JP3300492B2 - Dictionary inspection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dictionary inspection apparatus for a natural language processing apparatus for processing a natural language document using an electronic computer.

[0002]

2. Description of the Related Art In recent years, computer-aided natural language processing techniques have increased, and various document processing systems have been developed. In particular, a machine translation system that automatically translates a large number of documents into a document in another language has been put to practical use.

[0003] In a machine translation system, in addition to a translation dictionary (hereinafter referred to as a standard dictionary) generally provided as a default of the system, a technical term dictionary (hereinafter referred to as a specialized dictionary) for translating documents in various specialized fields is provided. In general, a user registration dictionary (hereinafter, referred to as a user dictionary) for the user to register words not registered in the standard dictionary or the specialized dictionary is used.

The user dictionary is a dictionary created by the user as described above. Generally, when a word is registered in the user dictionary, a word in the first language (Japanese in the case of Japanese-English translation) and a second
In addition to the headline of the language (English for Japanese-English translation), information such as the part of speech and how it is used is also registered. Also, words registered in the user dictionary are usually given priority over words registered in the standard dictionary and the specialized dictionary during translation processing.

[0005] However, words registered in the user dictionary have higher priority than words registered in other dictionaries, so that it may not be appropriate to register them in the user dictionary. For example, in the state where the word "no" is registered in the standard dictionary, the proper noun "no" is registered in the user dictionary.
If the word is registered, the morphological analysis may fail because “NODE” registered in the user dictionary has priority over “NODE” in the standard dictionary. For example, when a user tries to translate the sentence "Nodesan is a good person,"
Suppose that the translation did not go well because the proper noun "no" was not registered in the dictionary. In that case, the user registers the proper noun "node",
You can get the correct translation result for this sentence.
However, after performing this registration, if the user translates the sentence "I like her because she is kind," the proper noun "NODE" registered in the user dictionary takes precedence over "NODE" in the standard dictionary. Therefore, "NODE" is interpreted as a proper noun, and a correct translation result cannot be obtained. As described above, although registration in the user dictionary is useful for improving the quality of translation of a part that the user is paying attention to, there is a possibility that the quality of translation may deteriorate due to side effects as a whole. In particular, when a word registered by the user conflicts with an attached word during morphological analysis, the risk is high.

Further, by registering words in the user dictionary, consistency with the standard dictionary is lost, and as a result, the quality of translation may be reduced. For example, despite the fact that the standard dictionary already has the "sa" noun "restore" (a verb with the stem "no" and a suffix "su", which is hereinafter referred to as "sa noun"), a new "restore" If the noun is registered in the user dictionary, consistency with the standard dictionary will be lost. As a result, for example, when translating a sentence "Restore corrupted data.", "Restore" is not recognized as one word, and there is a risk that the quality of the translation may be reduced.

[0007] A similar problem also occurs when a new noun "restore" is newly registered in the user dictionary even though the user-defined dictionary has already registered the noun "restore".

[0008] Also, when the use of words is involved, registration in the user dictionary may be a problem. Consider a case where the user registers the proper noun "Kanda" in a state where the five-step inflection verb "cam (bite)" is registered in the standard dictionary. In this case, the proper noun "kana" registered by the user matches the conjunctive form of the five-conjugative verb "kamu""kan" plus the auxiliary verb "ta"("da" after the five-step conjugative verb in the line). Therefore, as in the previous case, it is useful for improving the translation quality of the part that the user is paying attention to, but may have side effects on the translation of other parts.

[0009]

The user dictionary is indispensable for the machine translation system. However, if the words registered in the user dictionary are not appropriate, the translation process may be adversely affected. The present invention has been made to solve such a problem, and an object of the present invention is to detect a word registered in a user dictionary when there is a possibility of adversely affecting the translation process so that a user can take appropriate action. An object of the present invention is to provide a dictionary inspection device that can take an action.

[0010]

According to the present invention, there is provided a dictionary for a machine translation apparatus or the like in which the contents to be registered in the dictionary may adversely affect the translation process. It checks whether it has any adverse effect by using the dictionary access function or morphological analysis function, and issues a warning to the user, interrupts the registration process according to an external instruction, It is characterized in that the contents of the inspection are notified by giving an instruction to make a change or the like.

[0011]

According to the present invention, when there is a possibility that the contents registered in the dictionary may adversely affect the translation process, the user can take appropriate action by detecting and notifying the translation process. This can prevent inappropriate words from being registered in the dictionary.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an example in which a dictionary registration item inspection device 101, which is a main part of the present invention, is combined with a normal dictionary registration device 102, an error message processing selection device 103, and a dictionary 104 to constitute a dictionary registration device with a registration item inspection device. It is.

In FIG. 1, a dictionary registration device 102 designates a headword, a part of speech, utilization information, semantic information, a translated word, a case pattern, a registration mode, a translation rule, and the like of a word to be registered in accordance with a user's instruction. , An integrated circuit or the like. The dictionary registration item specified by the dictionary registration device 102 is passed to the dictionary registration item inspection device 101 to check whether the content is suitable for registering in the user dictionary. Specifically, the content to be registered is written in an inspection registration item register 201 in the dictionary registration item inspection device described later. Dictionary registration item inspection device 101
Is similarly realized by computer software, an integrated circuit, and the like. The dictionary registration item inspection device 101 is a dictionary registration device 102
The dictionary registration items passed from are inspected by referring to the dictionary and performing morphological analysis, and the contents that are not suitable for registration,
Or if any potential content is found,
An error message indicating this and a processing option that can be selected by the user are returned to the dictionary registration device 102. If the received error message and the processing option are not empty, the dictionary registration apparatus 102 passes them to the error message display processing selection apparatus 103. If the error message and the processing option are empty, the dictionary registration apparatus 102 registers the error message and the processing option in the user dictionary as a dictionary registration item, and the dictionary registration ends. . If the error message and the processing options are not empty, the error message display processing selection device 103 presents the received error message and processing options to the user using a display, a printer, or the like. The error message display processing selection device 103 is realized by computer software, an integrated circuit, or the like. The user selects the error message display processing selection device 1
From the processing options indicated by 03, an appropriate one is selected using a keyboard, a mouse or the like. The error message display processing selection device 103 notifies the dictionary registration device 102 of the process selected by the user. The dictionary registration apparatus 102 performs the selected processing, for example, cancels registration, changes the registered contents, and forcibly registers the contents. If there is any content to be written in the user dictionary, it actually writes it in the dictionary 104.

