in which only the link information is shown in the second embodiment is also the input source document of the link source. It operates to add and display the translation result.

First, the tag identifying means 4 inputs the input document 3 with a tag from the network 1 via the input / output means 2 (step 501). Since the input document 3 includes the tag information, the tag identifying unit 4 obtains the translation target tag information 200 from the translation target tag information storage unit 5 and identifies the translation target of the input document 3 (step 1201). .

If the identified tag is an image tag that refers to non-language information (figure or image information) (step 1202), the link object acquisition means 9
Acquires the link object corresponding to the link tag from the network 1 via the input / output unit 2 (step 1203). Thereafter, the encoding means 10 recognizes the language information, that is, the character string, from the link object (step 1204). The tag identification means 4 obtains the character string recognized by the encoding means 10 via the link object acquisition means 9, and passes the character string of the link tag and the recognized character string to the translation means 6 as a translation target (step 120).
5).

On the other hand, if a negative result is obtained in the above step 1202, it is judged whether or not it is a link tag (step 1901). If the tag information identified by the tag identifying means 4 is a link tag (step 1901),
The link object acquisition unit 9 obtains the link object corresponding to the link tag from the network 1 via the input / output unit 2 (step 1902). After that, the link object acquisition means 9 transfers the acquired link object to the object storage means 11, and the object storage means 11 passes the stored object to the tag identification means 4 as a new input document (step 1).
903). The tag identifying means 4 is the object storing means 1
When the transfer of the link object from No. 1 is received, the operation up to that point is suspended, and the operation of using the transferred object as a new input document is started (step 1904).
When the tag identifying means 4 finishes the operation for the new input document, it returns to the suspended state and continues the operation (proceeds to step 503).

On the other hand, if the tag information identified by the tag identifying means 4 is not a link tag in the determination in step 1901, that is, if the tag identified by the tag identifying means 4 is neither an image tag nor a link tag. , The tag identification means 4 passes it to the translation means 6 (step 12).
06).

The translation means 6 translates the translation object obtained from the tag identification means 4 and passes the result to the output document generation means 7 (step 503). The output document generation means 7 shapes the translation result obtained from the translation means 6 while referring to the input document 3 and stores it in the output document 8 (step 504).

In the operations of the above steps 501 to 504, the tag identifying means 4 processes all of the input document 3,
The output document generation means 7 receives all the translation results from the translation means 6 and repeats until the shaping of the output document 8 is completed (step 505). When the input document 3 is all processed and the shaping of the output document 8 is completed, the output document generating means 7
Displays the output document 8 via the input / output unit 2 (step 506), and the machine translation system ends the operation (step 1905).

Next, the operation of the tag identifying means 4 will be described with reference to the drawings. Here, FIG. 20 is a flowchart showing the operation of the tag identifying means 4 of the third exemplary embodiment.

The tag identifying means 4 first determines whether or not an object is transferred from the object storage means 11 (step 2001). If there is an object transfer, the operation up to that point is suspended, a document is input from the object storage means 11, and the operation proceeds to step 601 in order to newly start the operation as a new input document (step 2002). It should be noted that when the tag identifying means 4 finishes the operation for the new input document, it returns to the suspended state and continues the operation.

While the object is not transferred from the object storage means 11 (step 2001), the tag identification means 4 first obtains the input document 3 from the network 1 via the input / output means 2 (step 501). ).

Next, in order to extract the expression relating to the tag information of the obtained input document 3 (which may be a document from the object storage means 11), it is stored by the translation target tag information storage means 5. Translation target tag information 200
(Step 601). Next, the input document 3 is searched for a character string that matches the obtained tag pattern (step 602), and it is determined whether or not a matching character string exists (step 603).

If a character string matching the tag pattern is present in the input document 3 (Yes in step 603), whether the tag information included in the matching character string is an image tag or a link tag. It is inspected (step 2003), and if the tag information is an image tag or a link tag, the character string is transferred to the link object acquisition means 9 (step 1301).

Next, the character string recognized through the operations of the link object acquisition means 9 and the encoding means 10 is received from the link object acquisition means 9 (step 130).
2). Then, the character string of the image tag and the received recognized character string are transferred to the translation means 6 as translation targets (step 1205). When the tag information is an image tag, these steps 1302 and 1205 function effectively, and when the tag information is a link tag, the operation of the link object acquisition means 9 causes the above step 2 to be performed.
002 is executed, and these steps 1302 and 1205 do not work.

On the other hand, if the tag identified by the tag identifying means 4 is neither an image tag nor a link tag in the judgment of step 2003, the tag identifying means 4 passes it to the translating means 6 (step 1206).

If no matching character string exists in the input document 3 in step 603, the operations of step 1206 and steps 1301 to 1205 are omitted.

