JP3847801B2 - Character processing apparatus and processing method thereof - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to a Japanese sentence, a character processing apparatus having a function of narrowing down a phrase structure of a target sentence, checking a phrase structure by phrase dependency, reading of words constituting the phrase, notation, and part of speech, and processing thereof It is about the method.
[0002]
[Prior art]
Conventionally, as a character processing device with a function to analyze the phrase structure of an input sentence for a Japanese sentence, input a kana reading, analyze the phrase structure, and convert it to an appropriate kanji character string There is a Kanji conversion device.
[0003]
In Kana-Kanji conversion, input reading sequences are analyzed to create possible phrase candidates, conversion candidates are determined from the combinations, and presented in the most likely order. Then, a candidate desired by the operator is selected from the presented conversion candidates. In order to increase the likelihood of conversion candidates presented as the first candidate, conventionally, a method of narrowing down conversion candidates to different parts of speech and a method of arranging them in order of word frequency have been implemented.
[0004]
Also, a method has been proposed in which a phrase structure is determined with reference to a sentence pattern case pattern called a valence pattern, and conversion processing to an appropriate kanji character string is performed.
[0005]
In addition, examples that appear in pairs, such as “read a book”, are registered in the example dictionary in advance, and the first candidate is determined according to the examples in the example dictionary by referring to the example dictionary at the time of conversion. A technique called example conversion has also been implemented. In addition to examples that describe the relationship between words such as “read a book”, examples that describe relationships with semantic classifications that group words together are proposed. For example, if the example of “write on [concrete object]” is registered, this example is also applied to “paper” which is [concrete object], and the first conversion candidate “write on paper” can be obtained. it can. By selecting a semantically possible word pair as an example to be registered in the example dictionary and using this example conversion mechanism, the possibility of obtaining a correct conversion result has increased.
[0006]
[Problems to be solved by the invention]
However, in the phrase structure analysis process that refers to the conventional example, only the applicable examples are referred to, so it is necessary to have all the semantically possible examples in the example dictionary, and the examples are not registered. On the other hand, even a combination that is clearly semantically inconsistent cannot be eliminated, and an unusually strange conversion result may be output. In addition, the number of semantically possible examples is enormous, and the capacity of the example dictionary memory for storing all these examples is enormous, making it difficult to achieve high-precision analysis efficiency at low cost. It was. Furthermore, depending on the document field, if an example is applied and priority interpretation is performed, analysis results that are semantically inconsistent may be obtained. For example, it is desirable not to give priority to ecstasy and metaphor expression in the technical field. To give an example, the idiom “grab a cloud” is not preferred in technical documents that discuss organisms.
[0007]
The present invention has been made in view of the above points, and its purpose is to distinguish an inappropriate word pair as the first analysis candidate of the phrase structure from the example suitable for the first analysis candidate. If the suppression example that matches the document field of the input sentence is detected when determining the phrase structure by storing the suppression example together with the field information to be suppressed, the first word pair that matches the suppression example It is an object to provide a character processing device and a processing method for controlling such that it is not output as an analysis candidate.
[0008]
[Means for Solving the Problems]
In order to achieve the above object, the character processing apparatus of the present invention has the following configuration.