FIG. 2 is a block diagram of the dictionary registration item inspection device. The dictionary registration item inspection device uses the inspection registration item register 20
1, an error checking device 202, an error message buffer 203, a processing option creation device 204, and a processing option storage buffer 205. Inspection registration item register 2
Reference numeral 01 stores the contents of dictionary registration items to be inspected that have entered the dictionary registration item inspection device 101, and is realized by a memory or the like on a computer. The inspection registration item register 201 can store all the contents of one dictionary item to be inspected, such as headings, stems, parts of speech, utilization information, translations, translation rules, registration modes, and the like. The color inspection device 202 refers to the inspection registration item register 201 and the dictionary 104, inspects whether the contents of the inspection registration item register 201 have no problem as the dictionary registration contents, and if any problem is found, the problem is found. The content of the problem is output to the error message buffer 203, and is realized by computer software, an integrated circuit, or the like. The error message buffer 203 stores the content of an error found by the error checking device 202, and is realized by a memory or the like on a computer. The processing option creating device 204 is the error checking device 2
02, referring to the contents of the error message buffer output in step S02, generating the user's processing options in accordance with the output error items and combinations thereof, and outputting the generated options to the processing option storage buffer 205. This is realized by computer software, an integrated circuit, or the like. The processing option storage buffer 205 stores processing options that can be selected by the user and created by the processing option creation device, and is realized by a memory or the like on a computer.

FIG. 3 is a configuration diagram of the error message buffer 203. The error message buffer 203 can store an error number unique to the error detected by each error detection routine to be described later and an error message corresponding to the error as a pair. Message storage area 3
02.

FIG. 4 is a block diagram of the error checking device 202. The error checking device includes an error checking unit A401, an error checking unit B402, and an error checking unit C403.
Each part is the inspection registration item register 201, the dictionary 1
04 can be referred to. Here, the dictionary 104 includes both a standard dictionary and a user dictionary. An error number is assigned to an error detected in each of the error checking unit A401, the error checking unit B402, and the error checking unit C403 according to the type. Each part outputs an error number for the detected error to the error message buffer 203 together with an error message for the detected error.

The error checking unit A401 detects an error which can be detected by referring to only the contents written in the check registration item register 201. Error checker B
Reference numeral 402 denotes an error which can be detected by referring to the dictionary 104 together with the contents written in the test registration item register 201.

The error checking unit C403 not only refers to the contents written in the test registration item register 201 and the dictionary 104, but also uses morphological processing to obtain the dictionary registration items written in the test registration item register 201. It is intended to detect errors that can be detected in consideration of inflected forms, derived forms, and inflected forms of the contents written in the dictionary.

FIG. 5 is a configuration diagram of the error checking unit A401. The error checking unit A401 is configured only by the error detecting unit 501. The error detector 501 refers to the contents of the test registration item register 201, and outputs an error number and an error message of the detected error to an error message buffer.

FIG. 6 is a flowchart showing the processing procedure of the error detection unit 501. The purpose of the processing is to detect that the ending of the sa-various noun ends with "do". In the figure, in step ST601, the part of speech of the test registration item register 201 is substituted for the variable "part of speech". In step ST602, the stem of the test registration item register 201 is substituted for the character string variable "stem". Step ST6
At 03, it is checked whether or not the value of the variable “part of speech” is “sa-variable noun”. If the value of "part of speech" is not "sa noun", the process ends. If the value of "part of speech" is "sa noun", the process proceeds to step ST604. Step ST6
In 04, it is checked whether the last two characters of the character string variable "stem" are "do". If the last two characters are not "do", the process ends. If the last two characters are “Yes”, the process proceeds to Step ST605. Step ST
At 605, an error number unique to the error detection unit 501 and an error message unique to the error
Write to 03.

FIG. 7 is a block diagram of the error checking unit B402. The error checking unit B402 includes a dictionary lookup unit 701, a dictionary lookup result storage buffer 702, an error detection unit 703, and a similar part-of-speech table 704. The relationship between the dictionary lookup unit 701, the dictionary lookup result storage buffer 702, the error detection unit 703, and the similar part-of-speech table 704, and the relationship between the dictionary 104, the test registration item register 201, and the error message buffer 203 will be described in detail below. Shown in

FIG. 8A shows a dictionary lookup result storage buffer 7.
02 is a configuration diagram. In the figure, a standard dictionary dictionary lookup result storage area 801 is used when the dictionary lookup unit 701 searches the standard dictionary for one or more items having the same heading as the inspection registration item register 201, including those having different parts of speech. The results (part of speech, utilization, translation, translation rules, registration mode, etc.) are stored as table elements.

In the same figure, one or a plurality of user dictionary dictionaries are stored in the user dictionary dictionary search result storage area 802 by the dictionary lookup unit 701. When a search is performed, the result (part of speech, inflection, translation, translation rule, registration mode, etc.) is stored as a table element.

FIG. 8B is a configuration diagram of the similar part-of-speech table 704. In the similar part-of-speech table 704, a plurality of pairs of part-of-speech that the user is likely to make mistakes at the time of registration are written.

The procedure of the process of the error checking unit B402 will be described with reference to the flowchart of FIG. In the figure, in step ST901, the heading of the test registration item register 201 is substituted for the variable "heading" of the dictionary lookup unit. In step ST902, the contents of the dictionary lookup result storage buffer 702 are cleared. In step ST903, the dictionary lookup unit 701 searches the standard dictionary for items having the same heading as the variable “heading”, including those having different parts of speech, and writes the searched items to the standard dictionary dictionary lookup result storage area 801. In step ST904, the dictionary lookup unit 701 searches the user dictionary for items having the same heading as the variable “heading”, including those having different parts of speech, and writes the searched items to the user dictionary dictionary lookup result storage area 802. In step ST905, the error detection unit 703 performs error detection by a method described later with reference to the dictionary lookup result storage buffer 702 and the test registration item register 201, and stores an error number and an error message corresponding to the detected error in the error message buffer 203. Write to.