After that, the translation target tag information storage means 5
It is checked whether or not the input document 3 is searched for in all the tag patterns of the translation target tag information 200 stored by (step 605), and if there is a tag pattern that has not been searched yet, steps 601 to 601 are executed. Step 1
205, or steps 601-1206 are repeated.

When the search using all the tag patterns is completed, the tag identifying means 4 ends the operation (step 1303).

Next, the operation of the link object acquisition means 9 of the third embodiment will be described with reference to the drawings. Here, FIG. 21 is a flow chart showing the operation of the link object acquisition means 9 of the third exemplary embodiment.

The link object acquisition means 9 first
The tag identifying means 4 receives the character string (link information) indicating the reference destination of the object transferred in step 1301 (step 1401). An image tag or a link tag is identified (step 2101).

If the transferred link information is an image tag, the link object acquisition means 9
Acquires a link object corresponding to the link information from the network 1 via the input / output unit 2 (step 1203). In this case, image information is obtained as a link object. Then, the obtained link object is transferred to the encoding means 10 (step 140).
2) The recognition character string recognized by the encoding means 10 from the link object is obtained from the encoding means 10 (step 14).
03). Finally, the link object acquisition means 9 returns the recognized character string to the tag identification means 4 (step 1
404), the operation ends (step 2103).

On the other hand, if the transferred link information is not an image tag but a link tag for referencing another document (step 2101), the link object acquisition means 9 causes the network 1 to operate via the input / output means 2. A link object corresponding to the link information is obtained from (step 1203). In this case, the referenced document with a tag is obtained as the link object. Next, the link object acquisition means 9 transfers the obtained link object to the object storage means 11 (step 210).
2) The operation is ended (step 2103).

Note that the tag identifying means 4 and the link object obtaining means 9 perform steps 1301 and 140.
1, and the steps 1404 and 1302 operate in synchronization with each other. Further, the link object acquisition means 9 and the object storage means 11 perform step 210.
2, the object storing means 11 and the tag identifying means 4 and the step 20 at the time when the object is stored.
02 works synchronously.

The translation means 6 in this third embodiment
Operates in the same manner as the translation means 6 in the first embodiment according to the flowchart shown in FIG. The output document generation means 7 in the third embodiment also operates in the same manner as the output document generation means 7 in the first embodiment according to the flowchart shown in FIG. Furthermore, the encoding means 10 in the third embodiment follows the flowchart shown in FIG. 15 and the encoding means 10 in the second embodiment.
Works the same as.

In the following, the tagged document 30 shown in FIG.
1 as an example of the input document 3 and the translation target tag information storage means 5 stores the translation target tag information 200 shown in FIG. 2, the operation of the machine translation system of the third embodiment will be described in detail. Explained.

First, the tag identifying means 4 obtains the input document 301 (step 501). Next, the tag identifying means 4 obtains the first tag pattern 201 of the translation target tag information 200 stored in the translation target tag information storage means 5 (step 6).
01). The obtained tag pattern is "<I>. + </ I
> ”, A character string matching this is input document 30
Search from 1 (step 602). As a result, "<I
> Parsing </ I>, “<I> transfer
"ring </ I>", "<I>generation"
</ I> ”and“ <I> transfer method ”
"d </ I>" is extracted as a character string that matches the tag pattern 201 (step 603).
Since there is no image tag or link tag, the tag identifying means 4 transfers these character strings to the translation means 6 as translation targets (step 1206).

The translation means 6 obtains these four translation objects (step 701), translates them, and then respectively translates them into "<I> analysis </ I>", "<I> transformation </ I>",
“<I> Generation </ I>”, “<I> Transfer method <
/ I> ”is obtained (step 702). The translation means 6 outputs these four translation results to the output document generation means 7
(Step 703).

The output document generation means 7 stores the four translation results obtained from the translation means 6 in a predetermined position (see FIG. 22) of the output document 2200 while referring to the format of the input document 301 (step 802). . After that, in the tag identifying means 4, since the tag pattern next to the translation target tag information 200 remains (Yes in step 605), the characters matching the tag patterns 202 and 203 of the input document 301 are similarly processed. Output document 22
00 is stored.

Next, the tag identifying means 4 determines the fourth of the translation target tag information 200 stored in the translation target tag information storage means 5.
The tag pattern 204 is obtained (step 601). As a result, "<A HREF =" ambiguity. ht
ml "> ambiguityprogram </A>"
Is extracted as a character string that matches the tag pattern 204 (step 603). Here, the tags included in the character string are link tags (<A HREF...></
A>) (step 2003), the character string “<A HREF =” ambiguity. html ”
> Ambiguity problem </a> ”
It is transferred to the link object acquisition means 9 (step 1301).

The link object acquisition means 9 indicates "<A
HREF = "ambiguity.html"> am
The link object 303 that conforms to the "biquity problem </A>" is obtained from the network 1 via the input / output unit 2 (step 1902), and the obtained link object 303 is transferred to the object storage unit 11 (step 2102).

