JP2007087189A - Translation device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inexpensive easy-to-use translation device translating and outputting a necessary part alone. <P>SOLUTION: This translation device is provided with an image input part 12 inputting image data including a translation source document, a translation part 14 performing translation of a character recognition result of the image data, an output part 15 outputting a translation result, and a control part 18 controlling the respective parts. The control part 18 carries out a primary output process, in which only an index part in the translation source document is extracted and an index list showing its translation sentence is outputted, and a secondary output process, in which main text corresponding to the index part selected from the index list by the user is translated and outputted. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a translation apparatus that translates content requested by a user using machine translation.

  With the development of international business and the spread of communication networks, a lot of information has become available in the original language. And the effective use of such information is becoming increasingly important at both the personal and social levels. However, the use of such information involves a difficult task of selecting necessary or interesting information from a lot of information in other languages.

  In such a case, it is conceivable to use machine translation by a computer. In fact, routine machine translation of foreign language information and providing it in a state where it can be used by users is a common practice in many organizations and in open areas such as the Internet. However, the information provided in this way is generally limited to information that has a common interest among many users.

There are already systems for providing translation services by machine translation for individual users. The user can use such a translation service for the purpose of collecting information he / she needs from predetermined foreign language information. However, translating unnecessary information is not only useful, but also increases translation costs, and it takes time to search for necessary information from a large amount of information even after translation.
For example, in Patent Document 1, when a sentence to be translated has the same part as a sentence that has been previously translated, it is possible to avoid unnecessary translation work by using a past translation sentence, thereby reducing the translation cost. Although a technique for reducing is disclosed, even if such a technique is used, it is impossible to avoid translation of information judged to be unnecessary for individual users.
Japanese Patent Laid-Open No. 3-282965

  The present invention has been made in view of the above circumstances, and provides a technique that enables easy and low-cost translation and output of only a portion determined to be necessary from a translation source document. For the purpose.

  In order to solve the above-described problems, the present invention provides an input unit that receives document image data representing a document, and performs layout analysis on the document image data that is input to the input unit. Extracting means for extracting an index part and a text corresponding to the index part; storage means for storing the index part extracted by the extracting means and the text corresponding to the index in association with each other; and The stored index part is translated into a pre-designated target language and the result of the translation is arranged in a list format. While the index list is output, the translated text of the index part arranged in the index list A specifying means for specifying the index portion selected by the user, and an identifier specified by the specifying means. Reading the text that corresponds to the index unit from said storage means, to provide a translation apparatus characterized by and an output means for outputting the translated into the target language.

  According to such a translation apparatus, even an output of an index list in which translated sentences of only the index part in the translation source document are arranged in a list format is selected by the user from among the index parts arranged in the index list. Only the text corresponding to the index part is translated and output. This makes it possible to easily select and translate only the information necessary for the user while reducing the translation work amount and reducing the cost.

In a more preferred aspect, in the index list, a predetermined annotation mark is attached to each translated sentence of the index portion, and the specifying means is configured to check whether the predetermined check mark is attached to the annotation mark. It is determined whether or not the index portion corresponding to the annotation mark is selected by the user when the check mark is added.
In such an aspect, since the selection instruction from the user to the translation apparatus is performed via the annotation mark attached to the translated sentence in the index portion, the operation is easily and reliably performed.

In order to solve the above-described problem, the present invention provides a computer apparatus that performs input analysis on document image data representing a document, and performs layout analysis on the document image data input to the input unit. Extraction means for extracting the index part of the document represented by the image data and the text corresponding to the index part, and storage means for storing the index part extracted by the extraction means and the text corresponding to the index in association with each other The index part stored in the storage means is translated into a pre-specified translation language, and an index list obtained by arranging the translation results in a list format is output, while being arranged in the index list. A specifying means for specifying an index part selected by a user among translated sentences of the index part, Reading the text that corresponds to the index portion specified by the constant unit from the storage unit, provides a program for causing to function as an output means for outputting the translated into the target language.
According to such a program, it becomes possible to give the same function as that of the translation apparatus to a general computer apparatus.

  According to the present invention, it is possible to easily translate and output only a portion determined to be necessary from a translation source document at low cost.

