JPH09265469A - Translation method for hyper text type document and translation device for html document - Google Patents

Abstract

PROBLEM TO BE SOLVED: To translate the anchor of a hyper link in a hyper text type document and to leave it without being translated at it is by making a user select an anchor translation mode or a non-translation mode. SOLUTION: When the user clicks the anchor of the hyper link included in the document displayed by a browser 1, the document name of a link destination and a translation command included in corresponding hyper link information are sent from the browser 1 to a supervisor 2. The supervisor 2 sets a flag F, transmits only the document name to a server 4 and receives the pertinent document from the server 4. Since the flag F is set, the supervisor 2 translates the transmitted document into Japanese after adding the translation command to each hyper link information included in the transmitted document. At the time, the information other than a text such as a document name, URL and a tag, etc., is not translated. Then, after resetting the flag F, a translation sentence is sent to the browser 1.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for translating a hypertext type document obtained through a network such as the Internet, and a text in a first language.
The present invention relates to an HTML document translation device for converting a TML document into an HTML document whose text is described in a second language.

[0002]

2. Description of the Related Art WWW (world wide web) is a distributed information system based on the concept of hypertext. When a user requests a document placed on the WWW server (Server) on the Internet, the document name URL (Un
Specify the iform Resource Locator from the browser. Then, from the browser, WWW with the document
A send request is sent to the server. W with that document
The document is sent from the WW server to the browser.

In a document (hypertext) sent from the WWW server, a link (hyperlink information) pointing to another document or the like is embedded. Hypertext is usually described in a language called HTML (Hypertext markup language). In HTML, the hyperlink information has a start tag <A ...>, as shown in FIG.
And the end tag </A>. And the linked information is U like "http://www.abc.co.jp/"
It is specified by RL.

A start tag <A ...> and an end tag </A>
A hyperlink anchor such as a keyword such as "ABC homepage" is inserted between the> and. When the keyword that is the anchor of the hyperlink is clicked, the browser sends a transmission request to the server having the linked document. The server having the linked document sends the document to the browser. Therefore,
The linked document is displayed.

By the way, when translating hypertext into, for example, English to Japanese, the tags enclosed by the symbols "<" and ">" or the symbols "</" and ">" are translated as the original text. Not done. The other English sentences are translated into Japanese. Therefore, the keyword that is the anchor of the hyperlink included in the hyperlink information is also translated.

[0006] Since the keyword which is the anchor of the hyperlink has a jump function to the contents related to it, it is difficult to specify an appropriate jump destination if the keyword is not translated properly. For example, if the keyword is "trial" meaning "trial" and it is translated as "trial" instead of "trial", assume "trial" from the Japanese sentence "trial". It is difficult. Therefore, it becomes difficult to obtain information on “trial”.

[0007]

SUMMARY OF THE INVENTION According to the present invention, a translated sentence in which a hyperlink anchor in a hypertext document is translated or a translation in which the hyperlink anchor is left untranslated according to a user's selection. An object of the present invention is to provide a method of translating a hypertext type document that can obtain a sentence.

Further, according to the present invention, an HTML document whose text is written in a first language and an HTML document whose text is written in a second language are provided.
It is an object of the present invention to provide an HTML document translation device that can be automatically converted into a TML document.

[0009]

According to the present invention, in a method of translating a hypertext type document, a user selects an anchor translation mode for translating an anchor of a hyperlink or an anchor non-translation mode in which the original text remains unchanged, When the anchor translation mode is selected, the hyperlink anchor is translated when the hypertext document is translated, and when the anchor non-translation mode is selected, the hypertext document is translated. At that time, the feature is that the anchor of the hyperlink is not translated and remains as the original text.

The hyperlink information includes, for example, a start tag indicating the hyperlink information, an end tag indicating the hyperlink information, and a hyperlink anchor sandwiched between the start tag and the end tag. When the anchor translation mode is selected, the start tag and the end tag are recognized as one word. The start tag and the end tag are stored in association with the anchor. After translating only the anchor, the start tag and the end tag are added to the translation of the anchor.

When the anchor non-translation mode is selected, the entire hyperlink information is recognized as one word, so that the hyperlink information is not translated and remains as the original text.

According to the method of translating a hypertext document according to the present invention, the translated sentence in which the hyperlink anchor in the hypertext document is translated or the hyperlink anchor is not translated according to the selection by the user. You can get the translated text that is left as it is.

A first HTML document translating apparatus according to the present invention converts an HTML document whose text is written in a first language,
The text is converted into an HTML document described in a second language.

In the first HTML document translation device according to the present invention, the original sentence is searched for the first tag that modifies the entire sentence, and if the first tag that modifies the entire sentence is present,
The sentence in which the first tag that modifies the entire sentence is removed from the original sentence is the target sentence of morphological analysis. Then, a tag that modifies the entire sentence is associated with the target sentence of the morphological analysis and stored.

Morphological analysis including word segmentation and grammatical analysis is performed on the target sentence of morphological analysis. Based on this morphological analysis result, from the target sentence of the morphological analysis,
If a second tag that modifies one or more words is sought, and a second tag that modifies one or more words is present,
The start tag of the start tag and the end tag forming the second tag is stored in association with the first word immediately after that, and the end tag is stored in association with the second word immediately before that.

A syntactic analysis and a semantic analysis are performed on the sentence from which the second tag has been removed from the morphological analysis target sentence. Based on this analysis result, the second sentence is added to each word constituting the sentence in which the second tag is removed from the target sentence of the morphological analysis.
Language translations are assigned. Also, each translated word is assembled to form a sentence in the second language, and a translated sentence in the second language is generated.

In the generated translated sentence, a start tag stored in association with the first word is added to the front side of the translated word corresponding to the first word, and the translated tag corresponds to the second word. The end tag stored in association with the second word is added to the rear side of the translated word. A first tag that modifies the entire sentence is added to the sentence thus obtained.

A second HTML document translating device according to the present invention is an HTML document whose text is described in a first language,
The text is converted into an HTML document described in a second language.