After that, the object storing means 11
Transfers the object to the tag identifying means 4 (step 2001), the tag identifying means 4 suspends the operation up to that point (that is, operates in the state in which the tag patterns 204 are inspected), and the object storing means 11 stores the object. The object 303 is input, and the process proceeds to step 601 in order to start a new operation using it as a new input document (step 2002).

Hereinafter, the tag identification means 4 is the first of the translation target tag information 200 stored in the translation target tag information storage means 5.
From the tag pattern 201 of No. 1 to the ninth tag pattern 209, the translating means 6 obtains the translation object, and the translating means 6 translates the expression related to the tag of the new input document 303 (step 503), and further outputs the output document. The generation unit 7 stores the translation result at a predetermined position (see reference numeral 2201 in FIG. 22) of the output document 2200 (step 504).

At this point, since the operation of the tag identifying means 4 for the object 303 is completed, the tag identifying means 4
Restarts the operation from the state in which the tag patterns 204 that have been suspended above are inspected.

In the same manner, character strings conforming to the tag patterns 205 to 207 of the input document 301 are sequentially translated and stored in the output document 2200. Next, the tag identifying means 4 and the translation target tag information storing means 5 Obtains the eighth tag pattern 208 of the translation target tag information 200 stored by (step 601). As a result, "<center><IM
G SRC = "TRANSFER.gif"></ ce
"nter>" is extracted as a character string that matches the tag pattern 208 (step 603). Here, the tag included in the character string is an image tag (<IMG ...>).
(Step 2003), the character string “<ce
inter><IMG SRC = “TRANSFER.g
if "></center>" is transferred to the link object acquisition means 9 (step 1301).

The link object acquisition means 9 displays "<c
enter><IMG SRC = “TRANSFER.
A link object 302 conforming to gif ">></center>" is obtained from the network 1 via the input / output unit 2 (step 1203). Specific operations performed by the tag identification means 4, the translation means 6, the output document generation means 7, the link object acquisition means 9, and the encoding means 10 for the obtained link object (FIG.) Are the same as those in the second embodiment. Yes, the translation result obtained from the translation unit 6 by the output document generation unit 7 refers to the format of the input document 301 at a predetermined position of the output document 2200 (see FIG. 22).
No. 1601).

Similarly, the character strings matching the tag pattern 209 of the input document 301 are sequentially translated, stored in the output document 2200, and finally the output document display screen 2300 shown in FIG. Is displayed via (step 506).

(C-3) Effects of the Third Embodiment The machine translation system of the third embodiment also has the technical idea of the second embodiment as it is.
It is possible to achieve the effect that the embodiment described above has.

In addition to this, according to the third embodiment,
A part limited to only the expressions related to the tag information of the link destination document 303 referred to by the link tag from the input document 301 is extracted and translated. Conventionally, unless the user actively follows the link tag. It becomes possible to obtain information that cannot be acquired, and the intention of the author of the input document 301 becomes clearer. This is clear by referring to the display screen 2300 of the output document 2200 shown in FIG.

Note that, in FIG. 23, "<A>" in the link tag information 2200 is displayed as a display corresponding to the document 303 in FIG.
The underlined "ambiguity problem" portion 2302 corresponding to "HREF =" ambiguity.html "> ambiguity problem </A>" and the translated content 2201 of the linked document 303 are also included. It is displayed.

(D) Fourth Embodiment Next, a fourth embodiment in which the document processing system according to the present invention is applied to a machine translation system for tagged documents will be described in detail with reference to the drawings.

The machine translation system of the fourth embodiment has a function of converting the character style in the translation result of the input document, in addition to the same function as that of the third embodiment. Here, the character style refers to a modification such as underline, bold, italic, or emphasized with respect to a character that forms a sentence.

In a tagged document, the fonts constituting the sentence are often modified by underlining, bold letters, italics, emphasis, etc., and the result of summarizing and translating such a document also has character style information. . However, the character style for the font is conscious of display and printing of unsummarized documents, and may not be suitable for the summarized translation result. Therefore, in the fourth embodiment, a character style conversion function is provided.

(D-1) Configuration of the Fourth Embodiment The machine translation system of the fourth embodiment has the configuration shown in FIG. 24 between the output document generating means 7 and the output document (buffer) 8 in the third embodiment. The character style conversion means 12 having the detailed configuration shown in FIG.

In FIG. 24, the character style conversion means 1
2 is a character style processing control unit 20, a character style registration / edit table 21, and a character style conversion reference table 2
2. The character style conversion collation table 23 and the character style conversion determination processing unit 24 are included.

The character style conversion judgment processing section 24 inputs the summary translation result, and uses the character style conversion collation table 23.
Using the data extracted from, determine the specified contents of the character style change registered by the user, and if the corresponding character style exists in the input translation result,
The character style specified by the user is changed, and the changed summary translation result is used as the output document 8.