FIG. 10 is a flow chart showing the procedure of the processing of the error detecting section 703. The error detection unit 703 not only refers to the test registration item register 201 but also stores the dictionary 1
04 is a central part of the error checking unit B402 for detecting an error that can be detected with reference to FIG. In FIG. 10, in step ST1001, a detection routine 10
“Detection of error 1 relating to part-of-speech error” by “01”. In step ST1002, "detection of error relating to registration mode 1" by detection routine 1002 described below is performed. In step ST1003, “detection of error relating to registration mode 2” by detection routine 1003 described below is performed. In step ST1004, "detection of error 2 relating to part-of-speech error" by detection routine 1004 described below is performed. Step ST1005
Then, “detection of an error related to an attached word” is performed by a detection routine 1005 described later.

FIG. 11 is a flowchart showing the processing procedure of a detection routine 1001 which is a component of the error detection unit 703. The detection routine 1001 detects that “the headword that is the same as the headword of the test registration item register is registered in the user dictionary as a part of speech that is likely to be wrong”, and in that case, the error message buffer 203 stores the information unique to the detection routine 1101. This is for writing the error number and error message.

In the figure, in step ST1101, the part of speech in the test registration item register 201 is substituted for the variable "part of speech 1". In step ST1102, the variable i is initialized to 1. In step ST1103, the user dictionary dictionary lookup result storage area 80 in the dictionary lookup result storage buffer 702
It is checked whether the i-th element exists in 2. If the i-th element does not exist, the process ends. If the i-th element exists, the process proceeds to step ST1104. In step ST1104, the part of speech of the i-th element is assigned to the variable “part of speech 2”. In step ST1105, "part of speech 1" and "part of speech 2" are previously stored in the similar part of speech table 704.
Is checked to see if it is a pair of part of speech prepared in. The similar part-of-speech table 704 contains pairs of parts of speech that are
"Noun and adjective verb" and "noun and adjective verb" are set. If “part of speech 1” and “part of speech 2” are pairs of parts of speech written in the similar part of speech table 704, the process proceeds to step ST1106. If not, the process proceeds to Step ST1107. In step ST1106, the error number unique to the detection routine 1001 and the unique error message are written in the error message buffer 203, and the process proceeds to step ST1107. Step ST110
At 7, the value of the variable i is incremented by 1, and the process returns to step ST1103.

The above processing will be described with an example. It is assumed that the user dictionary has already registered a semantic noun “separated and divided”, and the user is going to further register “separated and divided” as a noun. Only one word having the same heading as the word “separation division” to be registered in the user dictionary is extracted from the user dictionary by the dictionary lookup unit 701 and written in the user dictionary dictionary lookup result storage area 802. Since the heading of the sub-variable noun “separation division” in the user dictionary is “separation division”, the sub-variant noun “separation division” is written in the first element of the user dictionary dictionary lookup result storage area 802. In step ST1101,
Since the part of speech “noun” of the word “separation and division” to be registered in the inspection registration item register 201 is written, it is substituted for the variable “part of speech 1”. Step ST1102
Is substituted for 1 in the variable i. In step ST1103, since the first element exists, the process proceeds to step ST1104. In step ST1104, the part of speech “sa noun” of the first element in the user dictionary dictionary lookup result storage area 802 is assigned to the variable “part of speech 2”. In step ST1105, referring to the similar part-of-speech table 704, the value “noun” of the variable “part of speech” and the value “sa-variant noun” of the variable “part-of-speech 2” are found to be a pair of the part-of-speech that is likely to be wrong.
Proceed to 1106. In step ST1106, an error number and an error message are output to error message buffer 203. In step ST1107, the value of the variable i is increased to 2, and the process returns to step ST1103. In step ST1103, since the second element does not exist in the user dictionary dictionary lookup result storage area 802, the detection routine 1001 ends.

FIG. 12 is a flowchart showing the processing procedure of a detection routine 1002 which is a component of the error detection unit 703. The detection routine 1002 is performed in a mode in which the same entry word of the same part of speech as the item of the test registration item register is in the standard dictionary, and the item of the test registration item register is treated as a different item from the same entry word of the standard dictionary in the translation process. Is detected, and in that case, an error number unique to the detection routine 1002 and an error message unique to the detection routine 1002 are written in the error message buffer 203.

In the figure, in step ST1201,
The value of the variable “registration mode” in the test registration item register 201 is substituted for the variable “registration mode”. The variable “registration mode” in the test registration item register 201 is 1 when the item is handled as a different item from the same item in the standard dictionary, and is 0 when the item is merged. In step ST1202, the value of the variable “part of speech” in the test registration item register 201 is substituted for the variable “part of speech 1”. In step ST1203, it is checked whether the value of the variable “registration mode” is 0 or 1, and if it is 0, the process ends. In the case of 1, it proceeds to step ST1204. In step ST1204, the variable i is initialized to 1. In step ST1205, it is checked whether the i-th element in the standard dictionary dictionary lookup result storage area 801 exists. If the i-th element does not exist, the process ends. If the i-th element exists, the process proceeds to step ST1206. In step ST1206, the part of speech of the i-th element in the standard dictionary lookup result storage area 801 is assigned to the variable “part of speech 2”. Step ST12
At 07, it is checked whether "part of speech 1" and "part of speech 2" are the same. If they are the same, step ST1209
Proceed to. If they are not the same, the process proceeds to Step ST1208. In step ST1208, the value of the variable i is added by 1,
It returns to step ST1205. In step ST1209, the error number unique to the detection routine 1002 and the unique error message are written in the error message buffer 203, and the process ends.

FIG. 13 is a flowchart showing the processing procedure of a detection routine 1003 which is a component of the error detection unit 703. The detection routine 1003 sets a mode in which the same entry word of the same part of speech as the item under inspection is in the standard dictionary, and the same entry word of the same part of speech as the item in the inspection registration item register is to be handled in the translation process in the user dictionary. Is registered in a mode different from the item to be checked ", and in that case, an error number unique to the detection routine 1003 and an error message unique to the detection routine 1003 are written in the error message buffer 203. .

