JP4304146B2 - Dictionary registration device, dictionary registration method, and dictionary registration program - Google Patents

Description

  The present invention relates to a dictionary registration device, a dictionary registration method, and a dictionary registration program for registering words that are not registered in a dictionary.

  In recent years, a kana-kanji conversion system is generally used as a means for inputting Japanese sentences on a personal computer or a mobile phone. Recently, voice recognition systems that can input characters by voice have begun to be used. In these systems, when hiragana input characters are converted into kanji-kana mixed sentences, they are converted into an optimal notation string among word combinations registered in the system dictionary. Therefore, when a word with a notation desired by the user is not registered in the dictionary, the word is not correctly converted, and the notation of the word registered in the dictionary of the system is appropriately converted into an incorrect notation string.

  The dictionary registration device is a device for additionally registering unknown words, which are words that are not registered in the system dictionary, in the dictionary in order to eliminate such inconvenience. As a method of registering unknown words in the dictionary, the user inputs information such as word notation, reading and part of speech from the input screen one by one and registers them in the dictionary, and morphological analysis and extraction of the document specified by the user A method has been developed in which unknown words are registered in a dictionary.

  In the method of extracting and registering an unknown word from a document specified by the user, when a word that is not registered in the system dictionary is extracted, a continuous range of the same character type in the partial character string of the word is identified. A function for estimating a word as an unknown word has been developed (for example, Patent Document 1).

JP-A-2-163874

  However, according to the method of estimating a continuous range having the same character type as one unknown word, a word including a plurality of character types can be extracted as one word, such as a word combining katakana and Arabic numerals. Can not. For this reason, in order to correctly register a word including a plurality of character types as one word, there is a problem that the user needs to check and correct the extracted word.

  The present invention has been made in view of the above, and not only extracts unknown words by separating them with character types, but also extracts at least one word in front of and behind the extracted unknown words as an extracted unknown word. When a word similar to the generated extended unknown word is already registered in the dictionary by combining to generate the extended unknown word, the extended unknown word is also extracted as an unknown word, thereby including a plurality of character types Another object of the present invention is to provide a dictionary registration device, a dictionary registration method, and a dictionary registration program that can be simultaneously registered in a dictionary.

  In order to solve the above-described problems and achieve the object, the present invention includes a dictionary storage means for storing a dictionary that holds words, and morphological analysis of the input document, which is registered in the dictionary from the input document. Morpheme analyzing means for extracting no unknown word, unknown word range extending means for generating an extended unknown word in which at least one of the front and rear words of the unknown word extracted by the morpheme analyzing means is combined with the unknown word, Of the extended unknown words generated by the unknown word range expansion means, a partial match that searches for a registered word registered in the dictionary that matches the notation of the expanded portion of the unknown word. The partial match search based on a search attribute and a character attribute of a portion corresponding to the unknown word and a character attribute of the unknown word in the registered word searched by the partial match search portion A notation similarity determination means for determining a similarity between a notation of a portion corresponding to the unknown word and a notation of the unknown word, and the notation similarity determination means includes the registered word And a dictionary registering means for registering the extended unknown word in the dictionary when it is determined that the notation of the portion corresponding to the unknown word is similar to the notation of the unknown word.

  The present invention also provides a dictionary registration method and a dictionary registration program that can execute the above-described apparatus.

  According to the present invention, when an unknown word is extracted from a document specified by a user and registered in the dictionary, not only the notation of the word estimated as an unknown word by dividing the character type but also the front of the word estimated as an unknown word A word in which at least one of the back words is combined and a similar word is already registered in the dictionary can also be extracted as an unknown word. For this reason, words including a plurality of character types can be registered in the dictionary at the same time, and the accuracy of unknown word extraction can be improved. In addition, there is an effect that it is possible to reduce a user's work burden such as confirmation of the extracted unknown word.

  Exemplary embodiments of a dictionary registration device, a dictionary registration method, and a dictionary registration program according to the present invention will be explained below in detail with reference to the accompanying drawings.