The character style processing control unit 20 is activated by the user via the input / output means 2 and registers the contents of the character style change / designation in the translation result by the user using the character style registration / edit table 21. It controls the processing for editing and editing, the processing for referring to the character style conversion reference table 22 using the registered change designation content, and setting the data in the character style conversion collation table 23.

The character style registration / edit table 21 is
This is a table used by the user to register and edit the content for changing and designating the character style in the translation result. When registering or editing in this table, the user can register with a normal character style name without being aware of the special form handled in the tagged document. For example, in the document, bold (bold typeface) is represented by the tag information character strings "<B>" and "</ B>", but at the time of registration / editing, the user identifies the type name of the character style as "bold". It is designed so that you can enter it.

The character style conversion reference table 22 converts the character style change specification contents in the translation result registered / edited by the user in the character style registration / edit table 21 into corresponding tag information (character string). It stores information for doing so. The content stored in the character style conversion reference table 22 is the character style processing control unit 2
0 is referred to when the user sets data according to the stored contents of the character style registration / edit table 21 in the character style conversion / comparison table 23.

The character style conversion and collation table 23 is set with tag information which is treated as a special form in a tagged document, which corresponds to the content of change specification of the character style registered / edited by the user. The set data is used by the character style conversion determination processing unit 24.

FIG. 25 is an explanatory diagram showing a configuration example of the character style conversion reference table 22 described above.

The character style conversion reference table 22 is composed of a character style reference heading item 22A and a character style conversion reference heading item 22B.

The character style reference heading item 22A includes
The name of a normal character style (for example, "bold",
"Italic", "underline", etc. are stored, and the character style conversion reference heading item 22B is a tag used in a special form in the tagged document corresponding to the data of the character style reference heading item 22A. Information headings (for example, “<B>”, “</ B>”, etc.) are stored.

In the example of FIG. 25, "default character", "bold", "italic", "underline", "emphasis", and "strong emphasis" are stored as the data of the character style reference headline item 22A. As the data of the character style conversion reference heading item 22B corresponding to, there is no data in "default character", "<B>" and "</ B>" in "bold", and "<I in italic".
> ”And“ </ I> ”, and“ <U ”for“ underline ”
> ”And“ </ U> ”, and“ <EM> ”and“ <
/ EM> and “strong emphasis”, “<STRONG>” and “</ STRONG>” are stored. In addition, the "default character" has no designation of a character style (character style modification). Note that “<B>” is tag information indicating that the character immediately after it is bolded (bold), and “</ B>” is tag information indicating that the character immediately before it is bolded. , And other symbols (tag information).

FIG. 26 is an explanatory diagram showing the configuration of a character style registration / edit table 21 used by the user to register and edit the contents for changing and designating the character style in the translation result, and an example of the registration. is there.

The character style registration / edit table 21 is
It is composed of a character style heading item 21A and a character style conversion heading item 21B. The character style heading item 21A stores the name of the character style to be changed in the translation result input by the user, and the character style conversion heading item 21B stores the translation registered in the character style heading item 21A input by the user. The name of the changed character style corresponding to the name of the character style to be changed in the result is stored.

In the example of FIG. 26, “italic” and “bold” are registered as the character style heading data, and “bold” and “emphasis” are registered as the character style conversion heading data corresponding to each of them. There is.

FIG. 27 is an explanatory diagram showing an example of the structure and stored contents of the character style conversion collation table 23. Note that, in FIG. 27, the content stored in the character style conversion reference table 22 is the content shown in FIG. 2, and the content stored in the character style registration / edit table 21 is the content shown in FIG. 26. Shows the stored contents of.

In the character style conversion collation table 23,
When the work of changing and designating the character style of the user to the character style registration edit table 21 shown in FIG. 26 is completed, the character style processing control unit 20 refers to the character style conversion reference table 22 shown in FIG. Data obtained by converting the character style change designation content corresponding to the translation result registered / edited in the character style registration / edit table 21 by the user into tag information handled as a special form in the corresponding tagged document is set. .

The character style conversion collation table 23 is composed of character style conversion collation heading items 23A and character style conversion collation heading items 23B.

In the character style collation heading item 23A,
The data of the character style heading item 21A of the character style registration / edit table 21 is converted into a symbol used in a special form in the tagged document by referring to the character style conversion reference table 22 and set. On the other hand, in the character style conversion collation heading item 23B, the data of the character style conversion heading item 21B of the character style registration / edit table 21 is stored in the character style conversion reference table 2
2 is set, and data converted into a symbol used in a special form in the tagged document is set.

In the example of FIG. 27, in the character style collation heading item 23A, for the character string of FIG. 25, the italic character which is the data of the character style heading item 21A of the character style registration / edit table 21 of FIG. The style conversion reference table 22 is referred to and "<I>" and "</ I
> ”Is set, and for“ bold ”,“ <B> ”
And "</ B>" are set.