In the second HTML document translation device according to the present invention, the original sentence is searched for a tag that modifies only one character, and if the first tag that modifies only one character exists, the entire sentence and its tag are searched. The position of the first tag is changed so that the first tag modifies either one of the words including the character to be modified.

The original sentence after the position of the first tag is changed is searched for a second tag that modifies the entire sentence, and when a second tag that modifies the entire sentence is present, the entire sentence is modified from the original sentence. The sentence from which the second tag to be modified is removed is the target sentence of morphological analysis. Then, a tag that modifies the entire sentence is associated with the target sentence of the morphological analysis and stored.

Morphological analysis including word segmentation and grammatical analysis is performed on the target sentence of morphological analysis. Based on this morphological analysis result, from the target sentence of the morphological analysis,
If a third tag that modifies one or more words is sought and there is a third tag that modifies one or more words,
The start tag of the start tag and the end tag forming the third tag is stored in association with the first word immediately after that, and the end tag is stored in association with the second word immediately before that.

The sentence from which the third tag has been removed from the target sentence of morphological analysis is subjected to syntactic analysis and semantic analysis. Based on this analysis result, a translated word in the second language is assigned to each word that constitutes the sentence in which the third tag has been removed from the target sentence for morphological analysis. Also, each translated word is assembled to form a sentence in the second language, and a translated sentence in the second language is generated.

In the generated translated sentence, a start tag stored in association with the first word is added to the front side of the translated word corresponding to the first word, and the translated tag corresponds to the second word. The end tag stored in association with the second word is added to the rear side of the translated word. A second tag that modifies the entire sentence is added to the sentence thus obtained.

According to the HTML document translation apparatus of the present invention, an HTML document whose text is described in the first language is
The text can be automatically converted into an HTML document described in the second language. Therefore, it becomes possible to automatically convert the HTML document created in the language of the own country into the HTML document of the language of another country.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention applied to a document conversion system for converting an English document into a Japanese document will be described below with reference to the drawings.

(1) Description of the first embodiment

FIG. 1 shows the configuration of a document conversion system for converting an English document into a Japanese document.

This system comprises a personal computer (user side device) 10 provided on the user side and a WWW server (server side application) 4 connected to the personal computer 10 via the Internet 3. The personal computer 10 includes a browser (user-side application) 1 and a supervisor (data monitoring means) 2. As shown in FIG. 2, the supervisor 2 includes a data monitoring unit 21, a translation processing unit 22, and a cache memory 23.

FIG. 3 shows the flow of data among the browser 1, the supervisor 2 and the server 4. In FIG. 3, parenthesized numbers indicate sequence numbers.

First, the user selects W on the Internet 3.
The browser uses the name URL of the document placed on the WW server 4
From the browser 1, the document name is transmitted from the browser 1 to the server 4 having the document via the supervisor 2 (sequences NO. 1 and NO. 2).

The server 4 searches the database for a corresponding document (original text) based on the sent document name,
Transmit to supervisor 2 (sequence No. 3).
The supervisor 2 adds tag information indicating the English-Japanese translation instruction (hereinafter simply referred to as translation instruction) to each hyperlink information included in the sent document (hypertext), and then obtains the document ( The corrected original text) is sent to the browser 1 (sequence No. 4). FIG. 11 shows an example in which a translation command is added to the hyperlink information of FIG. In this example, "EJJ" is the translation instruction.

After that, when the user clicks the anchor (keyword or the like) of the hyperlink included in the document displayed by the browser 1, the document name of the link destination included in the corresponding hyperlink information. And the translation instruction is sent from the browser 1 to the supervisor 2 (sequence NO. 5). When the linked document name and the translation instruction are sent from the browser 1 to the supervisor 2, the supervisor 2 sets the flag F (F = 1). Then, the supervisor 2 selects from among the document name and the translation instruction sent from the browser 1,
Only the document name is transmitted to the server 4 having the document (sequence No. 6).

The server 4 retrieves the corresponding document from the database based on the sent document name and sends it to the supervisor 2 (sequence No. 7). Since the flag F is set, the supervisor 2 adds a translation command to each hyperlink information included in the sent document, and then translates the sent document into Japanese. At this time, information other than the text, such as the document name URL and tags, is not translated. Then, the flag F is reset (F =
Then, the translated document (translated sentence) is sent to the browser 1 (sequence No. 8).

After that, every time the user clicks the anchor of the hyperlink included in the displayed document, the sequence No. 5 to NO. The same process as 8 is performed.

The translation processing unit may be provided on the server 4 side. In this case, when the document name and the translation instruction are sent from the browser 1 to the supervisor 2, the supervisor 2 sends both the document name and the translation instruction to the server 4. When the document name and the translation command are sent, the server 4 reads the document from the database, translates the document, and sends the translated document to the supervisor 2. The supervisor 2 adds a translation command to each hyperlink information included in the sent document (translated sentence) and sends it to the browser 1.

(2) Description of the second embodiment

The system configuration is the same as in FIG. Second
In the first embodiment, when the supervisor 2 adds the translation instruction to each hyperlink information included in the document sent from the server 4 in the first embodiment, the supervisor 2 further The hyperlink information whose link destination is its own document is embedded in the document. A button (hereinafter, referred to as own document designation button) is used as an anchor of hyperlink information whose link destination is the own document. The processing of the supervisor 2 in this case will be described.

FIG. 4 shows the flow of processing when the user designates from the browser 1 the name URL of the document placed on the server 4 on the Internet 3.

When the user designates the document name URL from the browser 1, the browser 1 transmits the document name to the server 4 having the document via the supervisor 2 (sequence NO. 11, NO). .12).

The server 4 retrieves the corresponding document from the database based on the sent document name and sends it to the supervisor 2 (sequence No. 13). When the document specified by the server 4 is sent to the supervisor 2, the supervisor 2 adds a translation instruction to each hyperlink information included in the sent document and adds the translation command to the sent document. Embed the hyperlink information whose link destination is its own document. At this time, a translation command is added to the hyperlink information whose link destination is its own document. The document thus created is stored in the cache memory 23 and sent to the browser 1 as a modified original sentence (sequence No. 14).

