Dictionary editing device, dictionary editing method, and computer-readable storage medium

Abstract

Embodiments of the present invention relate to a dictionary editing device, a dictionary editing method, and a program. A technology capable of reducing the confirmation cost of word extraction results and pronunciation assignment results is provided. A dictionary editing device of one embodiment comprises a word extraction unit, a pronunciation assignment unit, and a correction candidate determination unit. The word extraction unit extracts words from text data. The pronunciation assignment unit assigns pronunciations to the words. The correction candidate determination unit determines a correction candidate when a correction is made to word information including the extracted word and the assigned pronunciation, the correction candidate being a word or pronunciation to be corrected in association with the correction.

Description

[0001] This application is based on Japanese Patent Application No. 2020-184918 (filing date: 11 / 5 / 2020) and claims priority from the aforementioned application, the entire contents of which are incorporated herein by reference. Technical Field

[0002] Embodiments of the present invention relate to a dictionary editing device, a dictionary editing method, and a program. Background Art

[0003] In order to improve the accuracy of voice recognition, it is important to pre-register professional terms that are often spoken in actual voice recognition scenarios and unknown words to the voice recognition engine in a dictionary that the voice recognition engine refers to. However, it is difficult to manually list professional terms and unknown words and assign pronunciations to them.

[0004] In contrast, as long as the text data associated with the scene using voice recognition is prepared to be read in (for example, if you want to recognize the voice of a university class, it is its handout material), professional terms and unknown words are automatically extracted from the text data, and the function of automatically giving pronunciation to the extracted professional terms and unknown words is provided, it will be easy to register professional terms and unknown words in the dictionary. However, the professional terms and unknown words automatically extracted and the pronunciations automatically given may be wrong. Therefore, it is necessary to manually perform the final confirmation of the professional terms and unknown words automatically extracted and the pronunciations automatically given. When the number of words automatically extracted is large, it is difficult to confirm all the automatically extracted words and the pronunciations automatically given. Summary of the invention

[0005] As described above, it takes a cost to manually confirm whether the technical terms and unknown words automatically extracted are correct, and whether the pronunciations automatically assigned to these words are correct.

[0006] The problem to be solved by the present invention is to provide a technology capable of reducing the confirmation cost of word extraction results and pronunciation assignment results.

[0007] A dictionary editing device according to one embodiment includes a word extraction unit, a pronunciation assignment unit, and a correction candidate determination unit. The word extraction unit extracts a word from text data. The pronunciation assignment unit assigns a pronunciation to the word. The correction candidate determination unit determines a correction candidate when a correction is made to word information including the extracted word and the assigned pronunciation, the correction candidate being a word or pronunciation to be corrected in association with the correction.

[0008] According to the dictionary editing device having the above configuration, it is possible to reduce the cost of confirming word extraction results and pronunciation assignment results. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a block diagram showing a dictionary editing device according to an embodiment.

[0010] Figure 2 It is shown Figure 1 A flowchart of an example of the operation of a dictionary editing device.

[0011] Figure 3 It is shown Figure 1 The display unit shown is a diagram of an example of a screen displaying a word list.

[0012] Figure 4 It is a diagram for explaining the emphasis display method of the embodiment.

[0013] Figure 5 It is a diagram for explaining the emphasis display method of the embodiment.

[0014] Figure 6 It is a diagram for explaining the emphasis display method of the embodiment.

[0015] Figure 7 It is a diagram for explaining the emphasis display method of the embodiment.

[0016] Figure 8 It is a block diagram showing the hardware configuration of the information processing device according to the embodiment.

[0017] Explanation of symbols

[0018] 100: dictionary editing device; 101: word extraction unit; 102: pronunciation assignment unit; 103: word list; 104: correction acceptance unit; 105: correction candidate determination unit; 106: display unit; 801: CPU; 802: RAM; 803: program memory; 804: storage device; 805: display device; 806: input device; 807: communication device; 808: bus. DETAILED DESCRIPTION

[0019] The following describes an embodiment with reference to the accompanying drawings. The embodiment provides a technique for supporting a user's operation when adding a word to a dictionary used for purposes such as voice recognition. In the following description, a dictionary used for voice recognition (hereinafter referred to as a voice recognition dictionary) is assumed. The voice recognition dictionary can be information that associates a word, a pronunciation of the word, and a phoneme corresponding to the pronunciation.