Embodiments of the present invention will be described below with reference to the drawings.
[Constitution]
FIG. 1 is a diagram illustrating a configuration of a translation apparatus 1 according to the present embodiment.
The translation device 1 inputs an input document (hereinafter referred to as a translation source document) described in a language (hereinafter referred to as a translation source language) different from the language used by the user, and then causes the user to select a necessary part of the input document. It is translated into the user's language (hereinafter referred to as the target language) and output. In this embodiment, both the input document and the output document are paper, but the present invention is not limited to this.

  The translation apparatus 1 includes an operation unit 11 that is operated by a user to input an instruction, an image input unit 12 such as an image scanner that reads an image of a paper document and acquires image data representing the image (hereinafter, document image data), A storage unit (RAM) 13 that temporarily stores document image data acquired by the image input unit 12, a translation unit 14 that translates a document represented by the document image data stored in the storage unit 13, and a translation An output unit 15 that outputs processed data, a main storage unit 16 that stores various data tables and programs, and a control unit that controls the operation of each unit, exchanges data, and processes data therebetween 18 Each of these parts includes a hardware part and a software part, and the software part is virtually constructed based on the function.

  The operation unit 11 is provided with input means such as a keyboard and a mouse, and display means such as a liquid crystal display. The main storage unit 16 is a non-volatile memory that can retain data even when power is not supplied from a power source (not shown), and a program that guides operations described later, a source language, and a target language are associated with each other. A dictionary D for translation is stored. The dictionary D corresponds to a plurality of types of languages according to the user's needs, and is selected and used from them.

[Operation]
When power is supplied from the power source to the apparatus, the control unit 18 reads out and executes the program P from the main storage unit 16, thereby allowing each unit of the translation apparatus 1 to be controlled. The translated sentence providing process performed by the control unit 18 operating according to the program P includes two steps. That is, the first step is a step (index list output step) in which index parts such as headings and titles are extracted from the translation source document, and an index list that is a list of the index parts is created and output. Then, in the second step, the body corresponding to the index portion designated by the user is extracted from the translation source document with respect to the index list created in the first step, and is translated into the translation destination language. This is a process (translation sentence output process) of outputting in a simple format.

  In the following, first, the index list creation step will be described with reference to the flowchart of FIG. 2, the translation source document example of FIG. 3, and the index list example of FIG. FIG. 3 illustrates a Japanese document as a translation source document for convenience. In the following, it is assumed that this translation source document is a document in a language other than Japanese (for example, English or Chinese). To explain.

First, the control unit 18 prompts the user to input the target language and, if necessary, for specifying the source language, and the user inputs this via the operation unit 11 (step 1). Next, the user sets a manuscript that is a translation source document (the document shown in FIG. 3 in this operation example) on the manuscript table (not shown) of the image input unit 12, and issues an instruction to start reading via the operation unit 11. (Step 2). The control unit 18 confirms that the document is set on the document table, acquires the document image data of the paper document set on the document table by the image input unit 12, and stores the acquired document image data in the storage unit 13. Is written and stored (step 3).
Next, the control unit 18 performs layout analysis on the document image data written in the storage unit 13 in step 3 (step 4), and divides the document image data into a text unit and graphics and other graphic units. Write to.

Next, the control unit 18 determines from the text portion written in the storage unit 13 in step S4, the title of the document, the index portion and the body corresponding to the index portion, and none of them, for example, a figure or table 5 is divided into titles and descriptions (step 5), the index part and the text corresponding to the index part are associated with each other to generate the management table shown in FIG. 5, and the translation source document is uniquely identified in the management table Identifier (in this operation example, the title of the document) is written in association with the storage unit 13. When generating the management table, the control unit 18 associates the index part with the text corresponding to the index part, and arranges the index part in the order of description of the index part in the translation source document to generate the management table. . This management table is used when the text corresponding to the index part selected by the user is specified in the translated text output process described later. In this embodiment, a case where a title is used as the identifier will be described. However, a unique reference number or unique barcode is assigned to each translation source document, and the reference number or barcode is used as the identifier. Also good.
Of course, when the index part and the text corresponding to the index part are written in the management table, the index part and the text corresponding to the index part are written after being subjected to OCR processing.