When the user clicks the own document designation button while the corrected original text is displayed by the browser 1, the document name of the own document is sent to the supervisor 2 together with the translation instruction (sequence No. 15). When the supervisor 2 receives the document name of the self-document and the translation instruction, the supervisor 2 reads the modified original sentence stored in the cache memory 23. Then, after removing the translation command added to the hyperlink information of the self-document as the link destination from the read original text, the obtained document is translated into Japanese. The document created in this manner is stored in the cache memory 23 as a translated sentence and is sent to the browser 1 (sequence NO. 16).

When the user clicks the own document designation button while the translated text is being displayed by the browser 1, in this case, the translation instruction is not added to the hyperlink information whose link destination is the own document. , Only the document name of its own document is sent to the supervisor 2 (sequence No. 17). Therefore, in this case, the supervisor 2 reads out the modified original text stored in the cache memory 23 and sends it to the browser 1 (sequence N).
O. 18).

When the user clicks the own document designation button while the corrected original text is displayed by the browser 1, in this case, the translation instruction is added to the hyperlink information whose link destination is the own document. , And a translation command together with the document name of its own document are sent to the supervisor 2 (sequence No. 19). Therefore, in this case, the supervisor 2 reads the translated sentence stored in the cache memory 23 and sends it to the browser 1 (sequence No. 20).

FIG. 5 shows the flow of processing when the user clicks on the hyperlink anchor included in the displayed document.

When the user clicks the anchor (keyword or the like) of the hyperlink included in the displayed document, the link destination document name and the translation instruction included in the corresponding hyperlink information are displayed. Browser 1
Sent to the supervisor 2 (sequence NO.
21). When the linked document name and the translation instruction are sent from the browser 1 to the supervisor 2, the supervisor 2 sets the flag F (F = 1). Then, the supervisor 2 transmits only the document name among the document name and the translation instruction sent from the browser 1 to the server 4 having the document (sequence NO.22).

The server 4 retrieves the corresponding document from the database based on the sent document name and sends it to the supervisor 2 (sequence NO.23). Since the flag F is set, the supervisor 2 first creates a modified original sentence and stores it in the cache memory 23, and then creates a translated sentence and stores it in the cache memory 23. Then, after resetting the flag F (F = 0), the translated sentence is sent to the browser 1. Therefore, the translated text is displayed by the browser.

That is, the supervisor 2 adds a translation command to each hyperlink information included in the sent document, and embeds the hyperlink information whose link destination is its own document in the sent document. . On this occasion,
A translation instruction is added to the hyperlink information whose link destination is its own document. The document created in this way is
The modified original text is stored in the cache memory 23.

Further, the supervisor 2 reads the modified original text from the cache memory 23. Then, after removing the translation instruction from the hyperlink information in which the link destination included in the read corrected original text is the own document, this document is translated into Japanese. The document created in this way is stored in the cache memory 23 as a translated document and sent to the browser 1 (sequence NO. 24).

When the user clicks the own document designation button while the translated text is displayed by the browser 1, in this case, the hyperlink information whose link destination is the own document has no translation instruction added. , Only the document name of its own document is sent to the supervisor 2 (sequence No. 25). Therefore, in this case, the supervisor 2 reads out the modified original text stored in the cache memory 23 and sends it to the browser 1 (sequence N).
O. 26).

When the user clicks the own document designation button while the corrected original text is displayed by the browser 1, in this case, the translation instruction is added to the hyperlink information whose link destination is the own document. , A translation command is sent to the supervisor 2 together with the document name of its own document (sequence NO.27).
Reads the translated sentence stored in the cache memory 23 and sends it to the browser 1 (sequence NO. 28).

The translation processing unit may be provided on the server 4 side. In this case, when the document name and the translation instruction are sent from the browser 1 to the supervisor 2, the supervisor 2 sends both the document name and the translation instruction to the server 4. When the document name and the translation instruction are sent, the server 4 retrieves the document from the database, translates the document, and sends the translated document to the supervisor 2. Supervisor 2 sends the sent document (translated sentence)
A translation command is added to each hyperlink information included in, and the hyperlink information whose link destination is its own document is embedded and sent to the browser 1.

(3) Description of the third embodiment

The system configuration is the same as in FIG. Third
In the embodiment, the user can select whether to display the original sentence or the translated sentence.

FIG. 6 shows the flow of processing when the user designates from the browser 1 the name URL of a document placed on the server 4 on the Internet 3.

When the user designates the document name URL from the browser 1, the browser 1 transmits the document name to the server 4 having the document via the supervisor 2 (sequence NO. 31, NO). .32).

The server 4 retrieves the corresponding document from the database based on the sent document name and sends it to the supervisor 2 (sequence NO.33). When the document specified by the server 4 is sent to the supervisor 2, the supervisor 2 embeds hyperlink information whose link destination is its own document in the sent document.

A translated sentence selection button or an original sentence selection button is used as an anchor of hyperlink information whose link destination is its own document. In this case, the translated sentence selection button is used as an anchor of the hyperlink information whose link destination is the own document. When the translated sentence selection button is used, the translation instruction is added to the hyperlink information whose link destination is the own document. The document created in this way is stored in the cache memory 23 as a modified original text and sent to the browser 1 (sequence N).
O. 34).

After that, when the user clicks the translated sentence selection button, the translation instruction is sent from the browser 1 to the supervisor 2 together with the name of the own document which is the link destination (sequence NO. 35). When the self-document name and the translation command are sent, the supervisor 2 reads the modified original text from the cache memory 23, and the hyperlink information other than the hyperlink information in which the link destination included in the modified original text is the own document. A translation command is added to the link information, and the translation command is removed from the hyperlink information whose link destination is its own document. Further, the translated sentence selection button is changed to the original sentence selection button. Then, the obtained document is translated into English and Japanese. The document created in this way is stored in the cache memory 23 as a translated document and is also sent to the browser 1 (sequence NO. 36).