[0020] Figure 1 The following schematically shows a dictionary editing device 100 according to one embodiment. Figure 1 As shown, the dictionary editing device 100 includes a word extraction unit 101 , a pronunciation assignment unit 102 , a word list 103 , a correction accepting unit 104 , a correction candidate specifying unit 105 , and a display unit 106 .

[0021] The word extraction unit 101 receives text data, extracts words as candidates to be added to the voice recognition dictionary from the text data, and sends the extracted words to the pronunciation assignment unit 102. The text data is, for example, text data associated with a scene that can be recognized by voice. A word can be composed of one or more morphemes. The word extraction unit 101 performs morpheme analysis on the text data, and extracts words as candidates to be added to the voice recognition dictionary based on the results of the morpheme analysis. The words output from the word extraction unit 101 can be words with substantial meanings such as nouns, verbs, adjectives, and adverbs. The words output from the word extraction unit 101 can include compound words (for example, compound nouns formed by connecting multiple nouns).

[0022] Specialized terms and / or unknown words may be candidates to be added to the voice recognition dictionary. Unknown words are words that do not exist in the voice recognition dictionary. In the present embodiment, the word extraction unit 101 extracts specialized terms and unknown words from text data. As a method for extracting specialized terms, for example, the following method can be used: a method of performing morphological analysis on text data to obtain multiple words, and extracting words with a high frequency of occurrence (for example, exceeding a threshold) in the text data as specialized terms; a method of performing morphological analysis on text data to obtain multiple words, and extracting words that are not easily present in text data in a field different from the field of the text data received by the word extraction unit 101 as specialized terms. As a method for extracting unknown words, for example, the following method can be used: a method of performing morphological analysis on text data to obtain multiple words, and extracting words that are not included in the voice recognition dictionary. These methods may be used in combination. For example, the word extraction unit 101 may extract specialized terms from text data, and extract words that are not included in the voice recognition dictionary from the extracted specialized terms. In addition, existing methods other than the above methods may also be used.

[0023] The pronunciation assigning unit 102 assigns pronunciations to the words extracted by the word extracting unit 101 , and registers word information including the extracted words and the pronunciations assigned to the words in the word list 103 .

[0024] As a method of assigning pronunciation to a word, for example, the following methods can be used: a method of using a word dictionary with pronunciation, a method of using a large amount of data to pre-learn the pronunciation of each character, and a statistical method of assigning pronunciation to a word using the learning result. In the method of using a word dictionary with pronunciation, when a word has been registered in the word dictionary, the word is assigned a pronunciation associated with the word, and when the word is a combination of multiple words registered in the word dictionary, the word is assigned a pronunciation obtained by connecting the pronunciations associated with these words while considering continuous voicing, etc. In addition, existing methods other than the above methods can also be used.

[0025] The word list 103 can store a plurality of word-pronunciation pairs. Each word-pronunciation pair is a pair of a word and a pronunciation.

[0026] The correction accepting unit 104 accepts corrections to the word information registered in the word list 103 from the user. The types of corrections include, for example, deletion of a word, partial deletion of a word, addition of a character to a word, addition of a word, revision of pronunciation, etc. Sometimes, partial deletion of a word and addition of a character to a word are collectively referred to as word revision. After making corrections to the word information registered in the word list 103, the correction accepting unit 104 updates the word list 103 according to the contents of the corrections.

[0027] When a correction is made to the word information registered in the word list 103, the correction candidate determination unit 105 determines a correction candidate that is a word or pronunciation to be corrected in association with the correction, and notifies the display unit 106 of the determined correction candidate. A method of determining the correction candidate will be described later.

[0028] Display unit 106 displays word list 103. Specifically, display unit 106 displays word information registered in word list 103. Furthermore, display unit 106 highlights the words or pronunciations determined as correction candidates by correction candidate determination unit 105 on the screen. The method of highlighting correction candidates will be described later.