Character style conversion collation heading item 2
3B includes a character style registration / edit table 2 shown in FIG.
For "bold" which is the data of the character style conversion heading item 21B of No. 1, the character style conversion reference table 22 of FIG. 25 is referred to, and "<B>" and "</ B
> ”Is set, and for“ emphasis ”,“ <EM> ”and“ </ EM> ”are set.

(D-4) Operation of the Fourth Embodiment Hereinafter, the operation of the character style processing control unit 20 and the character style conversion determination processing unit 24 for executing the characteristic processing of the fourth embodiment will be described in order. To do.

Here, FIG. 28 and FIG. 29 are processing flow charts of the character style processing control unit 20.

When the user activates the character style registration / editing process via the input / output means 2, the character style process control unit 20 starts a series of processes shown in the flowcharts of FIGS. 28 and 29.

First, the character style registration / edit table 2
The existing character style registration / edit table data is extracted from No. 1 and displayed so that the user can newly register or edit the character style data (step 2801).

After that, through interactive data exchange with the user, the user is allowed to perform the registration / editing work of the information for designating the character style change to the displayed character style registration / edit table data, and registering / editing the information. When the editing work is guided to completion and the completion command from the user is received, the control is moved to the next step 2804 (steps 2802 and 2803). As described above, when registering / editing in this table, the user can carry out the work with the name of a normal character style without being aware of the special symbol form handled in the tagged document. .

At the next step 2804, the character style registration / edit table data for which the registration / edit work has been completed is stored in the character style registration / edit table 21.

Thereafter, the stored data of the character style registration / edit table 21 is collated with the data of the character style reference heading item 22A of the character style conversion reference table 22 (step 2805), and the collation result is confirmed.
It is determined whether the data of the character style registration / edit table 21 corresponds to the data of the character style reference headline item 22A (step 2806).

As a result of the judgment, if there is even one data in the character style registration / edit table 21 for which the corresponding data does not exist (if there is no data of the character style reference headline item 22A matching the data), it is regarded as the target. Data in the character style registration / edit table 21 is deleted (step 2807), control is transferred to step 2801, and the data that does not exist in the character style reference heading item 22A becomes an empty table item. The character style registration / edit table data is extracted and displayed, the user is prompted to register the correct data, the process proceeds to steps 2802 to 2806, and the same determination is performed, and the character style registration / edit table 2 is displayed.
This processing loop is repeated until all the data of 1 match.

If all the data (style names) in the character style registration / edit table 21 correspond to the data of the character style reference headline item 22A, step 2
Character style registration / edit table 21 saved in 804
One piece of unextracted data is extracted from this data (step 2808), and it is determined whether or not there is unextracted data (step 2809).

The unextracted data extraction operation of step 2808 is performed once each time the processing loop of steps 2808 to 2812 is repeated. The extraction order of the unextracted data extracted at each time is, for example, It is as follows.

First, the character style registration / edit table 2
The data of the character style conversion heading item 21B is extracted first from the data of the character style heading item 21A of 1, and then the control is transferred to this step 2808, the data of the character style conversion heading item 21B corresponding to the data of the extracted character style heading item 21A is extracted. , And then the data next to the character style heading item 21A.

If unextracted data cannot be extracted from the character style registration / edit table 21 (if all data has been extracted and processed and there is no unextracted data),
The character style process control unit 20 ends the series of processes.

On the other hand, when unextracted data can be extracted from the character style registration / edit table 21, the extracted data (character style name) and the character style reference heading item 22A of the character style conversion reference table 22 are extracted.
Is compared with the data (character style name) (step 2810). Then, the data (tag) of the character style conversion reference headline item 22B corresponding to the data (character style name) of the matched character style reference headline item 22A is extracted, and the target data (character style name) is extracted (tag information). ) (Step 281)
1).

Then, the replaced target data (tag information) is set in the character style conversion collation heading item 23A or the character style conversion collation heading item 23B in the character style conversion collation table 23 (step 2812).
Returning to step 2808 described above.

Here, the character style conversion collation table 2
The order of setting in 3 corresponds to the order of extracting data from the character style registration / edit table 21, and is as follows, for example.

First, the character style collation heading item 23A
, The data is set to the character style conversion collation heading item 23B at the next step 1012, and the data is set to the character style collation heading item 23A at the next step 1012. And 23A are alternately switched.

As a result, the replacement data of the data extracted from the character style heading item 21A of the character style registration / edit table 21 is set in the character style matching heading item 23A of the character style conversion / matching table 23, and the character style registration / editing is performed. The replacement data of the data extracted from the character style conversion heading item 21B in the table 21 is set in the character style conversion heading item 23B corresponding to the character style conversion heading item 23A in the character style conversion heading table 23.