After that, when the user clicks the original text selection button, only the own document name as the link destination is displayed in the browser 1.
Sent to supervisor 2 (sequence NO.3
7). When only the self-written name is sent from the browser 1 in this way, the supervisor 2 reads the modified original text from the cache memory 23 and sends it to the browser 1 (sequence NO.38).

After this, when the user clicks the translated sentence selection button, the translation instruction is sent from the browser 1 to the supervisor 2 together with the name of the own document which is the link destination (sequence No. 39). When the self-document name and the translation command are sent, the supervisor 2 reads the translated document from the cache memory 23 and sends it to the browser 1 (sequence N).
O. 40).

If the user clicks the keyword of the hyperlink (excluding the translated text selection button) included in the displayed document while the corrected original text is displayed by the browser 1, the browser 1 Since only the document name of the link destination is sent from the supervisor to the supervisor 2, the processing flow is the same as in FIG.

FIG. 7 shows a process when the user clicks a hyperlink keyword (excluding the original text selection button) included in the displayed document while the translated text is being displayed by the browser 1. Shows the flow of.

When the user clicks the keyword of the hyperlink included in the displayed document while the translated text is being displayed by the browser 1, the link destination included in the corresponding hyperlink information is displayed. The document name and the translation instruction are sent from the browser 1 to the supervisor 2 (sequence NO.41). When the linked document name and the translation instruction are sent from the browser 1 to the supervisor 2, the supervisor 2 sets the flag F (F = 1). Then, the supervisor 2 transmits only the document name among the document name and the translation instruction sent from the browser 1 to the server 4 having the document (sequence NO.42).

The server 4 retrieves the corresponding document from the database based on the sent document name and sends it to the supervisor 2 (sequence No. 43). Since the flag F is set, the supervisor 2 first creates a modified original sentence and stores it in the cache memory 23. Then, after resetting the flag F (F = 0), the translated sentence is sent to the browser 1. Therefore, the translated text is displayed by the browser.

That is, the supervisor 2 first embeds hyperlink information whose link destination is its own document in the sent document. At this time, the translation instruction is added to the hyperlink information whose link destination is the own document. A translated sentence selection button is used as an anchor of hyperlink information whose link destination is its own document. The modified original text thus created is stored in the cache memory 23.

Next, the supervisor 2 reads the modified original text. Then, the translation instruction is added to the hyperlink information other than the hyperlink information in which the link destination included in the read corrected original text is the own document, and the translation instruction is removed from the hyperlink information in which the link destination is the own document. . Also, the button that is the anchor of the hyperlink information whose link destination is the own document is changed to the original text selection button. The document thus obtained is translated into English and then stored in the cache memory 23 as a translated sentence and sent to the browser 1 (sequence NO.44). Therefore, the translated text is displayed by the browser 1.

When the original sentence selection button is clicked while the translated sentence is displayed by the browser 1, the own document name is sent from the browser 1 to the supervisor 2 (sequence No. 45). The supervisor 2 reads the modified original text from the cache memory 23 and sends it to the browser 1 (sequence NO.46). Therefore, the corrected original text is displayed by the browser 1.

When the translated sentence selection button is clicked while the corrected original sentence is displayed by the browser 1, the browser 1 sends a translation instruction together with the name of the original document to the supervisor 2 (sequence No. 47). The supervisor 2 reads the translated sentence from the cache memory 23 and sends it to the browser 1 (sequence No. 48). Therefore, the translated text is displayed by the browser 1.

In each of the above embodiments, the tag information indicating the English-Japanese translation instruction is added to the hyperlink information.
Hyperlink information, tag information indicating Japanese-English translation instruction,
Tag information indicating another document conversion instruction, such as tag information indicating a Kanji code conversion instruction, may be added.

FIG. 8 shows a processing procedure when the translation processing unit 22 performs English-Japanese translation in each of the above embodiments. Here, as shown in FIG. 1, the translation processing unit 22 is provided in the supervisor 2.

When the hypertext is translated into English, the start tag or the symbol "" enclosed by the symbols "<" and ">"
The end tag enclosed between </ "and the symbol">"is not translated and remains as the original sentence. The other parts of the English sentence are translated into Japanese sentence. Therefore, it is generally included in the hyperlink information. Keywords that are anchors of hyperlinks are also translated.

Since the keyword that is the anchor of the hyperlink has a jump function to the content related to it, it is difficult to specify a proper jump destination if the keyword is not translated properly. For example, if the keyword is "trial" meaning "trial" and it is translated as "trial" instead of "trial", assume "trial" from the Japanese sentence "trial". It is difficult. Therefore, it becomes difficult to obtain information on “trial”.

In this embodiment, the user can select whether to translate the keyword that is the anchor of the hyperlink when performing English-Japanese translation (keyword translation mode) or not (keyword non-translation mode). ing.

A procedure when the translation processing unit 22 performs English-Japanese translation will be described. Here, as shown in FIG. 8, the original text to be translated is the word “Introd”.
ucing ", hyperlink information, word" is ", word" d "
It is assumed that the document is composed of "ifficult". The hyperlink information includes a start tag <A ...>, an end tag </A>,
It consists of two words "this" and "trial" inserted between them. "this trial" is the anchor (keyword) of the hyperlink.

When the original sentence to be translated is input (step 1), it is determined whether or not it is the mode for translating the keyword that is the anchor of the hyperlink (step 2). This determination is made depending on whether the user has set the keyword translation mode or the keyword non-translation mode.

When the keyword translation mode is set (YES in step 2), word division is first performed (step 3). In this case, as shown in FIG. 8, the input original sentence is divided for each tag and each word other than the tag.

Next, the tag is attributed (step 4). That is, the tag is attributed to the words before and after it. As shown in FIG. 8, the start tag <A ...> in the hyperlink information is attributed to the word “this” immediately after it, and the end tag </A> is attributed to the word “trial” immediately before it. It

After this attribute conversion, the main processing of translation is executed (step 5). In other words, sentences consisting of word strings other than tags are translated into English and Japanese. In this case, the keyword "this trial" is also translated.