[0029] In the word list 103, word pronunciation pairs can be associated with editing information. The editing information includes, for example, a candidate flag. The candidate flag is a flag indicating whether the word pronunciation pair is a candidate to be added to the voice recognition dictionary. For example, the candidate flag "0" indicates that the word pronunciation pair is a candidate to be added to the voice recognition dictionary, and the candidate flag "1" indicates that the word pronunciation pair is not a candidate to be added to the voice recognition dictionary. For example, when the user performs an operation of deleting a word, the candidate flag of the word is changed from "0" to "1". The dictionary editing device 100 outputs the word pronunciation pairing with the candidate flag of "0" in order to add it to the voice recognition dictionary at the time point when the correction work performed by the user is completed. In one example, the dictionary editing device 100 can register the word pronunciation pairing to the voice recognition dictionary. In another example, the dictionary editing device 100 can send the word pronunciation pairing to other devices (not shown) that register the word pronunciation pairing to the voice recognition dictionary.

[0030] Next, the operation of the dictionary editing device 100 will be described.

[0031] Figure 2An operation example of the dictionary editing device 100 is schematically shown. The word extraction unit 101 receives text data input by the user and extracts technical terms and unknown words from the text data ( Figure 2 step S201). The pronunciation assignment unit 102 assigns pronunciations to each of the words extracted as technical terms or unknown words (step S202). The extracted words and the assigned pronunciations are registered in the word list 103 in association with each other.

[0032] The display unit 106 displays the words and pronunciations registered in the word list 103 (step S203). The dictionary editing device 100 waits for the user to confirm the displayed words and pronunciations and correct a certain word or pronunciation (step S204).

[0033] When a correction is made by the user, the correction candidate determination unit 105 determines a correction candidate (step S205), which is a word or pronunciation that should be corrected in association with the correction. The display unit 106 highlights the word or pronunciation determined as the correction candidate (step S206). It is generally considered that the user makes further corrections through this highlighting display, so the process returns to step S204 where the dictionary editing device 100 waits for corrections made by the user. In the case of further corrections, the processes such as determining other correction candidates and highlighting them are similarly repeated.

[0034] In addition, in Figure 2 the example shown, pronunciation assignment is performed after extracting words, but word extraction can also be performed after pronunciation assignment. Extracting words from text data and assigning pronunciations to the words can be any of the processes of extracting words from text data and assigning pronunciations to the extracted words, or assigning pronunciations to text data and extracting words with pronunciations from the text data.

[0035] Next, a method for specifying correction candidates and a highlighting display method will be described. Here, the word extraction unit 101 is repeatedly referred to for extracting words "専門用", "語抽出", "Toshiba", "Toshiba Corporation" from text such as "...専門用語抽出...Toshiba...Toshiba Corporation...", and the pronunciation assignment unit 102 assigns pronunciations "せんもんよう", "ごちゅうしゅつ", "としば", "としばこーぽれーしょん" to these words (hereinafter, referred to as the reference example).

[0036] The dictionary editing device 100 provides a user interface for the user to make corrections. The display unit 106 displays word information registered in the word list 103 on the screen of the user interface. In the reference example, as Figure 3As shown, the display unit 106 displays a list of the following four word-pronunciation pairs: the pair of the word "専門用" (special-purpose) and the pronunciation "せんもんよう", the pair of the word "語抽出" (phrase extraction) and the pronunciation "ごちゅうしゅつ", the pair of the word "Toshiba" and the pronunciation "としば", and the pair of the word "Toshiba Corporation" and the pronunciation "としばこーぽれーしょん".

[0037] In the case of a correction of deleting a word or a part of a word, the correction candidate determination unit 105 determines, as correction candidates, words adjacent to the word registered in the word list 103 in the text data. It is generally considered that a correction of deleting a word or a part of a word is due to a morpheme analysis error, and it is generally considered that words adjacent to the word in the text data are also obtained due to a morpheme analysis error.

[0038] In the reference example, due to a morpheme analysis error, the phrase "専門用語抽出" (special-purpose term extraction) is divided into "専門用" (special-purpose) and "語抽出" (phrase extraction). In the case of a correction of deleting "専門用" or deleting "用" which is a part of "専門用", the correction candidate determination unit 105 determines the word "語抽出" adjacent to the corrected word "専門用" in the text data as a correction candidate. Suppose the word "語抽出" is determined as a correction candidate for deletion or for revision to "専門用語抽出" or "用語抽出". Here, whether the word "語抽出" becomes a deletion candidate or a revision candidate can depend on whether the revised word (in this example, "専門用語抽出" or "用語抽出") is worthy of being extracted as a technical term or an unknown word, whether the revised word is included in the word list 103, etc.