Steps 2808 to 28 described above
The processing loop consisting of 12 is repeated by the number of data existing in the character style registration / edit table 21, and if there is no unextracted data extracted from the character style registration / edit table 21, the character style processing control unit 20 A series of processing ends.

Next, the operation of the character style conversion determination processing section 24 that appropriately uses the stored contents of the character style conversion collation table 23 set as described above will be described with reference to the drawings.

30 and 31 are processing flow charts of the character style conversion determination processing section 24. The processing of FIGS. 30 and 31 shows the processing for one sentence having the translated text data with the tag output from the output document generation means 7.

The character style conversion determination processing section 24 is, to put it broadly, for inputting the output document (translated data with tag) output from the output document generating means 7, and for character style conversion extracted from the character style conversion collation table 23. Change the registered character style specified by the user using the data of the specified data, and if the corresponding character style (tag information) exists in the translated data, change to the character style specified by the user. And the changed data (when the change is unnecessary, the data output from the output document generation means 7) is used as the output document 8.

The character style conversion determination processing section 24 is shown in FIG.
0 and the process shown in FIG. 31 are started, first, the translated text data with a tag output from the output document generation means 7 is read (step 3001).

Thereafter, the character style conversion collation table 2
One data (tag information related to the character style) is to be extracted from the character style collation heading item 23A of 3 (step 3002). Next, in this step 300
When the control is shifted to 2, the next data is to be extracted from the character style matching index entry 23A of the character style conversion matching table 23 according to a predetermined order.

When such an extracting operation is executed, it is determined whether or not the unextracted data has been extracted by this extracting operation. In other words, the character style matching heading item 23A of the character style conversion matching table 23 does not include the unextracted data. Is present (step 300).
3).

Here, if the unextracted data cannot be extracted even by the extraction operation (the character style collation heading item 23
If there is no unextracted data in A), the process proceeds to step 3009 described later.

On the other hand, when the data (tag information relating to the character style) can be extracted from the character style matching heading item 23A of the character style conversion and matching table 23, in the tagged translation data read from the output document generating means 7, The process proceeds to the processing routine of steps 3004 to 3008 for replacing the character string related to the extracted data (tag information related to the character style) this time with other data (tag information related to the other character style).

In this processing routine, first, character data is extracted from the head side of the tagged translation text data read from the output document generating means 7 (step 3004), and it is determined that the character data is not the last character data (end-of-sentence data). While checking (step 3005), the character data extracted from the translated text data and the data of the character style matching heading item 23A of the character style conversion matching table 23 extracted in step 3002 (tag information related to the character style) are checked. It is collated (step 3006), and it is judged whether or not they match and the execution symbol is not added (step 3007). If they do not match or if they match but the execution symbol is attached, the above-mentioned is executed. Returning to step 3004, the next character data is extracted from the tagged translation data.

Here, the execution symbol means the following step 3
This is a symbol added to clearly indicate that the processing has been performed on the character data that has been processed in the processing of 008. This makes it possible to eliminate competition in processing of the target data. That is, the character data replaced in step 3008 (for example, "<I>" to "<B
>)) Is a symbol given at the time of the first replacement in order to prevent the replacement (eg, “<B>” to “<EM>”) in the subsequent step 3008.

Steps 3004 to 30 described above
Step 300 is repeated by repeating the processing loop of 07.
When character data in the translated text data that matches the data (tag information related to the character style) with the character style matching heading item 23A of the character style conversion matching table 23 extracted in 2 and is not attached with the execution symbol is found ( Affirmative result in step 3007), the character style conversion collation heading item 23 of the character style conversion collation table 23 corresponding to the data of the matched character style conversion heading item 23A of the character style conversion collation table 23
B data (tag information related to character style) is extracted,
It replaces the matched character data portion of the translated text data, and at the same time, adds an execution symbol to clearly indicate that the processing has been executed (step 3008).

Even when this process is completed, there is a possibility that there is still matching character data at the end of the sentence, so the process returns to step 3004 and the next character data is extracted.

As described above, the processing loop of steps 3004 to 3007, or step 3004
By repeating the processing loop of step 3008, the replacement of all the character data from the beginning of the sentence to the end of the sentence that matches the data extracted from the character style matching heading item 23A of the character style conversion matching table 23 in step 3002 this time is completed. Sometimes, a positive result is obtained in step 3005, and the process returns to step 3002 to extract the next unextracted data from the character style conversion collation table 23.

Here, the character style conversion collation table 2
The above process is repeated for all the data stored in the character style collation heading item 23A of No. 3, and when the data cannot be extracted in step 3002, the execution symbol given in step 3009 is removed from the input data and output. Then (step 3009), a series of processing is ended.

Although not described in a concrete example, it is not suitable for the summarized translation result due to the operation of the character style conversion means 12 provided between the output document generation means 7 and the output document (buffer) 8. Convert the character style for the font to a character style suitable for the summarized translation result.