(First embodiment)
The dictionary registration apparatus according to the first embodiment extracts an unknown word from a document specified by a user by morphological analysis, expands the range of the extracted unknown word, and expands the unknown word among the expanded unknown words If there is a word in the dictionary where the character type of the part corresponding to the unknown word is the same as the character type of the unknown word, the expanded unknown word is registered in the dictionary simultaneously with the extracted unknown word. Is.

  FIG. 1 is a block diagram illustrating a configuration of the dictionary registration device 100 according to the first embodiment. As shown in the figure, the dictionary registration apparatus 100 includes an input / output control unit 101, a morpheme analysis unit 102, an unknown word range expansion unit 103, a partial match search unit 104, a notation similarity determination unit 105, a dictionary, A registration unit 106 and a user I / F 110 are provided. Further, the dictionary registration device 100 according to the present embodiment stores a word string buffer 131 and an unknown word buffer 132 in a RAM (Random Access Memory) 130 and analyzes rules in a hard disk drive (HDD). 120 and the dictionary 121 are stored. The HDD corresponds to the dictionary storage means in the present invention.

  The input / output control unit 101 is a processing unit that controls the user I / F 110, and performs input / output instructions to the user I / F and exchange of input / output data between the user I / F 110 and other functional units.

  The morpheme analysis unit 102 divides the input document into words by performing morpheme analysis with reference to the analysis rules 120 and the dictionary 121, and stores the words in the word string buffer 131.

  The unknown word range expansion unit 103 generates an extended unknown word, determines the validity of registration of the extended unknown word in the dictionary in conjunction with the partial match search unit 104 and the notation similarity determination unit 105 described later, and Is determined, the extended unknown word is stored in the unknown word buffer 132 for dictionary registration. Here, the extended unknown word is an unknown word obtained by combining at least one word in front of, behind, or behind the extracted unknown word, that is, a word in front of, behind, or before and after the unknown word extracted by the morphological analysis unit 102. A word that is an extension of.

  The partial match search unit 104 searches the dictionary 121 for a word that matches the expanded part of the unknown word in the partial character string included in the extended unknown word generated by the unknown word range expansion unit 103.

  The notation similarity determination unit 105 determines whether a portion corresponding to the unknown word in the partial character string included in the word searched by the partial match search unit 104 is similar to the unknown word extracted by the morpheme analysis unit 102. Determine.

  The dictionary registration unit 106 includes a notation similarity determination unit 105 when a portion corresponding to an unknown word in the partial character string included in the word searched by the partial match search unit 104 is similar to the unknown word extracted by the morpheme analysis unit 102. The unknown word range expansion unit 103 registers the extended unknown word generated in the dictionary 121 when determined by the above.

  The user I / F 110 is a display device such as a display device and an input device such as a keyboard and a mouse. The user I / F 110 displays a document designation screen and an unknown word confirmation screen and accepts input operations from these screens.

  The analysis rule 120 describes rules necessary for morphological analysis, such as grammatical rules such as the degree of connection between parts of speech and priority selection rules for word selection.

  The dictionary 121 is a dictionary that holds words, and is a dictionary used in a general kana-kanji conversion system, a speech recognition system, or the like. FIG. 2 is an explanatory diagram showing an example of the structure of the dictionary 121. As shown in the figure, the dictionary 121 stores a dictionary number, a word notation, a word reading, and a word part of speech.

  The RAM 130 is a random accessible memory and functions as a storage unit for temporarily storing the word string buffer 131 and the unknown word buffer 132. The word string buffer 131 stores the word string extracted by the morpheme analyzer 102. The unknown word buffer 132 stores the unknown word extracted by the morphological analysis unit 102 and the extended unknown word generated by the unknown word range expansion unit 103.

  FIG. 3 is an explanatory diagram showing an example of the structure of the word string buffer 131. As shown in the figure, the word string buffer 131 stores a word number, a word notation, a word reading, and a word part of speech.

  FIG. 4 is an explanatory diagram showing an example of the structure of the unknown word buffer 132. As shown in the figure, the unknown word buffer 132 includes a notation of an unknown word or extended unknown word, and an analysis result word number that is a word number of a word corresponding to the unknown word or extended unknown word in the word string buffer 131. Is stored. In the case of an extended unknown word, the word numbers of the combined words are arranged and stored in the analysis result word number.