[0039] In the case of a correction of adding a character to a word, the correction candidate determination unit 105 determines, as correction candidates, partial words of the word obtained by the correction (a part of the word obtained by the correction). For example, in the case where a word matching the partial word exists in the word list 103, the correction candidate determination unit 105 may determine that word as a correction candidate (specifically, a deletion candidate), and in the case where a word matching the partial word does not exist in the word list 103, it determines the partial word as a correction candidate (specifically, an addition candidate).

[0040] In the reference example, when a correction is made to append the characters "語抽出" to the word "専門用", the word obtained by the correction can be "専門用語抽出", and partial word examples can be "語抽出", "抽出", "用語", "用語抽出", "専門用語", etc. "語抽出" exists in the word list 103, and the correction candidate determination unit 105 can determine the word "語抽出" as a deletion candidate. "抽出" does not exist in the word list 103, and the correction candidate determination unit 105 can determine the word "抽出" as an addition candidate. In this case, the correction candidate determination unit 105 can use the pronunciation assignment unit 102 to assign a pronunciation to the word "抽出" as an addition candidate, register the word "抽出" and the assigned pronunciation in the word list 103, and the display unit 106 performs an emphasized display on the word "抽出".

[0041] Alternatively, when a correction is made to append characters to a word, the correction candidate determination unit 105 can also determine, as correction candidates (specifically deletion candidates), words in the word list 103 that are adjacent to the word in the text data. For example, in the reference example, when a correction is made to append the character "語" to the word "専門用", among the words registered in the word list 103, the word "語抽出" that is adjacent to "専門用" in the text data is determined as a deletion candidate.

[0042] Alternatively, when a correction is made to append characters to a word, the correction candidate determination unit 105 can also determine, as correction candidates (specifically addition candidates), words that are adjacent to the word obtained by the correction in the text data and do not exist in the word list 103. For example, in the reference example, when a correction is made to append the character "語" to the word "専門用", the word "抽出" that is adjacent to "専門用語" in the text data is determined as an addition candidate.

[0043] In the case of correcting a newly added word, the correction candidate determination unit 105 can determine correction candidates in the same manner as the method described for the case of correcting a word by adding characters. Specifically, the correction candidate determination unit 105 determines a partial word (a part of the added word) of the added word as a correction candidate. For example, when a word that matches the partial word exists in the word list 103, the correction candidate determination unit 105 can determine that word as a correction candidate (specifically, a deletion candidate), and when a word that matches the partial word does not exist in the word list 103, it determines the partial word as a correction candidate (specifically, an addition candidate). For example, in the case of correcting by adding the word "専門用語抽出" (technical term extraction), the correction candidate determination unit 105 can determine the words "専門用" and "語抽出" registered in the word list 103 as deletion candidates. Additionally, for example, in the case of correcting by adding the word "専門用語" (technical term), the correction candidate determination unit 105 can determine the word "用語" that does not exist in the word list 103 as an addition candidate.

[0044] Alternatively, in the case of correcting a newly added word, the correction candidate determination unit 105 can also determine, as correction candidates (specifically, deletion candidates), words registered in the word list 103 that are adjacent to the added word in the text data. Alternatively, in the case of correcting a newly added word, the correction candidate determination unit 105 can also determine, as correction candidates (specifically, addition candidates), words that are adjacent to the added word in the text data and do not exist in the word list 103.

[0045] The correction candidate determination unit 105 can adjust the word extraction method of the word extraction unit 101 according to the content of the correction made by the user, and determine correction candidates based on the result of word extraction using the adjusted word extraction method. For example, in the case of a correction of revising or deleting a word, the correction candidate determination unit 105 adjusts the word extraction method of the word extraction unit 101 in a way that avoids extracting the corrected word from the text data or extracts the word obtained by the revision from the text data. For example, in the case of a correction of adding a word, the correction candidate determination unit 105 adjusts the word extraction method of the word extraction unit 101 to extract the added word from the text data. For example, when the word extraction method of the word extraction unit 101 is a method of extracting words based on certain thresholds, a method of changing the threshold to the score of the word after the correction and determining other words with scores below the threshold as correction candidates (specifically, deletion candidates) can be used. Also, when the added word or the word obtained by the revision is included in the text data and the score of the word is calculated during word extraction, the threshold can be lowered to this score, and other words with scores above the threshold can be determined as correction candidates (specifically, addition candidates).