Next, the tags are surface-layered (step 6). That is, the attributed tags are added to the translated text. As a result, a translated sentence "<A ...> It is difficult to introduce this trial </A>" is output. When such a translated sentence is sent to the browser, the actual sentence other than the tag is displayed on the monitor. Therefore, the sentence "It is difficult to introduce this attempt" is displayed.

When the keyword non-translation mode is set in step 2 above (N in step 2
O), first, word division is performed (step 7). In this case, as shown in FIG. 8, the hyperlink information <A ...> this trial </A> is treated as one word. For tags other than the tags used as hyperlink information, each tag is treated as one word. Words other than tags that are not keywords included in the hyperlink information “Introducin
Each of g "," is "and" difficult "is treated as one word.

Next, the main processing of translation is executed (step 8). In this case, since the word dictionary included in the translation processing unit 22 does not have a word corresponding to the hyperlink information, the word is not translated. Therefore, a sentence consisting of a word string other than hyperlink information is translated into English and Japanese. Therefore, a translated sentence "It is difficult to introduce <A ...> this trial </A>" is output. That is, the keyword of the hyperlink is not translated and remains as the original text. When such a translated sentence is sent to the browser, the actual sentence other than the tag is displayed on the monitor.
Therefore, the sentence "It is difficult to introduce this trial" is displayed. That is, the keyword is displayed as it is in the original text.

FIG. 9 shows a more detailed processing procedure when the translation processing unit 22 performs English-Japanese translation. Here, as shown in FIG. 1, the translation processing unit 22 is provided in the supervisor 2.

By the way, as tags used in HTML, in addition to tags for hyperlinks, tags <TITLE> and </ TITLE> indicating a title, and a tag <that indicates information to be placed in the head portion HEAD> and </ HE
AD>, a tag indicating that it is a graphic <IMG SRC
= "URL", a tag indicating that it is displayed in bold font <B
> And </ B>, the tag <CODE indicating that it is a code
> And </ CODE>, and a tag <Hn> that indicates a heading
Tags such as and </ H> are defined.

In the case of translating a document including such a tag, an incorrect translated sentence may be generated unless pre-processing that recognizes the tag is performed. In order to solve such a problem, it is conceivable to delete the tag information at the preprocessing stage. However, this method has a problem that attributes such as style information are missing and the style of the original sentence cannot be reflected in the translated sentence. In addition, even if the tag appearance position in the original sentence is stored in order to generate a translated sentence that maintains the original sentence style, when translating between languages in which the word order changes greatly, such as English and Japanese, the original sentence Is not effective because the positional relationship of the words in the translated sentence changes.

Therefore, a method is considered in which tag information such as style information is provided as a word attribute in the preprocessing stage. In the attributeization of tags, it is necessary to classify them into several categories according to the scope of modification by tags. For example,
In the following example sentence (1), the modification range of the tags <H3> and </ H3> representing the headline is the entire sentence. Also, a tag that represents a hyperlink <A HREF = "” hypertext4 / what2.
The modification range of html "> and </A> is only the word" here ".

<H3> Japanese Page is <A HREF="hypertext4/what2.html"> here </A> </ H3>… Example sentence (1)

In the following example sentence (2), tags <FONT SIZE = + 3> and </ FO
The range of NT> is only one letter "N".

<FONT SIZE = + 3> N </ FONT> ETWORK ... Example sentence (2)

That is, the tag modification range is (a)
Whole sentence, (b) word (however, if not modifying the whole sentence, including modifying two or more words), (c)
There are three types of letters. In the translation process of FIG. 9, a sentence attribute process is performed on a tag that modifies the entire sentence, a word attribute process is performed on a tag that modifies a word, and a character attribute process is performed on a tag that modifies a character. Be done.

The tag <A indicating the hyperlink information
...> and </A> may modify one or more words as shown in the example sentence (3) and may modify the entire sentence as shown in the example sentence (4).

<B> <FONT SIZE = + 3> J </ FONT> apanese Page is <A HREF="/misc/index.html"> Her e </A>. </ B>… Example sentence (3)

<A HREF="/misk/index.html"> Japanese Page is Here </A>. … Example sentence (4)

In the example sentence (3), the tags <B> and <
/ B> is a tag for displaying the English sentence (entire sentence) sandwiched between them in bold type. In the original text, tags <FONT SIZE = + 3> and </ FONT> are tags for designating the font size of the characters sandwiched between them. Also, the tag <A HREF = ”/ misc / index.ht
ml "> and </A> are tags indicating hyperlink information, and the word" Here "sandwiched between these tags indicates that they are anchors (keywords) of hyperlinks.

A procedure when the translation processing unit 22 performs English-Japanese translation will be described. Here, it is assumed that the original sentence to be translated is the English sentence shown in the above example sentence (3).

<B> <FONT SIZE = + 3> J </ FONT> apanese Page is <A HREF="/misc/index.html"> Her e </A>. </ B>… Example sentence (3)

When the original sentence to be translated is input (step 11), it is determined whether or not the mode is the translation of the keyword that is the anchor of the hyperlink (step 1).
2). This determination is made depending on whether the user has set the keyword translation mode or the keyword non-translation mode.

A case where it is determined in step 12 that the keyword translation mode is set will be described. When the keyword translation mode is set (YES in step 12), first, character attribute processing is performed (step 21). That is, the tag that modifies the character is searched. When a tag that modifies a character is found, the position of that tag is changed so that the tag modifies the entire sentence.

In the above original sentence (3), the tag <FONT SIZE = + 3>
Since / FONT> is a tag that modifies characters, the original text is converted into the following English text (5) by the character attribute processing.

<B> <FONT SIZE = + 3> Japanese Page is <A HREF="/misc/index.html"> Here </A>. </ B> </ FONT>… English (5)

In the character attribute processing, a tag that modifies a character modifies a word including the character,
The position of the tag may be changed.