  Next, an unknown word registration process performed by the dictionary registration apparatus 100 according to the first embodiment configured as described above will be described. FIG. 5 is a flowchart showing an overall flow of the unknown word registration process in the first embodiment.

  First, the input / output control unit 101 displays a document designation screen on the user I / F 110 (step S501). 6A and 6B are schematic diagrams illustrating the contents of the document designation screen. As illustrated in FIG. 6A, a reference button 601, a delete button 602, a next button 603, and a cancel button 604 are displayed on the document designation screen. When the reference button 601 is pressed, a file reference screen 605 as shown in FIG. 6-2 is displayed. On the file reference screen 605, a document file containing a word to be registered as a new word can be designated. Multiple document files can be specified. When the designated document file is selected and the delete button 602 is pressed, the designation of the document file can be canceled. When the next button 603 is pressed, the input document file specification is accepted and the unknown word registration process is started. When the cancel button 604 is pressed, the unknown word registration process is stopped.

  When the document file name is designated on the document designation screen and the next button 603 is pressed, the morphological analysis unit 102 performs morphological analysis on the document in the document file designated by the user, and the word obtained as a result of the analysis is displayed. Store in the word string buffer 131 (step S502). Next, the unknown word range expansion unit 103 acquires a word from the word string buffer 131 (step S503), refers to the part of speech information of the acquired word, and the acquired word is an unknown word or a word equivalent to an unknown word (hereinafter, It is determined whether or not it is simply an unknown word (step S504). Here, a word corresponding to an unknown word refers to a word that has no meaning as a word even in the dictionary, such as English letters and numbers.

  If the acquired word is not an unknown word (step S504: NO), the process proceeds to a determination process for determining whether all the words in the word string buffer 131 have been processed (step S515). If the acquired word is an unknown word (step S504: YES), the acquired unknown word is stored in the unknown word buffer 132 (step S505).

  Next, the unknown word range expansion unit 103 generates an extended unknown word expanded by combining the acquired unknown word with the acquired unknown word, and passes the generated extended unknown word to the partial match search unit 104 Then, the partial match search unit 104 searches the dictionary 121 for a word that matches forward except for the unknown word equivalent part of the extended unknown word (step S506).

  The unknown word range expansion unit 103 determines whether the corresponding word exists in the dictionary 121 (step S507). If it does not exist (step S507: NO), the process proceeds to a process of determining whether or not the first word has been processed, that is, whether or not the previous word has been combined by going back to the first word of the document (step S507). S514).

  If the corresponding word exists in the dictionary 121 (step S507: YES), the unknown word range expansion unit 103 stores the expanded unknown word further expanded by combining words after the unknown word in the unknown word buffer 132. (Step S508). In the first time immediately after the previous word is combined, the subsequent partial match search process and the notation similarity determination process are performed on the extended unknown word in which only the previous word is combined without combining the subsequent word. Thereafter, the same processing is performed on the extended unknown word obtained by sequentially combining the subsequent words.

  Next, the partial match search unit 104 searches the dictionary 121 for a word that partially matches a character string other than the unknown word equivalent portion of the extended unknown word (step S509). Further, the unknown word range expansion unit 103 determines whether or not the corresponding word exists in the dictionary 121 (step S510). If it does not exist (step S510: NO), the process proceeds to a process of determining whether or not processing has been performed up to the last word, that is, whether or not words have been combined up to the last word of the document (step S513).

  When the corresponding word exists in the dictionary 121 (step S510: YES), the notation similarity determination unit 105 determines that the character type of the unknown word corresponding part of the corresponding word is the character type of the unknown word acquired from the word string buffer 131. It is determined whether or not they are the same (step S511). If not identical (step S511: NO), the extended unknown word is deleted from the unknown word buffer 132 (step S512). This is because it is determined that it is not appropriate to additionally register the extended unknown word in the dictionary 121 because a word similar to the extended unknown word is not registered in the dictionary 121.