(D-3) Effects of the Fourth Embodiment The machine translation system of the fourth embodiment also provides the third embodiment.
Since it has the technical idea of the embodiment as it is, it is possible to obtain the effect that the third embodiment has.

In addition to this, according to the fourth embodiment,
Since the character style conversion means 12 is provided between the output document generation means 7 and the output document (buffer) 8, a character style that is not suitable as a summary in the summary translation result output from the output document generation means 7 is suitable for the summary. It also has the effect of being converted into a different character style.

(E) Other Embodiments In each of the above embodiments, the summary translation process is always executed. However, the user specifies the summary translation process or the normal translation process via the input / output unit 2. You may allow it. In this case, regarding the second embodiment,
Even when the abstract translation process is selected, whether or not to translate the character string in the image may be designated. Further, in the third embodiment, whether or not to translate the character string in the image may be designated even when the abstract translation process is selected.
It may also be possible to specify whether to translate the image and / or the linked document. Further, regarding the third embodiment, even when the link destination document is translated, it may be possible to specify the depth to which the link destination document is translated. Furthermore,
In the third embodiment, the original input document may be translated in full without being summarized, and the linked document may be translated in summary.

To switch the processing as described above, a branching process for taking in the processing method selected by the user and branching the processing according to the fetched processing method is appropriately provided in the series of processing shown in the above-mentioned flowchart. Can be achieved by

In each of the above embodiments, the link object is fetched via the network. However, when the machine translation system is constructed on an independent information processing system, a link document storage means is provided. The link object may be fetched from this link document storage means. Further, it is possible to allow both the fetching of the link object from the network and the fetching of the link object from the link document storage means provided in the machine translation system.

Further, in the second or third embodiment, when a non-language document which is a link destination document is included in an output document, it always includes a translation of a character string in the non-language document. However, the non-language document may be included as it is.

Furthermore, in the above embodiment, the translation target tag information storage means 5 has been described as one which can be registered and edited by the user, but it may be fixedly provided in the system, and user editing is possible. It is also possible to provide a fixed one and a fixed one.

In the above embodiment, the machine translation system for a document with a tag is shown. However, even if it is different from the tag information consisting of a text data string similar to the text portion, the tag in the document with a tag is different. The technical idea of the present invention can be applied to any machine translation of a document accompanied by an output format that performs the same function as information and information that defines a linked document.

Furthermore, in the above embodiment, the present invention is applied to the machine translation system, but the present invention can be applied to other document processing systems. Many document processing systems (natural language processing systems) execute analysis processing similar to that of a machine translation system without executing conversion to a target language, and the present invention is applied to such a document processing system. Can be applied, and the analysis target of the document can be limited using the tag information. Further, the machine translation system of the above embodiment may be configured with a document processing system (abstract creating system) having only a configuration for extracting a predetermined portion of a document using tag information. Even in this case, perform the same process for the linked document,
You may make it extract a predetermined part.

In the fourth embodiment described above, the character style of the summary document can be converted based on the tag information, but a style of a unit larger than the character may be converted.

[0213]

As described above, according to the present invention, display,
In a document processing system that processes documents with format specification information that specifies the format for print output, (1) Store a specific pattern of format specification information for extracting a character string of a specified type in the input document. Extraction target specific information storage means, and (2) a format specification information identification extraction means for extracting a portion of the input document that matches a specific pattern of the format specification information stored in the extraction target specific information storage means (3) Since it has an output document generating means for generating an output document by arranging the extracted portion or arranging the extracted portion processed thereafter, depending on the storage content of the extraction target specific information storage means, , Not only a predetermined part of the input document,
A non-language document or a linked document can be taken out, and an output document can be formed in which the special document information is also reflected.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating an overall configuration of a first embodiment.

FIG. 2 is an explanatory diagram showing a configuration example of a translation target tag information storage unit of the first exemplary embodiment.

FIG. 3 is a diagram showing an input document used to describe a specific operation example of the first exemplary embodiment.

FIG. 4 is a view showing a display screen of the input document of FIG.

FIG. 5 is a flowchart showing the overall operation of the first embodiment.

FIG. 6 is a flowchart showing the operation of the tag identifying means of the first embodiment.

FIG. 7 is a flowchart showing an operation of the translation means according to the first exemplary embodiment.

FIG. 8 is a flowchart showing the operation of the output document generation means of the first exemplary embodiment.

FIG. 9 is an example of an output document according to the operation of the first embodiment (FIG.
Corresponding to the input document of FIG.

10 is a diagram showing a display screen of the output document of FIG.

FIG. 11 is a block diagram showing an overall configuration of a second embodiment.

FIG. 12 is a flowchart showing the overall operation of the second embodiment.

FIG. 13 is a flowchart showing the operation of the tag identifying means according to the second embodiment.