When the character attribute process is completed, the sentence attribute process is executed for the sentence after the character attribute process is performed (step 22). That is, the tag that modifies the entire sentence is searched. When a tag is present at the beginning of a sentence and a tag paired with the tag does not exist in the middle of the sentence, the tag is determined to be the tag at the beginning of the sentence that modifies the entire sentence. If there is a tag at the end of the sentence, that tag is determined to be the tag at the end of the sentence that modifies the entire sentence.

In the above English sentence (5), start tags <B> and <FON
It is determined that T SIZE = + 3> is the tag at the beginning of the sentence that modifies the entire sentence. Also, the end tags </ B> and </ FONT> are discriminated as tags at the end of the sentence that modify the entire sentence. Then, of the English sentence (5), the next English sentence (6) excluding the tag that modifies the entire sentence is targeted for morphological analysis.

Japanese Page is <A HREF="/misc/index.html"> Here </A> ... English sentence (6) A tag at the beginning of a sentence that modifies the entire sentence and a tag at the end of a sentence that modifies the entire sentence. Are stored in association with the target of morphological analysis.

English sentence which is judged as a target of morphological analysis (6)
Is subjected to morphological analysis processing (step 2
3). In the morphological analysis in step 23, since the keyword translation mode is set, the start tag <A ...> And the end tag </A> in the hyperlink information are also treated as one word.

The morphological analysis result for English sentence (6) is shown in Table 1 below.

[0106]

[Table 1]

After the morphological analysis processing, word attribute processing is performed (step 24). In the word attribute processing, the start tag is attributed to the word immediately after it as the prefix tag information among the pair of tags that modify the word. That is, the start tag is stored as the leading tag information in association with the word immediately following it. Also, the end tag is attributed to the word immediately before it as the post tag information. That is, the end tag is stored as the trailing tag information in association with the word immediately before it.

Taking English (6) as an example, as shown in Table 2, the start tag <A HREF="/misc/index.html"> is attributed to the word "Here" immediately after it as the prefix tag information. Be converted. In addition, the end tag </A> is attributed to the word "Here" immediately before it as the postfix tag information.

[0109]

[Table 2]

When the word attribute processing is completed, syntax / semantic analysis is performed (step 25). Then, the conversion process is performed based on the result of the syntax / semantic analysis (step 26). That is, a translation is attached to each word in Table 2.
Thereafter, a syntax generation process is performed based on the syntax / semantic analysis result (step 27). That is, according to the Japanese grammar, the translated words obtained in step 26 are assembled to form a Japanese sentence. In the above example, the Japanese sentence "Japanese page is here." Is obtained.

Thereafter, a tag attribute restoration process is performed (step 28). That is, first, based on the word attribute processing result of step 24, the tag that modifies the word is restored. In the above example sentence, the tag that modifies the word is restored, so that "Japanese page is <A HREF =" / misc / ind
ex.html "> here </A>." is obtained.

Then, based on the result of the sentence attribute processing of step 22, the tag that modifies the entire sentence is restored. In addition,
In this example, the tag that modifies the character is also changed to the tag that modifies the entire sentence by the character attribute processing, so the tag that modifies the character is defined as the tag that modifies the entire sentence in the translated sentence. In the above example, the tags that qualify the entire sentence are restored, so that "<B><FONT SIZE = + 3> Japanese page is <A HREF="/misc/index.html"> here </ A
></B></FONT> ”is obtained.

The translated text thus obtained is sent to the browser (step 29). In the browser, of the sent translations, sentences other than tags are displayed on the monitor. Therefore, the Japanese sentence "Japanese page is here." Is displayed. In this case, the font is the font specified by the tag <B>, the font size is displayed by the font size specified by <FONT SIZE = + 3>, and “here” is the anchor of the hyperlink. It is displayed so that you can see.

The case where it is determined in step 12 that the keyword translation mode is not set will be described. Even when the keyword translation mode is not set, the processing steps 31 to 31 are almost the same as the processing steps 21 to 29 when the keyword translation mode is set.
9 is performed.

However, in the following two cases, the processing when the keyword translation mode is not set is different from the processing when the keyword translation mode is set.

(I) When the sentence targeted for morphological analysis includes a tag indicating hyperlink information (ii) When a tag indicating hyperlink information modifies the entire sentence

In the case of the above (i), the morphological analysis processing (step 33) performed when the keyword translation mode is not set, and the morphological analysis processing (step 33) performed when the keyword translation mode is set. Two
Different from 3).

In the morphological analysis in step 33, since the keyword translation mode is not set, the entire hyperlink information is treated as one word. Therefore, when the sentence targeted for morphological analysis is the English sentence (6), the morphological analysis result for it is as shown in Table 3 below.

Japanese Page is <A HREF="/misc/index.html"> Here </A> (6)

[0120]

[Table 3]

Therefore, in the word attribute processing (step 34) performed after the morphological analysis processing, the tag for the hyperlink is not attributed.

As a result, when the original sentence is the above example sentence (3), the translated sentence output in step 39 is "<B><F".
ONT SIZE = + 3> Japanese page is <A HREF = "/ misc / index.h
tml "> Here </A>. </ B></FONT>”.

In the case of (ii) above, the sentence attribute processing (step 32) performed when the keyword translation mode is not set, and the sentence attribute processing (step 32) performed when the keyword translation mode is set. 22).

For example, assume that the original sentence is the above-mentioned example sentence (4).

<A HREF="/misk/index.html"> Japanese Page is Here </A>. … Example sentence (4)

In such a case, if the keyword translation mode is set, <A HREF="/misk/index.html"> modifies the entire sentence by the sentence attribute processing of step 22 described above. It is identified as the tag at the beginning of the sentence. Also, </A>
Is determined to be the tag at the end of the sentence that modifies the entire sentence. Then, in the example sentence (4), the portion "Japanese Page is Here" excluding the tags that modify the entire sentence is targeted for morphological analysis. The tag at the beginning of the sentence that modifies the entire sentence and the tag at the end of the sentence that modifies the entire sentence are stored in association with the target of morphological analysis.