[0046] In the case of a correction of revising pronunciation, the correction candidate determination unit 105 can determine the pronunciation of a word that is similar in writing to the word after the pronunciation revision as a correction candidate. The writing of the first word being similar to the second word means that the first word contains at least a part of the second word. In the reference example, in the case of a correction of revising the pronunciation of the word "Toshiba" from "としば" to "とうしば", the correction candidate determination unit 105 determines the pronunciation "としばこーぽれーしょん" of the word "Toshiba Corporation" that contains the word "Toshiba" as a correction candidate. Also, in the reference example, in the case of a correction of revising the pronunciation of the word "Toshiba Corporation" from "としばこーぽれーしょん" to "とうしばこーぽれーしょん", the correction candidate determination unit 105 determines the pronunciation "としば" of the word "Toshiba" that contains a part of the word "ToshibaCorporation" as a correction candidate.

[0047] The display unit 106 highlights the correction candidates on the screen that displays the word information registered in the word list 103. As an example, the display unit 106 changes the background color of the field (sometimes also referred to as a cell, box, or text box) that stores the word or pronunciation determined as a correction candidate by the correction candidate determination unit 105. In the reference example, when the user deletes the word "専門用" and the correction candidate determination unit 105 determines the word "語抽出" as a deletion candidate, as Figure 4 As shown, the display unit 106 changes the background color of the field of the word "語抽出" to red, for example.

[0048] Furthermore, when the user revises the pronunciation of the word "Toshiba" to "とうしば" and the correction candidate determination unit 105 determines the pronunciation "としばこーぽれーしょん" as a correction candidate, as Figure 4 shown, the display unit 106 changes the background color of the field of the pronunciation "としばこーぽれーしょん" to yellow, for example.

[0049] The display unit 106 can perform color classification according to the type of correction candidate. For example, deletion candidates are displayed in red, revision candidates are displayed in yellow, and addition candidates are displayed in green. In addition to highlighting the words or pronunciations determined as correction candidates, the display unit 106 can also highlight the corresponding pronunciations or words. In the Figure 4 example shown, the background color of the field of the pronunciation "ごちゅうしゅつ" of the word "語抽出" determined as a deletion candidate is changed to the same color as the background color of the field of the word "語抽出".

[0050] Furthermore, the display unit 106 can also highlight the words or pronunciations that have been corrected. For example, the display unit 106 can change the background color of the field of the words or pronunciations that have been corrected to a color different from the color of the correction candidate. It can be that the correction candidate determination unit 105 determines the type of correction made by the user, and the display unit 106 changes the highlighting method (such as color classification) according to the determination result. In addition to highlighting the words or pronunciations that have been corrected, the display unit 106 can also highlight the corresponding pronunciations or words.

[0051] When the user makes a correction to delete the word "専門用", as Figure 4 shown, the display unit 106 changes the background color of the fields of the deleted word "専門用" and its pronunciation "せんもんよう" to gray, for example. When the user makes a correction to revise the pronunciation of the word "Toshiba" to "とうしば", the display unit 106 displays the revised pronunciation "とうしば" and changes the background color of its field to light blue, for example.

[0052] When the user adds the word "専門用語抽出" and the correction candidate determination unit 105 determines the words "専門用" and "語抽出" as deletion candidates, as Figure 5As shown, the display unit 106 changes the background color of the fields of the word "専門用" (special use) and its pronunciation "せんもんよう", and the word "語抽出" (phrase extraction) and its pronunciation "ごちゅうしゅつ" to red, for example, and changes the background color of the field of the word "専門用語抽出" (special term extraction) and its pronunciation "せんもんようごちゅうしゅつ" to yellowish green, for example.

[0053] The highlighting method is not limited to changing the background color of the field. For example, the highlighting method can also be thickening the frame of the field, changing the color of the frame of the field, enlarging the frame of the field, changing the color of the characters within the field, changing the size of the characters within the field, or changing the font of the characters within the field. The change of the font includes, for example, changing the font, making the font bold, italic, underlined, etc.