  After deleting the extended unknown word or when the character types are the same (step S511: YES), the unknown word range expansion unit 103 determines whether or not the last word of the document has been processed (step S513). If the last word has not been processed (step S513: NO), the process is repeated for the next subsequent word (step S508). If the last word has been processed (step S513: YES), the process proceeds to the next process.

  The unknown word range expansion unit 103 determines whether or not the first word of the document has been processed (step S514). If the unknown word range expansion unit 103 has not processed the first word (step S514: NO), the unknown word range expansion unit 103 sets the next word to the next previous word. The process is repeated (step S506). If the first word has been processed (step S514: YES), the process proceeds to the next process.

  The unknown word range expansion unit 103 determines whether or not all the words in the word string buffer 131 have been processed (step S515). If all the words have not been processed (step S515: NO), The word is acquired from the word string buffer 131 and the process is repeated (step S503). If all the words have been processed (step S515: YES), the process proceeds to the next process.

  The input / output control unit 101 displays on the user I / F 110 an unknown word confirmation screen for allowing the user to confirm whether or not to register the unknown word and the extended unknown word extracted in the above process in the dictionary 121 (step S516). .

  FIG. 7 is a schematic diagram showing the contents of the unknown word confirmation screen. As shown in the figure, the unknown word confirmation screen displays a list of unknown words or extended unknown words, readings, parts of speech, and a check box 701 on the left side of each unknown word or extended unknown word. . At the bottom of the unknown word confirmation screen, an all select button 702, an all cancel button 703, a correction button 704, a back button 705, a next button 706, and a cancel button 707 are displayed.

  When the user checks the check box 701, it is specified that the unknown word or the extended unknown word displayed on the right side is registered in the dictionary 121. When the select all button 702 is pressed, all the check boxes 701 are checked. When the all cancel button 703 is pressed, all check boxes 701 are unchecked. When the correction button 704 is pressed, a screen (not shown) for correcting reading and part of speech is displayed, and the user can correct reading and part of speech.

  When the return button 705 is pressed, the screen returns to the document designation screen and the document file can be designated again. When the next button 706 is pressed, the specified unknown word or extended unknown word is registered in the dictionary 121. When the cancel button 707 is pressed, the unknown word registration process is stopped.

  When the unknown word confirmation screen is displayed, the input / output control unit 101 determines whether or not the next button 706 has been pressed (step S517). If the next button 706 has not been pressed (step S517: NO), the next button 706 is waited for input. When the next button 706 is pressed (step S517: YES), the dictionary registration unit 106 registers the specified word in the dictionary 121 (step S518), and the unknown word registration process ends.

  As described above, in the dictionary registration device 100 according to the first embodiment, not only the notation of the word estimated as an unknown word by morphological analysis of the document specified by the user, but also the front of the word estimated as an unknown word, Words that are expanded by combining at least one of the rear words and similar to words already registered in the dictionary 121 can also be extracted as unknown words and registered in the dictionary.

(Second Embodiment)
The dictionary registration device according to the second embodiment calculates the similarity between the extended unknown word and the similar word according to a predetermined similarity determination rule, and when the calculated value is larger than the predetermined value The extended unknown word is registered in the dictionary simultaneously with the extracted unknown word.

  FIG. 8 is a block diagram illustrating a configuration of the dictionary registration device 800 according to the second embodiment. As shown in the figure, the dictionary registration device 800 includes an input / output control unit 101, a morpheme analysis unit 102, an unknown word range expansion unit 103, a partial match search unit 104, a notation similarity determination unit 105, a dictionary, A registration unit 106 and a user I / F 110 are provided. The dictionary registration apparatus 800 according to the present embodiment stores the word string buffer 131 and the unknown word buffer 132 in the RAM 130, and stores the analysis rule 120, the dictionary 121, and the similarity determination rule table 801 in the HDD. is doing. The HDD corresponds to a dictionary storage unit and a similarity determination rule storage unit in the present invention.

  The second embodiment is different from the first embodiment in that a similarity determination rule table 801 is added. Other configurations and functions are the same as those in FIG. 1 which is a block diagram showing the configuration of the dictionary registration apparatus 100 according to the first embodiment, and therefore, the same reference numerals are given and description thereof is omitted here.