  The control unit 18 extracts the title of the document, the index part and the body corresponding to the index part, and the title and description of a figure or table, for example, which do not belong to them, and extracts them from the text part. In this case, the extraction is performed while classifying based on the features. More specifically, the title of the document has the property that it is at the beginning of the document and is written in a larger character size than other sentences. The index part has several features different from the main text in the layout of the translation source document. For example, it exists immediately after the notation of chapter or section, has a larger character than the text, and is a short sentence separated from the text. On the other hand, the titles and explanations of figures and tables have a feature that they are laid out in the vicinity of the figures and tables. Based on these characteristics, the control unit 18 classifies the text part into an index part, a body corresponding to the index part, and a title or description of a figure or a table that does not belong to any of them. Note that the text corresponding to the index part is a sentence that follows the index part, and if there is a next index part that follows the sentence, it is a part up to immediately before the next index part. .

  Next, OCR (Optical Character Recognition) is performed on the title obtained in step 5 (in this operation example, the identifier associated with the management table described above) and the index portion stored in the management table. Text data is generated (step 6), and the text data is written into the storage unit 13. Further, the control unit 18 translates the obtained title and text data of the index part (step 7). In the example of the translation source document in FIG. 3, the title is “XX report”, and the index part is “Preface”, “Company A information”, “Characteristics of Company B product”, “Company C information”, “ "D market sales for the past 3 years", "E company sales table", "Company F information", and "postcard".

  Next, the control unit 18 creates an index list of a predetermined format as shown in FIG. 4 using the title and index part translated in step S7 (step 8). The index list includes at least a translated sentence with a title that is an identifier for identifying the translation source document, and an index serving as an index when the user selects a portion to be translated in the text of the translation source document Parts of translated sentences (each of which is called an item) are arranged in the order of description in the translation source document. When an index list of documents with the same title has already been output, a number such as “XX report 2” may be added.

  Each item in the index list is combined with an annotation mark (in the index list shown in FIG. 4, “◇” is an annotation mark). This annotation mark is used to transmit the user's selection contents to the translation apparatus 1. For example, the user checks (for example, fills) the annotation mark for an item for which the body text is desired to be translated. Thus, the selection of the item is transmitted to the translation apparatus 1. In the output example shown in FIG. 4, two annotation marks are associated with each item, the first annotation mark indicates whether or not translation is necessary, and the second annotation mark indicates that a summary is necessary. No. In this example, the items are “Company A information”, “Product characteristics of Company B”, “Company C information”, “D market sales for the past 3 years”, “Company E sales table”, and “Company F information”. Are listed. In this example, “Foreword” and “Afterword” (and similar indexes such as “Summary”) do not substantially contain new information, and are deleted from the index list.

  In addition to this, the index list includes a part for instructing a translation output method. In the output example shown in FIG. 4, color printing, saving printing (for example, translation for a plurality of pages on one recording material such as 2 up) It is possible to instruct whether or not it is necessary to print the sentence), enlargement printing, and additional writing of the original sentence. Further, when the control unit has a function of interpreting a free description, a description area where additional hope can be described may be provided as illustrated. The contents include inserting a map corresponding to the place name, displaying money amount information about the processing cost, and so on.

  The control unit 18 creates an index list with the above contents, images it, sends the data to the output unit 15, and prints it on a recording material such as printing paper (step 9). This completes the index list output process. The user refers to the item (that is, the translated text of each index part) described in the index list output as described above, determines whether the text corresponding to the item needs to be translated, and is necessary. This instruction is performed by checking annotation marks or writing in the description area. Then, the instruction is input by causing the scanner to read this.

Hereinafter, the translated text output process as the second process will be described with reference to FIG.
When image data corresponding to the index list is input via the image input unit 12 (step 11), the control unit 18 identifies an identifier (in this embodiment) representing document image data corresponding to the index list from the index list. , Title) is extracted (step 12), and the presence or absence of the original text of the corresponding translation source document is confirmed (step 13). Specifically, the control unit 18 checks whether or not a management table corresponding to the identifier is stored in the storage unit 13.

  When the determination result in step 13 is “No” (that is, when there is no corresponding translation source document), the control unit 18 displays that fact and ends the process. If the result is “Yes”, an annotation mark check relating to the necessity of translation is detected (step 14).