FIG. 14 is a flowchart showing an operation of the link object acquisition means of the second exemplary embodiment.

FIG. 15 is a flowchart showing an operation of the encoding means of the second exemplary embodiment.

FIG. 16 is a diagram showing an output document example (corresponding to the input document of FIG. 3) according to the operation of the second embodiment.

17 is a diagram showing a display screen of the output document of FIG.

FIG. 18 is a block diagram showing an overall configuration of a third embodiment.

FIG. 19 is a flowchart showing the overall operation of the third embodiment.

FIG. 20 is a flowchart showing the operation of the tag identifying means of the third exemplary embodiment.

FIG. 21 is a flow chart showing the operation of the link object acquisition means of the third exemplary embodiment.

FIG. 22 is a diagram showing an output document example (corresponding to the input document in FIG. 3) according to the operation of the third embodiment.

FIG. 23 is a diagram showing a display screen of the output document of FIG. 22.

FIG. 24 is a block diagram showing a detailed configuration of a characteristic part of the fourth embodiment.

FIG. 25 is an explanatory diagram of a configuration example of a character style conversion reference table according to the fourth embodiment.

FIG. 26 is an explanatory diagram of a registration example of a character style registration / edit table according to the fourth embodiment.

FIG. 27 is an explanatory diagram of a configuration and registration example of a character style conversion matching table according to the fourth embodiment.

FIG. 28 is a processing flowchart (1) of a character style processing control unit according to the fourth embodiment.

FIG. 29 is a processing flowchart (2) of the character style processing control unit according to the fourth embodiment.

FIG. 30 is a processing flowchart (1) of a character style conversion determination processing unit according to the fourth embodiment.

FIG. 31 is a processing flowchart (2) of the character style conversion determination processing unit of the fourth embodiment.

[Explanation of symbols]

1 ... Network, 2 ... Input / output means, 3 ... Input document (buffer), 4 ... Tag identification means, 5 ... Translation target tag information storage means, 6 ... Translation means, 7 ... Output document generation means, 8 ... Output document ( Buffer), 9 ... link object acquisition means, 10 ... encoding means, 11 ... object storage means,
12 ... Character style conversion means.

Claims (9)

[Claims] 1. A document processing system for processing a document with format designation information for designating a format at the time of display and print output, for extracting a character string of a predetermined type in an input document,
An extraction target specific information storage unit that stores a specific pattern of format specification information and a format that extracts a part of the input document that matches the specific pattern of the format specification information stored in the extraction target specific information storage unit A document processing system comprising: designated information identifying and extracting means; and output document generating means that prepares an extracted document by preparing an extracted portion or by preparing an extracted portion processed thereafter. 2. The format specification information for style specification, which is included in the output document and defines the modified state at the time of display and print output, is recognized, and when the format specification information for the predetermined style is specified, other 2. The document processing system according to claim 1, further comprising a style conversion unit that converts the style specification information for style specification or deletes the predetermined style specification format specification information. 3. A link destination document acquisition means for acquiring a link destination document by extracting and activating a link destination document in a certain document, and a link destination of a document including the format designation information as one type of the format designation information. Storing a specific pattern of the format designation information capable of extracting whether or not the input document includes the format designation information for designating such a linked document, in the extraction target specific information storage means. By the way, in the input document, the format designation information identification and extraction means
If there is a portion that matches the specific pattern of the format designation information for designating the linked document stored in the extraction target specific information storage means, the acquisition of the linked document by the linked document acquisition means is activated. The document processing system according to claim 1, wherein the document processing system is a document processing system. 4. When the link destination document acquired by the link destination document acquisition means is a language document, the format specification information identifying and extracting means also performs an extraction operation for the language document, and the output document generation. Output means prepares a portion extracted from the input document and the acquired link destination document, or prepares an extracted portion from the input document and the acquired link destination document subjected to a predetermined process after extraction The document processing system according to claim 1, wherein the document processing system generates a document. 5. The output document generation means, when the link destination document acquired by the link destination document acquisition means is a non-language document such as a figure or a table, includes the non-language document as it is in the output document. The document processing system according to any one of claims 1 to 4, wherein: 6. The method according to claim 1, further comprising processing means for executing a predetermined process on the portion of the input document extracted by the format designation information identifying and extracting means. Document processing system described. 7. The document processing system according to claim 1, wherein the processing means is a translation means for translating a source language into a target language. 8. The encoding means for recognizing a character string pattern, which may be linguistic information, from the image data of the non-language document acquired by the linked document acquisition means, and encoding the character string pattern, and the translating means. 9. The document processing system according to claim 7, wherein the output document generation means includes the translated word of the encoded character string in the output document. 9. The document with the format designation information for designating the format at the time of display and print output is a tagged document in which the format designation information is represented by a character code string similar to the text body of the document. The document processing system according to any one of claims 1 to 8.