In the figure, in step ST1301, the value of the registration mode of the test registration item register 201 is substituted for the variable “registration mode 1”. In step ST1302, the part of speech in the test registration item register 201 is substituted for the variable “part of speech 1”. In step ST1303, 1 is substituted for the variable i. In step ST1304, it is checked whether or not the i-th element exists in the standard dictionary dictionary lookup result storage area 801. If not, the process ends. If it exists, it proceeds to step ST1305. Step ST1
In 305, the part of speech of the i-th element in the standard dictionary dictionary lookup result storage area 801 is substituted for the variable “part of speech 2”. In step ST1306, it is checked whether the variable “part of speech 1” and the variable “part of speech 2” are the same. If they are not the same, the process proceeds to Step ST1307. If they are the same, the process proceeds to Step ST1308. In step ST1307, the value of the variable i is incremented by 1, and the process returns to step ST1304. In step ST1308, 1 is substituted for the value of the variable i. In step ST1309, it is checked whether the i-th element exists in the user dictionary dictionary lookup result storage area 802. If not, the process ends. If it exists, it proceeds to step ST1310. Step ST131
At 0, the part of speech of the i-th element in the user dictionary dictionary result storage area 802 is substituted for the variable “part of speech 2”. In step ST1311, the user dictionary dictionary lookup result storage area 80
The value of the registration mode of the i-th element of 2 is assigned to the variable “registration mode 2”. In step ST1312, it is checked whether the variable “part of speech 1” and the variable “part of speech 2” are the same, and whether the variable “registration mode 1” and the variable “registration mode 2” are the same. If the condition is satisfied, step ST
Proceed to 1314. When the condition is not satisfied, the process proceeds to Step ST1313. In step ST1313, the value of the variable i is added by 1, and the process returns to step ST1309. In step ST1314, an error number unique to the detection routine 1003 and an error message unique to the detection routine 1003 are written into the error message buffer 203, and the process ends. FIG. 14 is a flowchart showing the procedure of the processing of the detection routine 1004, which is a component of the error detection unit 703. is there. The detection routine 1004 detects that “a word having the same headword as the headword of the item in the test registration item register, and the part of speech of the item in the test registration item register and the part of speech that is easily mistaken exists in the standard dictionary”. In this case, an error number unique to the detection routine 1004 and a unique error message are written in the error message buffer 203.

In the figure, in step ST1401, the part of speech in the test registration item register 201 is substituted for the variable “part of speech 1”. In step ST1402, 1 is substituted for the variable i. In step ST1403, it is checked whether or not the i-th element exists in the standard dictionary dictionary lookup result storage area 801. If the i-th element does not exist, the process ends. If the i-th element exists, step ST1
Proceed to 404. In step ST1404, the part of speech of the i-th element in the standard dictionary dictionary lookup result storage area 801 is assigned to the variable “part of speech 2”. In step ST1405, it is checked whether or not “part of speech 1” and “part of speech 2” are pairs of parts of speech prepared in the similar part of speech table 704 in advance. In the similar part-of-speech table 704, one or a plurality of pairs of part-of-speech that are likely to be wrong are set in advance. “Part of speech 1” and “part of speech 2” are similar parts of speech table 704
If it is the pair of parts of speech prepared in step ST1406, the process proceeds to step ST1406. If “part of speech 1” and “part of speech 2” are not pairs of parts of speech prepared in the similar part of speech table 704, the process proceeds to step ST1407. Step ST14
In 06, an error number unique to the detection routine 1004 and a unique error message are written in the error message buffer 203, and the process proceeds to step ST1407. In step ST1407, the value of the variable i is incremented by 1, and the process returns to step ST1403.

FIG. 15 is a flowchart showing the procedure of processing of a detection routine 1005 which is a component of the error detection unit 703. The detection routine 1005 detects that "the same entry as the entry of the inspection registration item register and its part of speech is an annex is present in the standard dictionary". This is for writing the error number and the unique error message.

In the figure, in step ST1501, 1 is substituted for a variable i. In step ST1502, it is checked whether the i-th element exists in the standard dictionary dictionary lookup result storage area 801. If the i-th element does not exist, the process ends. If the i-th element exists, the process proceeds to step ST1503. In step ST1503, the part of speech of the i-th element in the standard dictionary dictionary lookup result storage area 801 is assigned to the variable “part of speech”. Step ST1504
Then, it is checked whether the variable "part of speech" represents an annex. If it represents an ancillary word, step ST
Proceed to 1505. If it does not represent an ancillary word, it proceeds to step ST1506. In step ST1505, an error number and an error message unique to the detection routine 1005 are written in the error message buffer 203, and the process proceeds to step ST1506. Step ST1
In step 506, the value of the variable i is incremented by 1, and step ST150
Return to 2.

FIG. 16A shows the components of the error detection unit C403. The error detection unit C403 uses the utilization generation unit 16
01, a utilization result storage buffer 1602, a morphological analysis unit 1603, a morphological analysis result storage buffer 1604, an error detection unit 1605, and a utilization result counter i 1606. Interactions 1601 to 1606 and Dictionary 1
04, the detection registration item register 201, and the error message buffer 203 will be described in detail in the following description.

FIG. 16B is a configuration diagram of the morphological analysis result storage buffer 1604. As shown in the figure, the morphological analysis result storage buffer can be used to write the headings, parts of speech, inflected forms, inflected forms, from which dictionaries, etc. of words obtained as a result of the morphological analysis. Also, it is possible to understand how a phrase is formed by one or a plurality of words. The figure shows a state in which the character string “kana” has been subjected to morphological analysis.

The processing procedure of the error investigation unit C403 will be described with reference to the flowchart of FIG. In the figure, in step ST1701, the value of the part of speech in the test registration item register is substituted for the variable “part of speech”. Step ST1702
Then, it is checked whether the variable "part of speech" represents the part of speech to be used. If it represents a part of speech to be used, the process proceeds to step ST1703. If it does not represent the part of speech to be used, the process proceeds to step ST1704. In step ST1703, the utilization generation unit 1601 refers to the heading, part of speech, and utilization information of the inspection registration item register 201.
Writes in the utilization result storage buffer 1602 all the utilized forms for the items in the examination registration item register 201. The utilization generation unit 1601 has a table in which, for example, stems, nouns, utilization forms (patterns), and the like are stored, and is written into the utilization result storage buffer 1602 by referring to the table. In step ST1704, the header of the test registration item register 201 is written as it is to the first element of the utilization result storage buffer 1602. Step ST
At 1705, 1 is assigned to the utilization result counter i1606. In step ST1706, it is checked whether or not the i-th element exists in the utilization result storage buffer 1602. If the i-th element does not exist, the processing of the error checking unit C403 ends. If the i-th element exists, the process proceeds to step ST1707. Step ST1
In step 707, the morphological element analysis unit 1603 morphologically analyzes the character string of the i-th element in the utilization result storage buffer 1602 with reference to the dictionary 104. The result of the morphological analysis is written to the morphological analysis result storage buffer 1604. If the morphological analysis is not successful, the morphological analysis result storage buffer 1604 is cleared. In Step ST1708, the error detecting unit 1605 is executed. At this time, the error detecting unit 1605 uses the utilization result storage buffer 1602, the morphological analysis result storage buffer 1604, and the test registration item register 2
01, refer to utilization result counter i1606. If an error is detected, the error number and the error message are written to the error message buffer 203. Step ST1
In step 709, the value of the utilization result counter i1606 is incremented by 1, and the process returns to step ST1706. Note that the contents of the utilization result storage buffer 1602 and those subjected to morphological analysis by the morphological analysis unit 1603 are introduced into the test registration item register 2011.