  The similarity determination rule table 801 holds the character similarity for each character type of the comparison source character and the comparison destination character. FIG. 9 is an explanatory diagram showing an example of the structure of the similarity determination rule table 801. As shown in the figure, the similarity determination rule table 801 stores comparison source characters, comparison destination characters, and character similarity.

  Thus, the similarity determination rule table 801 stores the character similarity for each corresponding character. Accordingly, the similarity between the entire character strings is calculated by obtaining the average of the character similarities of the characters in the character strings to be compared. As an example, a case where the similarity between the character string “ABC” and the character string “Def” is calculated according to the similarity determination rule table 801 is shown below.

  The character type of the first character “A” of the character string “ABC” and the first character “D” of the character string “Def” are both uppercase letters, and the character similarity defined in the similarity determination rule table 801 is 100. It is. Since the character type of the second character “B” of the character string “ABC” is uppercase, and the character type of the second character “e” of the character string “Def” is lowercase, the character similarity is 90. It is. Furthermore, since the character type of the last character “C” of the character string “ABC” is uppercase English, and the character type of the last character “f” of the character string “Def” is lowercase English, the character similarity is 90. . The average value of these character similarities (100 + 90 + 90) / 3 = 93 represents the similarity between the character string “ABC” and the character string “Def”.

  If the number of characters in the character string to be compared is different, if the character type is the same as that of the immediately preceding character, a predetermined ratio with respect to the character similarity of the immediately preceding character, for example, 80% is set as the similarity of the character. Also good. If there is a possibility that different degrees of similarity may be calculated depending on the position of the character to be compared, the maximum value may be used as the degree of similarity.

  Next, an unknown word registration process performed by the dictionary registration apparatus 800 according to the second embodiment configured as described above will be described. FIG. 10 is a flowchart showing an overall flow of the unknown word registration process in the second embodiment.

  The document designation screen display process, the morphological analysis process, and the extended unknown word search process from step S1001 to step S1010 are the same as the process from step S501 to step S510 in the dictionary registration device 100 according to the first embodiment. The description is omitted.

  When a word that partially matches a character string other than the unknown word equivalent portion of the extended unknown word exists in the dictionary 121 (step S1010: YES), the notation similarity determination unit 105 does not know the word that partially matches the unknown word. The similarity of the word equivalent part is calculated according to the similarity determination rule table 801 (step S1011). Next, the notation similarity determination unit 105 determines whether or not the calculated value is larger than a predetermined value (step S1012).

  If the similarity is smaller than a predetermined value (step S1012: NO), the extended unknown word is deleted from the unknown word buffer 132 (step S1013). If the similarity is larger than a predetermined value (step S1012: YES), the process proceeds to a process of determining whether or not the last word of the document has been processed (step S1014).

  The process completion check process, the unknown word confirmation screen display process, and the dictionary registration process from step S1014 to step S1019 are the same processes as steps S513 to S518 in the dictionary registration apparatus 100 according to the first embodiment. The description is omitted.

  11-1 to 11-7 illustrate an example in which an unknown word and an extended unknown word are detected from a document file designated by a user in the dictionary registration apparatus 800 according to the second embodiment in accordance with the above-described unknown word registration process. It is explanatory drawing which showed the example of the process registered into the dictionary 121. FIG.

  In the example illustrated in FIG. 11A to FIG. 11-7, the case where the user specifies a document file storing the document illustrated in FIG. 11A is illustrated. First, the morphological analysis unit 102 performs morphological analysis on the document shown in FIG. 11A, and the obtained word string is stored in the word string buffer 131 (step S1002). FIG. 3 described above shows a part of the contents of the word string buffer 131 at this time. As shown in FIG. 3, in this example, the word “DME” corresponding to the word number 51 is extracted as an unknown word.

  Referring to the word string buffer 131, the unknown word range expansion unit 103 stores the word “DME” corresponding to the word number 51 in the unknown word buffer 132 (step S1005). At this time, the unknown word buffer 132 is in the state shown in FIG.