[0054] The display unit 106 can perform highlighting using any one of the above highlighting methods, or can perform highlighting by combining two or more of the above highlighting methods. In other words, the display unit 106 can change at least one of the background color of the field storing the correction candidate, the size of the frame of the field, the color of the frame of the field, the color of the characters within the field, the size of the characters within the field, and the font of the characters within the field. The display unit 106 can also use a highlighting method different from the above highlighting methods.

[0055] When highlighting the correction candidate, the display unit 106 can also be displayed in a form that associates the word or pronunciation determined as the correction candidate with the corrected word or pronunciation in response to the correction made by the user. In one example, the display unit 106 can move the word or pronunciation determined as the correction candidate directly below the corrected word or pronunciation. In another example, the display unit 106 can also Figure 6 as shown, connect the corrected word or pronunciation with the correction candidate with a line, etc., to clearly show that the correction candidate is linked to the corrected word or pronunciation. In Figure 6 the example shown, since the user deleted the word "専門用" and the word "語抽出" was determined as the correction candidate, the word "専門用" and the word "語抽出" are connected with a line. Furthermore, since the user revised the pronunciation "としば" and the pronunciation "としばこーぽれーしょん" was determined as the correction candidate, the word "Toshiba" corresponding to the pronunciation "としば" and the word "Toshiba Corporation" corresponding to the pronunciation "としばこーぽれーしょん" are connected with a line. In yet another example, the display unit 106 can also rank the correction candidate at the top of the list.

[0056] The correction candidate determination unit 105 can generate correction schemes for the words or pronunciations determined as the correction candidates, and the display unit 106 can display the correction schemes generated by the correction candidate determination unit 105 in addition to emphasizing the words or pronunciations determined as the correction candidates. The display unit 106 can emphasize the correction schemes. In this case, the correction candidate determination unit 105 sends information indicating the correction schemes to the display unit 106. For example, in a case where the pronunciation of the word "Toshiba" is revised from "としば" to "とうしば", the correction candidate determination unit 105 determines the pronunciation "としばこーぽれーしょん" as the correction candidate, and generates a correction scheme "とうしばこーぽれーしょん" based on the content of the revision. Figure 7 As shown, the display unit 106 may also display the correction plan side by side with the original pronunciation. In addition, when the highlighted field is clicked, the correction plan may be displayed in the form of a drop-down list. In addition, when the cursor is aligned with the highlighted field, the correction plan may be displayed using a pop-up screen. Conversely, when the correction plan is highlighted and the cursor is aligned with the field where the correction plan is displayed, the original pronunciation may be displayed using a pop-up screen.

[0057] As described above, when the dictionary editing device 100 extracts words from text data, assigns pronunciations to the extracted words, and performs corrections on word information including the extracted words and the assigned pronunciations, it determines the words or pronunciations that should be corrected in association with the corrections, and prompts the user. As a result, after correcting the words or pronunciations, the user can easily find the words or pronunciations that should be confirmed next. As a result, the cost of confirming and correcting the word extraction results and the pronunciation assignment results can be reduced.

[0058] Regarding the dictionary editing device 100 , the above-described processing can be realized by executing a program by a general-purpose circuit such as a CPU (Central Processing Unit).

[0059] Figure 8 FIG. 1 schematically shows an example of the hardware configuration of the dictionary editing device 100. Figure 8 In the example shown, the dictionary editing device 100 is a computer including a CPU 801, a RAM (Random Access Memory) 802, a program memory 803, a storage device 804, a display device 805, an input device 806, a communication device 807, and a bus 808. The CPU 801 exchanges signals with the RAM 802, the program memory 803, the storage device 804, the display device 805, the input device 806, and the communication device 807 via the bus 808.

[0060] CPU801 is an example of a general-purpose circuit. RAM802 is used by CPU801 as a working memory. RAM802 includes volatile memories such as SDRAM (Synchronous Dynamic Random Access Memory). Program memory 803 stores programs such as dictionary editing programs executed by CPU801. Programs include commands that can be executed by a computer. As program memory 803, for example, ROM (Read-Only Memory) is used.