A number of morphological analysis methods have been conventionally proposed, and a description thereof will be omitted.
FIG. 18 is a flowchart illustrating the procedure of the process of the error detection unit 1605.

The error detection unit 1605 not only refers to the contents written in the test registration item register 201 and the dictionary 104 as described above, but also uses morphological processing to
An error checking unit C403 for detecting an error that can be detected in consideration of the utilization form and the derivative form of the dictionary registration item written in the examination registration item register 201 and the utilization form and the derivative form of the contents written in the dictionary. Is an error detection unit. In FIG. 18, in step ST1801, detection routine 1 is performed.
In step 801, “error detection related to use of test items and words in dictionary” is performed. In step ST1802, “utilization of inspection items and detection of errors related to words in dictionary” are performed by detection routine 1802. Step ST
In step 1803, the detection routine 1803 performs "detection of an error relating to the use of inspection items and the use of words in the dictionary". In step ST1804, the detection routine 180
4, "Detection of errors relating to the use of inspection items and the use of ancillary words in the dictionary" is performed.

FIG. 19 is a flowchart showing the processing procedure of a detection routine 1801, which is a component of the error detection unit 1605. The detection routine 1801 indicates that “when the registered item to be inspected is a word that is not used,
04 is used to write a unique error number and a unique error message to the detection routine 1801 in the error message buffer 203 in that case. is there.

In the figure, in step ST1901, the part of speech in the test registration item register 201 is substituted for the variable “part of speech”. In step ST1902, it is checked whether the “part of speech” represents a part of speech that is not used. If the “part of speech” indicates the part of speech to be used, the process ends. If it represents a part of speech that is not used, step ST19
Go to 03. In step ST1903, it is checked whether the result is stored in the morphological analysis result storage buffer 1604 and the morphological analysis is completed within one phrase. If no result is stored in the morphological analysis result storage buffer 1604 or the morphological analysis has been completed for two or more phrases, the process ends. If the result is stored in the morphological analysis storage buffer 1604 and the morphological analysis has been completed within one phrase, the process proceeds to step ST1904. Step ST1
At 904, it is checked whether the head item of the morphological analysis storage buffer 1604 is a part of speech to be used. If it is not the part of speech to be used, the process ends. If it is a part of speech to utilize, the process proceeds to step ST1905. Step ST190
In step 5, an error number specific to the detection routine 1801 and a specific error message are stored in the error message buffer 20.
Write to 3.

The above process will be described with an example. Consider a case in which the user attempts to register the proper noun "Kanda" in a state where the verb "cam (bite)" is registered in the standard dictionary. In this case, the word that the user is trying to register matches the conjunctive form of the verb "kamu" with "kan" plus the auxiliary verb "ta"("da" after the five-row inflection verb). Care must be taken and the user needs to be notified. The procedure of the detection routine 1801 in this case will be described. Before entering the detection routine, the morphological analysis unit 1603 performs a morphological analysis on the character string “kana” in advance, and the resulting verb “kamu” plus the auxiliary verb “ta” is stored in the morphological analysis result storage buffer 1604. In step ST1901, the word “kana” being registered in the examination registration register 201 is registered.
Is assigned to the variable “part of speech”. In step ST1902, the value “proper noun” of the variable “part of speech” represents the part of speech that is not used, so
Proceed to 3. In step ST1903, the morphological analysis result storage buffer 1904 contains the resulting verb "kamu" plus the auxiliary verb "ta", and it is known that the morphological analysis has been completed within one phrase, so the flow proceeds to step ST1904. In step ST1904, morphological analysis buffer 1
Since the first item of 604, the verb “kam”, is found to be a part of speech to be used, the process proceeds to step ST1905. In step ST1905, the error number and the error message are output to the error message buffer, and the process of the detection routine 1801 ends.

FIG. 20 is a flowchart showing the processing procedure of a detection routine 1802, which is a component of the error detection unit 1605. The detection routine 1802 detects that “the use of the registration item to be checked matches the word that is not used in the dictionary 104”, and in that case, the error number and the specific error message unique to the detection routine 1802 are stored in the error message buffer 203. Is for writing.

In the figure, in step ST2001, the value of the utilization result counter i1606 is substituted for a variable n.
In step ST2002, the examination registration item register 201
Is assigned to the variable “part of speech 1”. Step ST2
In 003, it is checked whether or not the variable “part of speech 1” represents the part of speech to be used. If it does not represent the part of speech to be used, the process ends. If it represents a part of speech to be used, the process proceeds to step ST2004. Step S
In T2004, the character string of the n-th item in the utilization result storage buffer 1602 is substituted for the character string variable “utilization”. In step ST2005, the morphological analysis result storage buffer 1
It is checked whether the result is stored in 604. If no result is found, the process ends. If the result is included, the process proceeds to Step ST2006. In step ST2006, the heading and the part of speech of the first item in the morphological analysis result storage buffer are substituted for the character string variable "dictionary heading" and the variable "part of speech 2", respectively. In step ST2007, it is checked whether or not the variable “part of speech 2” represents an unused part of speech. If “part of speech 2” is to be utilized, the process ends. If "part of speech 2" is not utilized, step ST
Proceed to 2008. In step ST2008, it is checked whether or not the character string variable “utilization” and the character string variable “dictionary heading” are the same. If they are not the same, the process ends. If they are the same, the process proceeds to Step ST2009. Step S
In T2009, an error number unique to the detection routine 1802 and an unique error message are written to the error message buffer 203.

FIG. 21 is a flowchart showing the processing procedure of a detection routine 1803, which is a component of the error detection unit 1605. The detection routine 1803 detects that “when the registered item to be inspected is a word to be used, the used form matches the used form of the word to be used in the dictionary 104”.
Is to write a unique error number and a unique error message in the detection routine 1803.