  Next, a word that matches forward with “Toshiba” that is not the unknown word equivalent part of the extended unknown word “Toshiba DME” combined with the immediately preceding word “Toshiba” is searched from the dictionary 121 (step S1006). For example, if the word shown in FIG. 2 is registered in the dictionary 121, “Toshiba”, “Toshiba AVE”, and “Toshiba AVE Co., Ltd.” are searched for as a word that matches forward. Since there is a forward matching word, the extended unknown word “Toshiba DME” is stored in the unknown word buffer 132 (step S1008). At this time, the unknown word buffer 132 is in the state shown in FIG.

  When the similarity between the unknown word equivalent part of each word and the unknown word "DME" is calculated, "Toshiba" has no unknown word equivalent part, so the similarity is 0, and the unknown word equivalent part "AVE" of "Toshiba AVE" is Since the number of characters and the character type all match, the similarity is 100. Since the unknown word equivalent part “AVE Inc.” of “Toshiba AVE Inc.” has a large number of characters, the similarity is 43 (= 300/7) when the character similarity of that part is calculated as 0. If the reference value of similarity that is determined to be similar is 75, the word “Toshiba AVE” (similarity 100) that satisfies the condition exists in the dictionary 121, and therefore the expanded unknown word “Toshiba DME” is unknown without being deleted. It remains in the word buffer 132 (step S1012).

  Next, the extended unknown word “Toshiba DME stock” combined with the backward word “stock” is stored in the unknown word buffer 132 (step S1008). At this time, the unknown word buffer 132 is in the state shown in FIG.

  The word “Toshiba” and “stock” other than the unknown word equivalent part of the extended unknown word are searched from the dictionary 121, but since there is no corresponding word, the extended unknown word “Toshiba DME stock” Is deleted from the unknown word buffer 132 (step S1013).

  Similarly, when the extended unknown word “Toshiba DME Co.” combined with the next word “company” is stored in the unknown word buffer 132 and a partial match search is executed, a similar word “Toshiba AVE Co., Ltd.” exists. Since the similarity of the unknown word equivalent portion “AVE” becomes 100, the extended unknown word is left in the unknown word buffer 132. At this time, the unknown word buffer 132 is in the state shown in FIG.

  Thereafter, the expanded unknown word “Toshiba DME Co., Ltd.” combined with the backward word “” ”is stored in the unknown word buffer 132 (FIG. 11-6), and the same processing is repeated until the end of the sentence. If it is, the next forward word is combined and the same process is repeated up to the beginning of the sentence (FIG. 11-7).

  Thus, in the dictionary registration apparatus 800 according to the second embodiment, the similarity between the extended unknown word and the registered word is determined according to the similarity determination rule table 801 storing the character similarity for each character type of the character to be compared. Can be determined. Thereby, even an extended unknown word whose character type is not the same can be extracted as an unknown word, and the accuracy of unknown word extraction can be improved.

  In the dictionary registration device according to the first and second embodiments, the unknown word and the extended unknown word are displayed on the screen so that the user can check them. You may comprise so that an unknown word and an extended unknown word may be automatically registered as it is, without selecting a word.

  Further, in the dictionary registration device according to the first and second embodiments, first the forward words are combined to generate extended unknown words, similarity is determined, and then the backward words are combined to expand unknown. Although the word is generated, it may be configured such that the backward word is combined first, or only the forward or backward word may be combined.

  The dictionary registration device according to the first or second embodiment includes a control device such as a CPU, a storage device such as a ROM (Read Only Memory) and a RAM, an external storage device such as an HDD and a CD drive device, and a display. It has a display device such as a device and an input device such as a keyboard and a mouse, and has a hardware configuration using a normal computer.

  The dictionary registration program executed by the dictionary registration apparatus according to the first or second embodiment is an installable format or executable file, and is a CD-ROM, flexible disk (FD), CD-R, DVD. (Digital Versatile Disk) or the like recorded on a computer-readable recording medium.