[0061] The CPU 801 expands the program stored in the program memory 803 in the RAM 802, interprets and executes the program. When the dictionary editing program is executed by the CPU 801, the CPU 801 executes the above-mentioned processing related to the dictionary editing device 100. In other words, the CPU 801 functions as the word extraction unit 101, the pronunciation assignment unit 102, the correction acceptance unit 104, the correction candidate determination unit 105 and the display unit 106 according to the dictionary editing program. The word list 103 is implemented by the RAM 802 and / or the storage device 804.

[0062] The program such as the dictionary editing program can be provided to the dictionary editing device 100 in a state where it is stored in a computer-readable storage medium. In this case, for example, the dictionary editing device 100 is provided with a drive for reading data from the storage medium, and obtains the program from the storage medium. Examples of the storage medium include a magnetic disk, an optical disk (CD-ROM, CD-R, DVD-ROM, DVD-R, etc.), a magneto-optical disk (MO, etc.), and a semiconductor memory. Alternatively, the program may be stored in a server on the network, and the dictionary editing device 100 may download the program from the server.

[0063] The storage device 804 stores data. The storage device 804 includes a non-volatile memory such as a hard disk drive (HDD) or a solid state drive (SSD). A portion of the storage device 804 may also be used as a program memory 803.

[0064] The display device 805 may be, for example, a liquid crystal display, an OLED (Organic Light Emitting Diode) display, etc. The display device 805 displays an image generated by the display unit 106 , such as a screen of a user interface for correction.

[0065] The input device 806 is a device that allows the user to input information. The input device 806 includes, for example, a keyboard and a mouse. The input device 806 is used to modify word information.

[0066] The communication device 807 is an interface for communicating with an external device and includes, for example, a wired communication module and / or a wireless communication module.

[0067] In addition, regarding the dictionary editing device 100, at least a part of the above-mentioned processing may be implemented by using a dedicated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0068] It is also possible to adopt a configuration in which a terminal device operated by a user is provided separately from the dictionary editing apparatus 100. In such a configuration, the dictionary editing apparatus 100 uses the communication device 807 to communicate with the terminal device.

[0069] Several embodiments of the present invention have been described, but these embodiments are presented as examples and are not intended to limit the scope of the invention. These new embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the gist of the invention. These embodiments and their variations are included in the scope and gist of the invention, and are included in the invention described in the patent claims and the scope equivalent thereto.

[0070] Furthermore, the above-mentioned embodiments can be summarized into the following technical solutions.

[0071] Technical Solution 1

[0072] A dictionary editing device, comprising:

[0073] A word extraction unit extracts words from text data;

[0074] a pronunciation assigning unit that assigns a pronunciation to the word; and

[0075] The correction candidate determination unit determines, when a correction is made to word information including the extracted word and the assigned pronunciation, a correction candidate that is a word or a pronunciation to be corrected in association with the correction.

[0076] Technical Solution 2

[0077] In the above-mentioned technical solution 1, the correction candidate determination unit determines, as the correction candidate, a word that is adjacent to the word subjected to the correction on the text data among the extracted words.

[0078] Technical Solution 3

[0079] In the above-mentioned technical solution 1, when the correction is the addition of the first word or the deletion or revision of the second word among the extracted words, the correction candidate determination unit adjusts the word extraction method of the word extraction unit in a manner not to extract the second word from the text data, or in a manner to extract the first word from the text data or the third word obtained by the revision, and determines the correction candidate based on the result of the word extraction of the text data by the word extraction unit using the adjusted word extraction method.

[0080] Technical Solution 4

[0081] In the above-mentioned technical solution 1, the correction candidate determination unit determines a word that is similar in writing to the word that has undergone the correction as the correction candidate.

[0082] Technical Solution 5

[0083] In the above-mentioned technical solution 1, the correction candidate determination unit determines, as the correction candidate, the pronunciation of a word that is written similarly to the word corresponding to the pronunciation that has been corrected.

[0084] Technical Solution 6

[0085] In any one of the above-mentioned technical solutions 1 to 5, the dictionary editing device further includes a display unit that highlights and displays the correction candidates.

[0086] Technical Solution 7

[0087] In the above-mentioned technical solution 6, the correction candidate determination unit determines the type of the correction, and the display unit changes the mode of emphasized display according to the determination result of the type.