In the figure, in step ST2101, the part of speech in the test registration item register 201 is substituted for the variable “part of speech”. In step ST2102, it is checked whether or not the variable “part of speech” represents the part of speech to be used. If it does not represent the part of speech to be used, the process ends. If it is a part of speech to utilize, the process proceeds to step ST2103. In step ST2103, the morphological analysis result storage buffer 1
A morphological analysis result is entered in 604, and it is checked whether morphological analysis has been performed for one word. If the result is not stored in the morphological analysis result storage buffer 1604 or the morphological analysis is not performed for one word, the process ends. If the result is stored in the morphological analysis result storage buffer 1604 and morphological analysis is performed with one word, step ST2104
Proceed to. In step ST2104, it is checked whether or not the part of speech of the word stored in the morphological analysis result storage buffer 1604 is the part of speech to be used. If it is not a part of speech to be used, the process ends. If the part-of-speech is to be used, step S
Proceed to T2105. In step ST2105, the error number unique to the detection routine 1803 and the unique error message are written to the error message buffer 203.

FIG. 22 is a flowchart showing the processing procedure of a detection routine 1804 which is a component of the error detection unit 1605. The detection routine 1804 detects that “when the registered item to be inspected is a word to be used, the used form matches the used form of the auxiliary word in the dictionary 104”. This is for writing a unique error number and a unique error message in the detection routine 1804.

In the figure, in step ST2201, the part of speech in the test registration item register 201 is substituted for the variable “part of speech”. In step ST2202, it is checked whether or not the variable “part of speech” represents the part of speech to be used. If it does not represent the part of speech to be used, the process ends. If it is a part of speech to utilize, the process proceeds to step ST2203. In step ST2203, morphological analysis result storage buffer 1
A morphological analysis result is entered in 604, and it is checked whether morphological analysis has been performed for one word. If the result is not stored in the morphological analysis result storage buffer 1604 or the morphological analysis is not performed for one word, the process ends. If the result is stored in the morphological analysis result storage buffer 1604 and morphological analysis is performed with one word, step ST2204
Proceed to. In step ST2204, it is checked whether or not the part of speech of the word stored in the morphological analysis result storage buffer 1604 is an ancillary word to be used. If it is not an annex to be used, terminate. If it is an ancillary word to be used, the process proceeds to step ST2205. In step ST2205, the error number unique to the detection routine 1804 and the unique error message are written to the error message buffer 203.

FIG. 23 is a block diagram of the processing option creation device 204. The processing option creation device 204 includes a processing option creation unit 2301 and an error classification table 2302. The processing option creation unit 2301 refers to the error number storage area 301 and the error classification table 2302 of the error message buffer 203, and stores in the processing option storage buffer 205 a user-selectable countermeasure for one or more detected errors. Output.

FIG. 24 is a configuration diagram of the error classification table 2302. The errors that may be found by the error checking device 202 are determined in advance according to whether the errors are serious or not as shown in the figure, and whether the system can be automatically corrected or not. Classified into tables. An error number specific to the error is written in the table. Therefore, all error numbers written in the error number storage area 301 of the error message buffer 203 can be classified by referring to this table.

FIG. 25 is a configuration diagram of the processing option storage buffer 205. The option area 2051 is composed of three variables: a variable “compulsory registration”, a variable “correction registration”, and a variable “cancellation of registration”. When each variable is 1, the option represented by those variables can be selected. Express that. When each variable is 0, it indicates that the option represented by those variables is not selectable. Reference numerals 2502 and 2503 in FIG. 4 are used to store error numbers of errors to be corrected, and have meaning only when the variable “correction registration” is 1. The corrected error error number storage area 2502 is an area for writing the error number of the error to be corrected when the error is corrected and registered. The correction candidate error error number storage area 2503 is an area for writing an error number of an error for which the user is left to determine whether to correct the error when registering the error. An error number of a serious error among the detected correctable errors is written to a correction error error number storage area, and a non-critical error is written to a correction candidate error error number storage area.

FIG. 26 is a flowchart showing the procedure of the process of the processing option creating unit 2301. In the figure, in step ST2601, the processing option storage buffer 205 is cleared. Specifically, the processing option storage buffer 20
All five variables and regions are cleared to zero. In step ST2602, it is checked whether one or more error numbers are written in the error number storage area 301 of the error message buffer 203. If no error number has been written, the processing of the processing option creating unit 2301 ends. If one or more error numbers have been written, the process proceeds to step ST2603. Step ST260
In step 3, it is checked with reference to the error number storage area 301 and the error classification table 2302 whether a serious and uncorrectable error has been detected. Although there is a part with no possibility of modification here, for example, as shown in the above-mentioned example, it indicates an attached word itself, an object containing an attached word, an adverb, and the like. If there is a serious and uncorrectable error, the process proceeds to step ST2604. If there is no serious and uncorrectable error, the process proceeds to step ST2605. In step ST2604, the variable “registration cancellation” in the processing option storage buffer 205 is set to 1, and the processing of the processing option creation unit 2301 ends. In step ST2605, the error number storage area 301 and the error classification table 23
It is checked whether all of the error numbers written in the error number storage area 301 are error numbers for non-critical non-correctable errors by referring to both of the numbers 02.
If all the errors are non-critical non-correctable errors, the process proceeds to step ST2606. Otherwise, the mobile terminal proceeds to step ST2607. Step ST2
In step 606, two variables “forced registration” and “correction registration” in the processing option storage buffer 205 are set to 1 and step S
Proceed to T2611. In step ST2607, it is checked with reference to the error number storage area 301 and the error classification table 2302 whether there is an error number for a serious error in the error numbers written in the error number storage area 301. If there is a serious error, step ST
Proceed to 2608. When there is no serious error, the process proceeds to Step ST2610. In step ST2608, the variable “correction registration” and the variable “cancellation of registration” in the processing option storage buffer 205 are reduced to one, and the process proceeds to step ST2609.
Proceed to. In step ST2609, referring to the error number storage area 301 and the error classification table 2302, an error number of a serious error that can be corrected is written in the correction error error number storage area 2502 of the processing option storage buffer 205, and the process proceeds to step ST2611. Step ST2
At 610, the variable “forced registration”, the variable “correction registration”, and the variable “registration cancellation” of the processing option storage buffer 205 are made one, and the process proceeds to step ST2611. Step ST
At 2611, referring to the error number storage area 301 and the error classification table 2302, the processing option storage buffer 2
The error number of a non-critical error that can be corrected is written in the correction candidate error error number storage area 2503 of the process 05, and the processing of the processing option creating unit 2301 ends.