  Further, the dictionary registration program executed by the dictionary registration apparatus according to the first or second embodiment is stored on a computer connected to a network such as the Internet and is provided by being downloaded via the network. It may be configured. Further, the dictionary registration program executed by the dictionary registration apparatus according to the first or second embodiment may be provided or distributed via a network such as the Internet.

  The dictionary registration program according to the first or second embodiment may be provided by being incorporated in advance in a ROM or the like.

  The dictionary registration program executed by the dictionary registration device according to the first or second embodiment includes the above-described units (input / output control unit, morpheme analysis unit, unknown word range expansion unit, partial match search unit, notation similarity) The module configuration includes a determination unit and a dictionary registration unit). As actual hardware, the CPU (processor) reads the dictionary registration program from the storage medium and executes it to load each unit onto the main storage device. In addition, an input / output control unit, a morphological analysis unit, an unknown word range expansion unit, a partial match search unit, a notation similarity determination unit, and a dictionary registration unit are generated on the main storage device.

  As described above, the dictionary registration device, the dictionary registration method, and the dictionary registration program according to the present invention are suitable for a document creation system, a kana-kanji conversion system, and a speech recognition system having a function of additionally registering words that are not registered in the dictionary. Yes.

It is a block diagram which shows the structure of the dictionary registration apparatus concerning 1st Embodiment. It is explanatory drawing which shows an example of a dictionary. It is explanatory drawing which shows an example of a word string buffer. It is explanatory drawing which shows an example of an unknown word buffer. It is a flowchart which shows the unknown word registration process in the dictionary registration apparatus concerning 1st Embodiment. It is a schematic diagram which shows an example of a document designation | designated screen. It is a schematic diagram which shows an example of a document designation | designated screen. It is a schematic diagram which shows an example of an unknown word confirmation screen. It is a block diagram which shows the structure of the dictionary registration apparatus concerning 2nd Embodiment. It is explanatory drawing which shows an example of a conversion notation rule table. It is a flowchart which shows the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment. It is a schematic diagram which shows an example of the unknown word registration process in the dictionary registration apparatus concerning 2nd Embodiment.

Explanation of symbols

100, 800 Dictionary registration device 101 Input / output control unit 102 Morphological analysis unit 103 Unknown word range expansion unit 104 Partial match search unit 105 Notation similarity determination unit 106 Dictionary registration unit 110 User I / F
120 analysis rule 121 dictionary 130 RAM
131 Word string buffer 132 Unknown word buffer 601 Reference button 602 Delete button 603 Next button 604 Cancel button 605 File reference screen 701 Check box 702 Select all button 703 Cancel all button 704 Modify button 705 Return button 706 Next button 707 Cancel button 801 Similarity judgment rule table

Claims (9)