On the other hand, in the sentence attribute processing when the keyword translation mode is not set, when it is determined that the tag that modifies the entire sentence is the tag for the hyperlink, that tag With respect to, no sentence attribute processing is performed. That is, the entire example sentence (4) is targeted for morphological analysis.

The document conversion system for converting an English document into a Japanese document has been described above, but the present invention can also be applied to a document conversion system for converting a Japanese document into an English document. .

In the document conversion system for converting a Japanese document into an English document, the translation processing unit 22 performs Japanese-English translation. FIG. 10 shows a translation processing procedure when the translation processing unit 22 performs Japanese-English translation. Here, the case where the hyperlink anchor is also translated will be described.

Here, the translation processing procedure will be described by taking the following original sentence (7) as an example. <B><FONT SIZE = ＋ 3> English </ FONT> The English page is <A HREF＝”english.html”> Here </A>. </ B>… Original (7)

In the original sentence (7), the tags <B> and <
/ B> is a tag for displaying the text sandwiched between them in bold font. In the original sentence (7), tags <FONT SIZE = + 3> and </ FONT> are tags for designating the font size of the characters sandwiched between them. In the original sentence (7), the tag <A HREF = "en"
“Glish.html”> and </A> are tags indicating hyperlink information, and the word “here” between the tags indicates that they are anchors (keywords) of hyperlinks.

When the original text to be translated is input (step 41), character attribute processing is performed (step 4).
2). That is, the tag that modifies the character is searched.
When a tag that modifies a character is found, the position of that tag is changed so that the tag modifies the entire sentence.

In the original sentence (7), the tag <FONT SIZE = + 3>
And </ FONT> are tags that modify characters, so
By the character attribute processing, the original sentence is converted into the following Japanese sentence (8).

<B> <FONT SIZE = + 3> English page is <A HREF=”english.html”> Here </A>. </ B> </ FONT>… Japanese (8)

In the character attribute processing, the tag which modifies a character modifies a word including the character,
The position of the tag may be changed.

When the character attribute process is completed, the sentence attribute process is executed for the sentence after the character attribute process is performed (step 43). That is, the tag that modifies the entire sentence is searched. When a tag is present at the beginning of a sentence and a tag paired with the tag does not exist in the middle of the sentence, the tag is determined to be the tag at the beginning of the sentence that modifies the entire sentence. If there is a tag at the end of the sentence, that tag is determined to be the tag at the end of the sentence that modifies the entire sentence.

In the above Japanese sentence (8), start tags <B> and <FON
It is determined that T SIZE = + 3> is the tag at the beginning of the sentence that modifies the entire sentence. Also, the end tags </ B> and </ FONT> are discriminated as tags at the end of the sentence that modify the entire sentence. Then, of the Japanese sentence (8), the next Japanese sentence (9) excluding the tag that modifies the entire sentence is targeted for morphological analysis.

The English page is <A HREF =”english.html”> here </A>. ... Japanese sentence (9) The tag at the beginning of the sentence that modifies the entire sentence and the tag at the end of the sentence that modifies the entire sentence are stored in association with the target of morphological analysis.

Japanese sentence (9) determined as a target of morphological analysis
Is subjected to morphological analysis processing (step 4).
4). Here, the keywords are also translated.

The morphological analysis result for Japanese sentence (9) is shown in Table 4 below.

[0141]

[Table 4]

After the morphological analysis processing, word attribute processing is performed (step 45). In the word attribute processing, the start tag is attributed to the word immediately after it as the prefix tag information among the pair of tags that modify the word. That is, the start tag is stored as the leading tag information in association with the word immediately following it. Also, the end tag is attributed to the word immediately before it as the post tag information. That is, the end tag is stored as the trailing tag information in association with the word immediately before it.

Taking the Japanese sentence (9) as an example, as shown in Table 5, the start tag <A HREF="english.html"> is attributed to the word "here" immediately after it as the prefix tag information. . In addition, the end tag </A> is attributed to the word "is" immediately before it as the suffix tag information.

[0144]

[Table 5]

When the word attribute processing is completed, syntax / semantic analysis is performed (step 46). Then, conversion processing is performed based on the result of the syntax / semantic analysis (step 47). That is, a translation is attached to each word in Table 5.
Thereafter, a syntax generation process is performed based on the syntax / semantic analysis result (step 48). That is, according to the English grammar, the translated words obtained in step 47 are assembled to form an English sentence. In the above example, "English pa
ge is here. ”

Thereafter, a tag attribute restoration process is performed (step 49). That is, first, based on the result of the word attribute processing in step 45, the tag that modifies the word is restored. In the example sentence above, the tag that modifies the word is restored, so that “English pageis <A HREF =” englis
You will get the translation h.html "> here </A>.".

Then, based on the result of the sentence attribute processing in step 43, the tag that modifies the entire sentence is restored. In addition,
In this example, the tag that modifies the character is also changed to the tag that modifies the entire sentence by the character attribute processing, so the tag that modifies the character is defined as the tag that modifies the entire sentence in the translated sentence. In the above example, the tag that qualifies the entire sentence is restored, so that “<B><FONT SIZE = + 3> En
glish page is <A HREF =”english.html"> here </A>
. </ B></FONT> ”is obtained.

The translated text thus obtained is sent to the browser (step 50). In the browser, of the sent translations, sentences other than tags are displayed on the monitor. Therefore, the English sentence "English page is here." Is displayed. At this time, the font is displayed in the font specified by the tag <B>, the font size is displayed in the size specified by the tag <FONT SIZE = + 3>, and “here” is the anchor of the hyperlink. It is displayed so that you can see it.

The translation process of FIG. 9 can also be applied when automatically translating an HTML document created in English into a Japanese HTML document. FIG.
The translation process can also be applied when automatically translating an HTML document created in Japanese into an English HTML document.

[0150]

According to the method of translating a hypertext document according to the present invention, a translated sentence obtained by translating a hyperlink anchor in a hypertext document or a hyperlink anchor in a hypertext document is translated according to a user's selection. You will be able to get the translated text left without any changes.