This will be described with reference to an example. For example, when the user has registered the noun “write” (used like “write to disk”) in the standard dictionary,
Is registered in the user dictionary as a mode that is treated as a different word from the same entry in the standard dictionary in the translation process.
You try to register by mistake. In this case, an error is detected by the detection routine 1002 in the error checking device 202, and an error number “4” unique to the detection routine 1002 is written in the error number storage area 301 of the error message buffer 203. Also, the noun “write” to be registered matches the conjunctive form of the verb “write” in the standard dictionary.
The error is detected by the detection routine 1801 in the error message buffer 02, and the error number “6” unique to the detection routine 1801 is written in the error number storage area 301 of the error message buffer 203. Thus, two error numbers are written in the error number storage area 301 of the error message buffer 203. Error classification table 23
According to 02, the error number 4 is classified as a correctable non-critical error, and the error number 6 is classified as an uncorrectable non-critical error. The processing option creation unit 2301 performs the processing in FIG. 26 while referring to both the error number storage area 301 and the error classification table 2302. In this case, the processing is performed in ST2601, ST2602, ST2603, ST26.
05, ST2607, ST2610, and ST2611, and as a result, all of the three conversions “forced registration”, “correction registration”, and “registration cancellation” in the option area 2501 of the option processing buffer 205 become “1”. The error number 4 is written in the correction candidate error error number storage area 2503. In this case, nothing is written in the correction error error number storage area 2502.

Description will be made with another example. For example, consider the case where the user attempts to add a new translation to the particle "at" (used as "wedding was performed in a church"). Adding a new translation for the user can help improve the translation of the part of the user who is currently focused on, but can cause side effects when translating other sentences. In the case of an annex such as "at," the side effects are particularly likely to be significant. In this case, an error is detected by the detection routine 1005 in the error checking device 202, and an error number “2” unique to the detection root 1005 is written in the error number storage area 301 of the error message buffer 203. At the same time, “de” is the verb “naru (similar)”.
The error is also detected by the detection routine 1801 in the error checker 202, and the error number unique to the detection routine 1801 is stored in the error number storage area 301 of the error message buffer 203. 6 "is written. According to the error classification table 2302, the error number 2 is classified as an uncorrectable serious error, and the error number 6 is classified as an uncorrectable non-critical error. The processing option creation unit 2301 stores the error number storage area 301 and the error classification table 2302
26 is performed with reference to both of the above. In this case, the processing is performed in ST2601, ST2602, ST2603, S
The process proceeds in the order of T2604, and as a result, among the three variables in the option area 2501 of the processing option processing buffer 205, only the variable “registration cancellation” becomes 1. In this case, nothing is written in the correction error / error number storage area 2502 and the correction candidate error / error number storage area 2503.

The processing option storage buffer 205 created by the processing option creating device 204 is sent to the dictionary registration device 102 together with the error message buffer 203.
The dictionary registration device 102 sends the received processing option storage buffer 205 and error message buffer 203 to the error message display processing selection device 103.

FIG. 27 is a flowchart showing the procedure of the process performed by the error message display process selecting device 103. In the figure, in step ST2701, the dictionary registration device 10
The error message and the option button are displayed on the display with reference to the error message buffer 203 and the processing option storage buffer 205 received from the server 2. At this time, the option area 250 of the processing option storage buffer 205
Only the button corresponding to the variable that is 1 out of the 3 variables in 1 is presented to the user. Step ST270
In 2, the user selects one of the option buttons displayed on the display using a mouse, a keyboard, or the like. In Step ST2703, it is checked whether or not the button selected by the user is a button corresponding to the variable “correction registration”. If the button selected by the user is a button corresponding to “correction registration”, the process proceeds to step ST2705. Otherwise, it proceeds to step ST2704. In step ST2704, the process corresponding to the selected button is transmitted to the dictionary registration device 102, and the process of the error message display process selection device 103 ends. In step ST2705, it is checked whether an error number is written in the correction candidate error error number storage area 2503 of the processing option storage buffer 205. If an error number has been written, the process proceeds to Step ST2706. If no error number is written, step ST2
Proceed to 707. In step ST2706, the user is inquired of which of the correction candidate errors to correct using a display or the like, and the user selects one or a plurality of errors to be corrected using a mouse, a keyboard, or the like. In step ST2707, the error correction and registration, and which error is to be corrected are transmitted to the dictionary registration device 102, and the processing of the error message display processing selection device 103 ends.

FIG. 28 is an error message and option button window presented to the user when the user tries to register the noun “write” in the user dictionary in a mode in which the word is treated as a different word from the same word in the standard dictionary. . In this case, three buttons of “forced registration”, “correction registration”, and “registration cancellation” are presented to the user along with two error messages.

FIG. 29 is a window of an error message and an option button presented to the user when a new translation is to be added to the particle "". In this case, two error messages are displayed. Also, only the button of "cancel registration" is presented to the user.

As described above, when the contents to be registered in the dictionary are added to the existing dictionary information and the morphological analysis is performed, the possibility that ambiguity will occur or increase will be predicted by the number. It is. This includes contents that cannot be modified.

The present invention is not limited to the above embodiment. First of all, the conditions to be inspected in the error inspection units A, B, and C are only examples, and similar inspection items can be added as appropriate.

The error classification table 2302 shown in FIG.
In FIG. 25, a medium error classification is provided in addition to the serious error and the non-critical error, and the processing option storage buffer 20 shown in FIG.
By providing a variable corresponding to a new button in the fifth option area 2501 and making any change to the flowchart of FIG. 26, the combination of processing options that can be specified by the user can be different from the above-described embodiment. It is.

As shown in FIG. 30, the dictionary registration item inspection device 101 which is a main part of the present invention is replaced with a dictionary search device 300.
1, error message display processing selection device 103, dictionary 1
When used in conjunction with the user dictionary 04, a user dictionary inspection device that inspects words already registered in the user dictionary can be configured.