[0009]
Word dictionary means for storing word readings and notations in association with each other, notation of first and second words constituting the example, and an example type indicating whether the example should be prioritized or suppressed , An example storage means for indicating a specific field to which the example is to be applied and indicating the field, and indicating that it should be generally applied if there is no field to be applied; Input means for inputting a character string, and field designation information for designating a field to be converted with priority given to the character string input by the input means. The field designation information accepting means received from the operator and the field designation information received by the field designation information accepting means The field designation information storage means for storing, the word dictionary means, the character string input by the input means is divided into phrase candidates, the phrase candidate creating means, and the phrase candidate created by the phrase candidate creating means The search means for searching for a combination of words corresponding to the example stored in the example storage means, and whether the combination of the words found by the search means corresponds to the example to be prioritized or the example to be suppressed The application field of the example to which the combination of the example type determination means for determination and the word found by the search means corresponds should be generally applied, or is a specific field and stored in the field designation information storage means The field determination means for determining whether the field matches the field specified by the field specification information, and the example type determination means should prioritize the type of the example. When it is determined as an example, the combination of words corresponding to the example is given priority as a phrase candidate, and when the example type determination unit determines that the example type should be suppressed, the field determination unit Therefore, if it is determined that it should be generally applied or matches the specified field, the word combination corresponding to the example is suppressed as a phrase candidate and determined not to match the specified field In such a case, the combination of words corresponding to the example is not suppressed as a phrase candidate, and has phrase candidate determination means for determining a phrase candidate.
[0012]
The processing method of the character processing apparatus according to the present invention includes the following steps.
[0013]
Word dictionary means for storing word readings and notations in association with each other, notation of first and second words constituting the example, and an example type indicating whether the example should be prioritized or suppressed , An example storage means for indicating a specific field to which the example is to be applied and indicating the field, and indicating that it should be generally applied if there is no field to be applied; Input means for inputting a character string, field designation information storage means for storing field designation information for designating a field to be converted with priority given to a character string input from the input means, and a phrase for storing phrase candidates Candidate storage means; processing means for executing various processes based on a program; A phrase candidate creation means, a search means, an example type judgment means, a field judgment means, which are realized by cooperating with the processing means and the program storage means. Means for determining phrase candidates A processing method of a character processing device comprising: Create phrase candidates Means for dividing the character string input from the input means into phrase candidates with reference to the word dictionary means, and storing the created phrase candidates in the phrase candidate storage means; Search A means for searching for a combination of words corresponding to the example stored in the example storage unit among the phrase candidates stored in the phrase candidate storage unit; Example type judgment An example type determination step in which means refers to an example type of an example to which a combination of words found by the search step corresponds from the example storage unit, and determines whether the example type is a priority example or a suppression example. And said Field judgment The means refers to the example application field of the example to which the combination of words found in the search step corresponds from the example storage unit, and the example application field should be generally applied or is a specific field. A field determination step of determining whether or not the field specified by the field specification information stored in the field specification information storage means matches the field Determination of phrase candidates When it is determined in the example type determination step that the example type of the example is a priority example, the means prioritizes a combination of words corresponding to the example as a phrase candidate, and the example type determination step If the example type is determined to be an example to be suppressed, the field determination step should generally apply, or if it is determined to match the specified field, the word corresponding to the example If it is determined that the combination is not matched with the specified field, the combination of words corresponding to the example is not suppressed as a phrase candidate, and the phrase candidate is determined. Process.
[0015]
【Example】
Hereinafter, the present invention will be described in detail with reference to the drawings, taking as an example a kana-kanji conversion device in which an input target is limited to a Japanese kana reading sequence.
[0016]
FIG. 1 is a diagram showing an overall configuration of an embodiment according to the present invention.
[0017]
In the figure, reference numeral 1 denotes a microprocessor (CPU) which performs operations for character processing, logical determination, etc., and is connected via an address bus (AB), a control bus (CB), and a data bus (DB). Control each component. Here, the address bus transfers an address signal indicating a component to be controlled by the CPU 1. The control bus transfers and applies a control signal of each component to be controlled by the CPU 1. The data bus transfers data between the component devices. Reference numeral 2 denotes a read-only fixed memory (ROM), and reference numeral 2a denotes a program area (PA) for storing control procedures (FIGS. 9 to 15) by the CPU 1.