According to the HTML document translation device of the present invention, an HTML document whose text is described in the first language is
The text can be automatically converted into an HTML document described in the second language. Therefore, it becomes possible to automatically convert the HTML document created in the language of the own country into the HTML document of the language of another country.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a document conversion system.

FIG. 2 is a block diagram showing a configuration of a supervisor.

FIG. 3 is a sequence diagram showing a flow of processing according to the first embodiment of the present invention.

FIG. 4 is a sequence diagram showing a flow of processing according to the second embodiment of the present invention.

FIG. 5 is a sequence diagram showing the flow of processing according to the second embodiment of the present invention.

FIG. 6 is a sequence diagram showing the flow of processing according to the third embodiment of the present invention.

FIG. 7 is a sequence diagram showing the flow of processing according to the third embodiment of the present invention.

FIG. 8 is a flowchart showing a translation processing procedure by a translation processing unit.

FIG. 9 is a flowchart showing a translation processing procedure when an English HTML document is converted into a Japanese HTML document.

FIG. 10 is a flowchart showing a translation processing procedure when converting a Japanese HTML document into an English HTML document.

FIG. 11 is a diagram showing hyperlink information added with a translation instruction.

FIG. 12 is a diagram showing an example of hyperlink information.

[Explanation of symbols]

 1 Browser 2 Supervisor 3 Internet 4 WWW Server 21 Data Monitoring Unit 22 Translation Processing Unit 23 Cache Memory

Claims (4)

[Claims] 1. In a method of translating a hypertext document, a user selects an anchor translation mode in which an anchor of a hyperlink is translated or an anchor non-translation mode in which the original text remains unchanged, and the anchor translation mode is selected. If you are translating a hypertext document, translate the hyperlink's anchor, and if anchor non-translation mode is selected, translate the hypertext document's anchor. A method of translating a hypertext type document, which is characterized by keeping the original text without translating it. 2. The hyperlink information comprises a start tag indicating hyperlink information, an end tag indicating hyperlink information, and a hyperlink anchor sandwiched between the start tag and the end tag. When the anchor translation mode is selected, the start tag and the end tag are recognized as one word, the start tag and the end tag are stored in association with the anchor, and only the anchor is translated. After that, by adding the above start tag and the above end tag to the translated word of the anchor and recognizing the entire hyperlink information as one word when the anchor non-translation mode is selected, the hyperlink information is The method for translating a hypertext document according to claim 1, wherein the original text is not translated and remains as the original text. 3. An HTML document translation device for converting an HTML document, the text of which is described in a first language, into an HTML document, the text of which is described in a second language. A first tag for modifying an entire sentence from an original sentence. And a first tag that modifies the entire sentence is present, the sentence in which the first tag that modifies the entire sentence is removed from the original sentence is the target sentence of morphological analysis, and the target sentence of the morphological analysis is First means for associating and storing tags for modifying the entire sentence; second means, second means for performing morphological analysis including word division and grammatical analysis on the target sentence of the morphological analysis obtained by the first means On the basis of the morphological analysis result according to, the second sentence that modifies one or a plurality of words is searched from the target sentence of the morphological analysis, and if there is a second tag that modifies one or a plurality of words, The second tag A third means for storing a start tag of the constituent start tags and end tags in association with a first word immediately after that, and storing an end tag in association with a second word immediately before it, the subject of morphological analysis Fourth means for performing syntactic analysis and semantic analysis on the sentence from which the second tag has been removed, and the second tag has been removed from the target sentence of the morphological analysis based on the analysis result by the fourth means. Generated by the fifth means and the fifth means for allocating a translated word of the second language to each word constituting the sentence and assembling each translated word to form a sentence of the second language to generate a translated sentence of the second language. In the translated sentence, a start tag stored in association with the first word is added to the front side of the translated word corresponding to the first word, and the second tag is added.
The sixth means for adding the end tag stored in association with the second word to the rear side of the translated word corresponding to the word, and the sentence stored by the first means in the sentence obtained by the sixth means. An apparatus for translating an HTML document, comprising: a seventh means for adding a first tag for modifying the whole. 4. A tag for modifying only one character from an original sentence is searched for in an HTML document translation device for converting an HTML document whose text is written in a first language into an HTML document whose text is written in a second language. However, if there is a first tag that modifies only one character, the first tag modifies either one of the entire sentence and the word that includes the character that the tag modifies. The first means for changing the position of, and the original sentence after the processing by the first means is searched for a second tag that modifies the entire sentence, and if a second tag that modifies the entire sentence exists, By the second means and the second means, the sentence in which the second tag that modifies the entire sentence is removed from the original sentence is set as the target sentence of morphological analysis, and the target sentence of the morphological analysis is associated with the tag that modifies the entire sentence The obtained pair of morphological analysis Third means for performing morpheme analysis including word division and grammatical analysis on the quadrangle, and modifying one or more words from the sentence for morpheme analysis based on the morpheme analysis result by the third means If there is a third tag that searches for tags and modifies one or more words, associates the start tag of the start tag and end tag that form the third tag with the first word immediately following it. A fourth means for storing the end tag in association with the second word immediately before it, and for performing a syntactic analysis and a semantic analysis on the sentence obtained by removing the third tag from the target sentence of the morphological analysis. 5 means, based on the analysis result by the 5th means, the translated word of the second language is assigned to each word constituting the sentence from which the second tag is removed from the target sentence of the morphological analysis, and each translated word is assigned to the second language. Construct a sentence A sixth means for assembling as described above to generate a translated sentence in a second language, and a translated sentence generated by the sixth means, in front of the translated word corresponding to the first word, in the first word, In addition to adding a start tag stored in association with the second tag,
The seventh means for adding the end tag stored in association with the second word to the rear side of the translated word corresponding to the word, and the sentence stored by the second means in the sentence obtained by the seventh means. An HTML document translation device comprising: an eighth means for adding a second tag for modifying the whole.