Referring to FIG. 30, a dictionary search device 3001 retrieves, modifies and writes contents already registered in the user dictionary 104 one by one, and is realized by computer software, an integrated circuit, or the like. Dictionary search device 300
The dictionary registration item extracted from the user dictionary in step 1 is passed to the dictionary registration item inspection device 101. As in the case of FIG. 1, the dictionary registration item inspection device 101 inspects the passed dictionary registration items while performing morphological analysis with reference to the dictionary.
The error message and the processing options are stored in the dictionary search device 300.
Return to 1. If the received error message and processing options are not empty, the dictionary search device 3001 passes them to the error message display processing selection device 103. If the error message and the processing options are empty, the process proceeds to the next dictionary item check. If the error message and the processing options are not empty, the error message display processing selecting device 103 uses a display, a keyboard, a mouse, and the like, and allows the user to select an appropriate processing option from the processing options as in the case of FIG. . Error message display processing selection device 1
Reference numeral 03 denotes a process selected by the user in the dictionary search device 3
Tell 001. The dictionary search device 3001 performs selected processing, for example, cancels registration, changes registered contents, and updates the contents of the user dictionary 104 if necessary at that time.

Further, the present invention is not limited to the above embodiment. For example, it goes without saying that the dictionary inspection device of the present invention can be applied to general dictionaries that use natural language processing technology, such as a document search system and a word processor, in addition to a dictionary of a machine translation device.

[0067]

As described above, according to the present invention, when there is a possibility that the contents registered in the user dictionary may have an adverse effect on the translation process, this is detected, and the user takes an appropriate action, thereby allowing the user to take appropriate measures. It is possible to prevent inappropriate contents from being registered in the dictionary and prevent the translation processing from being adversely affected by the registered contents in the user dictionary.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a dictionary registration device with a registration item inspection device, in which a dictionary registration item inspection device is combined with a normal dictionary registration device or the like.

FIG. 2 is a block diagram showing a configuration of a dictionary registration item inspection device.

FIG. 3 is a block diagram showing a configuration of an error message buffer.

FIG. 4 is a block diagram showing a configuration of an error checking device.

FIG. 5 is a block diagram showing a configuration of an error checking unit A.

FIG. 6 is a flowchart illustrating a procedure of a process performed by an error detection unit 501;

FIG. 7 is a block diagram illustrating a configuration of an error checking unit B.

FIG. 8 is a block diagram showing the configuration of a dictionary lookup result storage buffer and an explanatory diagram of a similar part-of-speech table.

FIG. 9 is a flowchart illustrating a procedure of a process performed by an error detection unit B;

FIG. 10 is a flowchart showing a procedure of processing of an error detection unit 703.

FIG. 11 is a flowchart showing the procedure of processing of a detection routine 1001.

FIG. 12 is a flowchart showing a procedure of a detection routine 1002;

FIG. 13 is a flowchart showing the procedure of a detection routine 1003;

FIG. 14 is a flowchart showing the procedure of processing of a detection routine 1004.

FIG. 15 is a flowchart showing the procedure of processing of a detection routine 1005.

FIG. 16 is a block diagram showing a configuration of an error checking unit C and a diagram showing a configuration of a morphological analysis result storage buffer.

FIG. 17 is a flowchart illustrating a procedure of a process performed by an error detection unit C;

FIG. 18 is a flowchart illustrating a procedure of processing of an error detection unit 1605.

FIG. 19 is a flowchart illustrating a procedure of a detection routine 1801;

FIG. 20 is a flowchart illustrating a procedure of a detection routine 1802;

FIG. 21 is a flowchart showing the procedure of processing of a detection routine 1803;

FIG. 22 is a flowchart showing the procedure of the processing of a detection routine 1804.

FIG. 23 is a block diagram illustrating a configuration of a processing option creation device.

FIG. 24 is an explanatory diagram of an error classification table.

FIG. 25 is a block diagram showing a configuration of a processing option storage buffer.

FIG. 26 is a flowchart illustrating a procedure of a process performed by a processing option creating unit;

FIG. 27 is a flowchart showing a procedure of processing of the error message display processing selection device.

FIG. 28 is a diagram showing a window of an error message and a processing option button to be submitted to the user when registering the noun “write” in a mode in which the noun “write” is handled as a different one from the standard noun “write”.

FIG. 29 is a view showing an error message presented to the user when trying to add a new translation to the particle “at” and a window of a processing option button.

FIG. 30 is a block diagram showing a configuration of a user dictionary inspection device in which a dictionary registration item inspection device is combined with a dictionary search device and the like.

Continuation of the front page (56) References JP-A-3-29054 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 17/21-17/28

Claims (3)

(57) [Claims] An inspection item storage unit for storing an item to be inspected for a word to be registered in a dictionary; a similar part of speech storage unit for storing a combination of parts of speech that are likely to be wrong in advance; A dictionary lookup unit for extracting items to be inspected for words that have the same heading and are registered in the dictionary in advance; and a storage unit for storing the inspection items for words registered in the dictionary in advance extracted by the dictionary lookup unit. A dictionary lookup result storage unit, and a part of speech information stored as an inspection item in the storage unit and a part of speech information stored as an inspection item in the dictionary lookup result storage unit, the error stored in the similar part of speech storage unit. Error detecting means for detecting as an error when the combination of easy-to-speak parts of speech, and when an error is detected by this error detecting means,
A dictionary inspection apparatus, comprising: a message display unit for displaying an error message. 2. Inspection item storage means for storing an item to be inspected for a word to be registered in the dictionary, and a word to be inspected for a word having the same heading as the word to be registered and registered in the dictionary in advance A dictionary lookup unit for extracting items; a dictionary lookup result storage unit for storing test items of words registered in advance in the dictionary extracted by the dictionary lookup unit; When the stored part-of-speech information is a particle, an error detecting means for detecting as an error, and when an error is detected by the error detecting means,
A dictionary inspection apparatus, comprising: a message display unit for displaying an error message. 3. An inspection item storage unit for storing an item to be inspected for a word to be registered in the dictionary; a morphological analysis unit for morphologically analyzing the word to be registered in the dictionary; A morphological analysis result storage unit for storing an analysis result of a word to be morphologically analyzed and an inspection item obtained from the morphological analysis result, wherein a part of speech stored as an inspection item in the storage unit is not used. Then, morphological analysis succeeds within one phrase,
Error detection means for detecting as an error when the part of speech stored as the inspection item of the first word obtained from the morphological analysis result is not utilized; and when an error is detected by this error detection means,
A dictionary inspection apparatus, comprising: a message display unit for displaying an error message.