[0018]
Reference numeral 3 denotes a writable random access memory (RAM) composed of 16 bits per word, and is used for temporary storage of various data from each component. Reference numeral 3a denotes a word dictionary (WDIC) for performing kana-kanji conversion, and FIG. 3 shows its configuration. Reference numeral 3b denotes an example dictionary (YRDIC) storing combinations of phrases and phrases, and FIG. 4 shows the configuration thereof. An input reading buffer memory (YBUF) 3c stores readings input from the keyboard. Reference numeral 3d denotes a document buffer (TBUF), which is a memory for storing document information input from the keyboard. 3e is a phrase candidate table (BCT) that stores phrase candidates in the middle of kana-kanji conversion. 3f is a homophone buffer memory (DBF) for storing unconfirmed phrase candidates for Kana-Kanji conversion, and FIG. 6 shows the configuration thereof. 3g is a homophone pool memory (DBP) for storing other candidates for kana-kanji conversion corresponding to the homophone buffer memory, and FIG. 7 shows its configuration. 3h is a flag (BFLG) that designates preferential conversion of words to which field information is given among words stored in the word dictionary.
[0019]
Reference numeral 4 denotes an external memory (DISK) for storing a standard document. The created document is stored, and the stored document is called when necessary by an instruction from the keyboard. Reference numeral 5 denotes a keyboard (KB), which includes various function keys such as alphabetic keys, hiragana keys, character symbol input keys such as katakana keys, and conversion keys for instructing conversion. Here, 5a is a key for inputting a reading (YOMI), 5b is a conversion instruction key (CON) for converting the input reading, and 5c is a next candidate for changing a conversion candidate and converting it to a next candidate. A conversion instruction key (NXT), 5d is determined in the current homophone display direction, and at the same time a selection key (SEL) for instructing to learn the candidate notation, and 5e is stored in the word dictionary WDIC. This is a biological field designation key (BIO) for designating priority conversion of the biological field.
[0020]
Reference numeral 6 denotes a cursor register (CR), whose contents are read and written by the CPU 1. A CRT controller, which will be described later, displays a cursor at a position on the display device with respect to the address stored here. Reference numeral 7 denotes a display buffer memory (DBUF), which stores patterns such as document information stored in the document buffer 3d. Reference numeral 8 denotes a CRT controller (CRTC), which plays a role of displaying the contents stored in the cursor register 6 and the display buffer 7 on the display device. Reference numeral 9 denotes a display device (CRT) using a cathode ray tube or the like, and a dot configuration pattern and cursor display in the display device are controlled by a CRT controller 8. Reference numeral 10 denotes a character generator (CG) which stores character and symbol patterns to be displayed on the display device CRT.
[0021]
The character processing device including the respective components operates in response to various inputs from the keyboard 5. When a key input is supplied from the keyboard 5, an interrupt signal is first sent to the CPU 1, and the CPU 1 Reads various control signals stored in the ROM 2 and performs various controls in accordance with these control signals.
[0022]
An example of performing kana-kanji conversion in the apparatus of the present embodiment having the above-described configuration will be described below with reference to FIG.
[0023]
In FIG. 2, the symbol “/” in the left input reading string represents a conversion key for instructing conversion. First, in (a), when “Moji is Ishiyake” is entered, “character is written on stone” is converted as the first candidate. This is because the example “write in [substance]” is registered in the example dictionary 3b, and “stone” is described as [substance] in the word dictionary 3a. Other “wills” and “doctors” with the same reading “Ishi” are not [substance], so this example is not applicable. Without such an example, a semantically inconsistent conversion such as “write to will” would result.
[0024]
Next, when “Mihon Mizukunikaku” is entered in (b), “Water” is also designated as “Substance” in the word dictionary 3a as in (a), but in the example dictionary 3b, Since “write in water” is registered as an example to be suppressed, the example “write in [substance]” is not applied, and “write in water” is excluded from the first conversion candidates. Further, in (c), the input “grab the cloud” is input, and the example “grab a cloud” is applied as a first conversion candidate when inputting a general document. However, in the input of a biological field document, “grab the cloud”. Will be suppressed and become "grab".