Dictionary storage means for storing a dictionary holding words;
Morphological analysis of the input document, and extraction of unknown words that are not registered in the dictionary from the input document;
An unknown word range expansion means for generating an extended unknown word in which at least one of the front and rear words of the unknown word extracted by the morphological analysis means is combined with the unknown word;
Of the extended unknown words generated by the unknown word range expansion means, a partial match that searches for a registered word registered in the dictionary that matches the notation of the expanded portion of the unknown word. Search means;
Of the registered words searched by the partial match search means, the registered registration searched by the partial match search means based on the character attribute of the notation corresponding to the unknown word and the character attribute of the unknown word notation. A notation similarity determination means for determining the similarity between the notation of the part corresponding to the unknown word and the notation of the unknown word,
Dictionary registration means for registering the extended unknown word in the dictionary when the notation similarity determination means determines that the notation of the portion corresponding to the unknown word in the registered word is similar to the notation of the unknown word When,
A dictionary registration device comprising:   The notation similarity determination means, when the character type of the notation corresponding to the unknown word in the registered word and the character type of the notation of the unknown word are the same, the unknown of the registered word The dictionary registration device according to claim 1, wherein the notation of a portion corresponding to a word and the notation of the unknown word are determined to be similar. A similarity determination rule storage means for storing a similarity determination rule table that stores character similarity for each character type of the comparison source character and the comparison target character;
The notation similarity determination means is a similarity value between a character type of a notation corresponding to the unknown word and a character type of the unknown word notation among the registered words calculated based on the similarity determination rule table. 2. The dictionary according to claim 1, wherein when the value is larger than a predetermined value, it is determined that a notation of a portion corresponding to the unknown word in the registered words is similar to a notation of the unknown word. Registration device. A morpheme analysis step in which the morpheme analysis unit performs morpheme analysis on the input document and extracts from the input document unknown words that are not registered in the dictionary stored in the dictionary storage unit that stores a dictionary that holds words When,
An unknown word range extending unit generates an extended unknown word by combining at least one of the front and rear words of the unknown word extracted by the morpheme analyzing step with the unknown word; and
A partial match search means, among the extended unknown words generated by the unknown word range expansion step, the registered words that are registered in the dictionary and are words that match the notation of the expanded part of the unknown words A partial match search step for searching from a dictionary;
The notation similarity determination means , based on the character attribute of the notation of the part corresponding to the unknown word and the character attribute of the notation of the unknown word, among the registered words searched by the partial match search step A notation similarity determination step for determining the similarity between the notation of the part corresponding to the unknown word and the notation of the unknown word among the registered words searched by the step;
When the dictionary registration means determines that the notation similarity determination step is similar to the notation of the portion corresponding to the unknown word and the notation of the unknown word in the registered word, the extended unknown word is stored in the dictionary. A dictionary registration step to register;
A dictionary registration method comprising: In the notation similarity determination step , when the notation similarity determination unit has the same character type of the notation of the part corresponding to the unknown word and the character type of the notation of the unknown word, The dictionary registration method according to claim 4, wherein the notation of a portion corresponding to the unknown word in the registered words is determined to be similar to the notation of the unknown word. In the notation similarity determination step , the notation similarity determination means calculates based on a similarity determination rule table that holds the character similarity for each character type of the comparison source character and the comparison destination character stored in the storage means When the similarity value between the character type of the portion corresponding to the unknown word and the character type of the unknown word in the registered word is greater than a predetermined value, the registered word The dictionary registration method according to claim 4, wherein the notation of a portion corresponding to an unknown word is determined to be similar to the notation of the unknown word. Morphological analysis means, and morphological analysis of the input document, the unknown word not registered in the dictionary stored in the dictionary storage means for storing a dictionary for holding a word, morphological analysis procedure for extracting from the input document When,
An unknown word range extension means for generating an extended unknown word in which at least one word before and behind the unknown word extracted by the morphological analysis procedure is combined with the unknown word; and
A partial match search means, among the extended unknown words generated by the unknown word range expansion procedure, is a word that matches the notation of the expanded part of the unknown word and is registered in the dictionary Partial match search procedure to search from dictionary,
The notation similarity determination unit is configured to perform the partial match search based on a character attribute of a portion corresponding to the unknown word and a character attribute of the unknown word notation among the registered words searched by the partial match search procedure. A notation similarity determination procedure for determining the similarity between the notation of the part corresponding to the unknown word and the notation of the unknown word among the registered words searched by the procedure;
When the dictionary registration means determines that the notation similarity determination procedure indicates that the notation of the portion corresponding to the unknown word in the registered word is similar to the notation of the unknown word, the extended unknown word is stored in the dictionary. Dictionary registration procedure to register,
Dictionary registration program for executing on a computer. In the notation similarity determination procedure , when the notation similarity determination unit has the same character type of the notation of the part corresponding to the unknown word and the character type of the notation of the unknown word, 8. The dictionary registration program according to claim 7, wherein it is determined that a notation of a portion corresponding to the unknown word in the registered words is similar to a notation of the unknown word. In the notation similarity determination procedure , the notation similarity determination means calculates based on a similarity determination rule table that holds the character similarity for each character type of the comparison source character and the comparison destination character stored in the storage means When the similarity value between the character type of the portion corresponding to the unknown word and the character type of the unknown word in the registered word is greater than a predetermined value, the registered word The dictionary registration program according to claim 7, wherein the notation of a portion corresponding to an unknown word is determined to be similar to the notation of the unknown word.

Images