[0088] Technical Solution 8

[0089] In the above-mentioned technical solution 6 or 7, the correction candidate determination unit generates a correction proposal for the word or pronunciation determined as the correction candidate, and the display unit further displays the correction proposal.

[0090] Technical Solution 9

[0091] In the above-mentioned technical solution 8, the display unit displays the correction proposal in a form associated with the correction candidate.

[0092] Technical Solution 10

[0093] In any one of the above-mentioned technical solutions 6 to 9, the display unit displays the word or pronunciation determined as the correction candidate in a form associated with the word or pronunciation that has undergone the correction.

[0094] Technical Solution 11

[0095] In any one of the above-mentioned technical solutions 6 to 10, the display unit changes at least one of the background color of the field storing the correction candidate, the size of the frame of the field, the color of the frame of the field, the color of the characters in the field, the size of the characters in the field, and the font of the characters in the field.

[0096] Technical Solution 12

[0097] A dictionary editing method, comprising:

[0098] Extract words from text data;

[0099] assigning a pronunciation to the word; and

[0100] When the word information including the extracted word and the assigned pronunciation is corrected, a correction candidate is determined, which is the word or the pronunciation to be corrected in association with the correction.

[0101] Technical Solution 13

[0102] A program that causes a computer to function as:

[0103] Extracting word units from text data;

[0104] a unit for assigning a pronunciation to said word; and

[0105] A means for determining a correction candidate, which is a word or a pronunciation to be corrected in association with the correction, when the word information including the extracted word and the assigned pronunciation is corrected.

Claims

1. A dictionary editing device, comprising: A word extraction unit extracts words from text data; a pronunciation assigning unit that assigns a pronunciation to the word; and a correction candidate determination unit that, when the word information including the extracted word and the assigned pronunciation is corrected, determines a correction candidate that is a word or a pronunciation to be corrected in association with the correction, When the correction is a correction of one of the extracted words, the correction candidate determination unit determines a word among the extracted words that is adjacent to the word subjected to the correction on the text data as the correction candidate.

2. The dictionary editing device according to claim 1, wherein: In a case where the correction is the addition of a first word or the deletion or revision of a second word from the extracted word, the correction candidate determination unit adjusts the word extraction method of the word extraction unit in a manner not to extract the second word from the text data or in a manner to extract the first word or the third word obtained by the revision from the text data, and determines the correction candidate based on the result of the word extraction unit performing word extraction on the text data using the adjusted word extraction method.

3. The dictionary editing device according to claim 1, wherein: The correction candidate determination unit determines a word having a similar writing style to the word to be corrected as the correction candidate.

4. The dictionary editing device according to claim 1, wherein: The correction candidate determination unit determines, as the correction candidate, the pronunciation of a word that is similar to the written word corresponding to the corrected pronunciation.

5. The dictionary editing device according to any one of claims 1 to 4, wherein: The dictionary editing device further includes a display unit configured to highlight and display the correction candidates.

6. The dictionary editing device according to claim 5, wherein: The correction candidate determination unit determines the type of the correction, The display unit changes the manner of highlighting display according to the determination result of the type.

7. The dictionary editing device according to claim 5, wherein: The correction candidate determination unit generates a correction proposal for the word or pronunciation determined as the correction candidate, The display unit further displays the correction proposal.

8. The dictionary editing device according to claim 6, wherein: The correction candidate determination unit generates a correction proposal for the word or pronunciation determined as the correction candidate, The display unit further displays the correction proposal.

9. A dictionary editing method comprising: Extract words from text data; assigning a pronunciation to the word; and When the word information including the extracted word and the assigned pronunciation is corrected, a correction candidate is determined, which is a word or a pronunciation to be corrected in association with the correction. The determining of the correction candidates includes, when the correction is a correction of one of the extracted words, determining a word adjacent to the word subjected to the correction on the text data among the extracted words as the correction candidate.

10. A computer-readable storage medium storing a program for causing a computer to function as the following means: Extracting word units from text data; a unit for assigning a pronunciation to said word; and means for determining a correction candidate, which is a word or a pronunciation to be corrected in association with the correction, when the word information including the extracted word and the assigned pronunciation is corrected, The unit for determining the correction candidate includes a unit for determining, when the correction is a correction of one of the extracted words, a word adjacent to the word subjected to the correction on the text data among the extracted words as the correction candidate.

Images