Next, the control unit 18 determines whether or not there is a check mark (painting in this embodiment) (step 15), and if there is a check (that is, if the annotation mark is painted), it responds. The text is read from the management table described above, converted into text data by performing OCR processing, and then translated and stored (step 16). Further, the presence / absence of a summary instruction for the item is detected (step 17), and if necessary, a summary is created (step 18).
When the text corresponding to the annotation mark with the check mark is read from the management table, the control unit 18 first determines the order in which the annotation mark with the check mark is written in the index list. The index part corresponding to the annotation mark is specified, and the text corresponding to the index part is read from the management table.

  Thereafter, the control unit 18 creates an image in which the translated sentence and the original sentence chart created in step S16 and the summary if created are laid out (step 19), and whether or not all annotation marks are checked. Is detected (step 20), and if not, the process returns to step 13 to repeat the subsequent steps. When the translated text is created for all of the checked items, the control unit 18 prints the image created in step A19 (step 21), and the translated text output process ends. FIG. 7 shows an example of translation output when “Company A information”, “D market sales for the past three years”, and “Company F information” are selected.

  As described above, the present invention has been described along the embodiment, but the present invention is not limited to this. In this example, a paper document is used as the input document, but digitized data can be used. Such data can be input using a medium such as a magnetic disk or a communication line such as the Internet, and an input unit corresponding to them is used.

  Further, although a paper document is used as the index list output, an appropriate display means such as a display of the operation unit 11 can be used. In this case, the annotation mark may be a check box that can be input with a mouse or the like. Such an index list output and an instruction method for the index list can also be performed via a communication line such as the Internet. Similarly, the translated sentence output may be provided by writing electronic data representing the translated sentence (for example, image data representing the image of the translated sentence or text data of the translated sentence) on a recording medium such as a magnetic disk. In addition, such electronic data may be provided via a communication line such as the Internet.

  In the embodiment described above, a program for causing the control unit 18 to execute a translated sentence providing process characteristic of the translation apparatus according to the present invention is stored in the main storage unit 16 of the translation apparatus 1 in advance. explained. However, it goes without saying that the program may be written and provided on a computer-readable recording medium such as a CD-ROM or a magnetic disk. By installing the program on a general computer device using such a recording medium, the computer device is caused to execute the translated sentence providing process (that is, the computer device is the same as the translation device according to the present invention). Can be added).

It is a block diagram which shows the structure of the translation apparatus 1 which concerns on one Embodiment of this invention. It is a flowchart which shows the flow of the index list output process which the control part 18 of the translation apparatus 1 performs. It is a figure which shows the example of a translation origin document. It is a figure which shows the example of the index list output by the index list output process. It is a figure which shows the example of the management table written in the memory | storage part 13 in the index list output process. It is a flowchart which shows the flow of the translation output process which the control part 18 performs. It is a figure which shows the example of the translation sentence output by the same translation output process.

Explanation of symbols

12 Image input unit 14 Translation unit 15 Output unit 18 Control unit

Claims (3)

Input means for inputting document image data representing a document;
An extraction unit that performs layout analysis on the document image data input to the input unit, and extracts a document index part represented by the document image data and a body corresponding to the index part;
Storage means for storing the index part extracted by the extraction means and the text corresponding to the index in association with each other;
The index part stored in the storage means is translated into a pre-specified translation language, and an index list formed by arranging the translation results in a list format is output, while the indexes arranged in the index list A specifying means for specifying the index part selected by the user among the translations of the part;
An output unit that reads out the text corresponding to the index part specified by the specifying unit from the storage unit, translates it into the target language, and outputs it;
A translation apparatus comprising: In the index list, a predetermined annotation mark is attached to each translated sentence in the index part,
The specifying means is:
It is determined whether or not a predetermined check mark is attached to the annotation mark, and when the check mark is attached, it is specified that the index portion corresponding to the annotation mark is selected by the user. The translation apparatus according to claim 1. Computer equipment,
Input means for inputting document image data representing a document;
An extraction unit that performs layout analysis on the document image data input to the input unit, and extracts a document index part represented by the document image data and a body corresponding to the index part;
Storage means for storing the index part extracted by the extraction means and the text corresponding to the index in association with each other;
The index part stored in the storage means is translated into a pre-specified translation language, and an index list formed by arranging the translation results in a list format is output, while the indexes arranged in the index list A specifying means for specifying the index part selected by the user among the translations of the part;
An output unit that reads out the text corresponding to the index part specified by the specifying unit from the storage unit, translates it into the target language, and outputs it;
A program characterized by functioning as

Images