[0025]
FIG. 3 is a diagram illustrating an example of the word dictionary 3a of the character processing device according to the present embodiment. The word dictionary 3a includes “reading (YM)”, “notation (HK)”, “part of speech (GI)”, “word likelihood (FQ)”, “semantic classification (SF)”, and “field information (CT)” for one word. "7 fields of field likelihood (CF)". Here, a value of 1 to 5 is stored in the word likelihood as information indicating the likelihood of the word itself such as frequency information. Likelihood value 5 is interpreted as most likely, and as the value decreases, the likelihood of being the first conversion candidate is lacking. The semantic classification describes the meaning of the word and stores “concept”, “substance”, and the like. This semantic classification is given only to words whose part of speech is a noun. The field likelihood is a value indicating the likelihood of the field information indicated in the field information.
[0026]
FIG. 4 is a diagram illustrating an example of the example dictionary 3b of the character processing device according to the present embodiment. The example dictionary 3b includes “first word information (LW)”, “second word information (RW)”, “particle connection information (CJ)”, “example application type (TP)”, “example application field ( BY) ". Here, the first and second word information stores a pointer to the head of the corresponding word in the word dictionary 3a. Each word information stores a semantic classification in addition to the word, and a word group based on the semantic classification can be designated. Further, the particle connection information describes a particle that connects the first word information and the second word information. The example application type stores either “normal type” or “suppression type”. The “normal type” example is an example in which the registered pattern should be given priority, and the “suppression type” example is an example in which the registered pattern does not become the first candidate. In the example application field, field information on which the example should be applied or suppressed is stored. For example, the example of “write in [substance]” in the figure is an example that can be generally applied to words having a semantic classification [substance]. In addition, “grab a cloud” is an example that can be applied outside the biological field, but in the biological field, “grab a cloud” is an example that should be suppressed from becoming a first candidate.
[0027]
FIG. 5 is a diagram illustrating an example of the field designation flag 3h of the character processing device according to the present embodiment. The field designation flag 3h is set to indicate whether one field corresponds to one bit and whether priority conversion is performed or not. In the state shown in the figure, the “biology” field whose bit value is “1” is easier to be preferentially converted than the other fields.
[0028]
FIG. 6 is a diagram illustrating an example of the phrase candidate table 3e of the character processing device according to the present embodiment. The phrase candidate table 3e is a table in which a sequence of phrases that can be analyzed from the beginning to the end of the input reading sequence is represented by a tree structure having one phrase candidate as a node, and is composed of five fields. In the figure, BCID is an ID number that makes each phrase candidate unique. BCJW represents an independent part word of a phrase candidate and is a pointer to the word dictionary 3a where the independent part word exists. BCFW is an adjunct word string that constitutes an appendix of phrase candidates, and stores an index on the input reading buffer 3c for an adjunct start position and adjunct end position. Here, the symbol “φ” means that there is no attached word string. SLNK links another phrase candidate whose phrase start reading position is the same as that phrase candidate, and stores a linked phrase candidate ID. A SLNK value of “−1” is a phrase candidate ID that does not exist, means a termination code, and indicates that there is no more same reading start position phrase. DLNK links phrase candidates subsequent to the phrase candidate, and stores the phrase candidate ID of the same link destination as SLNK. When the value of DLNK is “−1”, it means that there is no further phrase candidate as in SLNK.
[0029]
For example, the phrase candidate ID “0” “Mizu ni” has “without looking” or “without checking” as the phrase candidate at the same reading start position, and “each” and “without checking” as the subsequent phrase candidates. , Has the same reading start position clause "nuclear""write""missing".
[0030]
FIG. 7 is a diagram illustrating an example of the homophone buffer 3f of the character processing device according to the present embodiment. In FIG. 7, HID is a homophone ID, and is an ID number that makes each of the homophones created for each phrase unique. JW represents a self-supporting word of the homophone and stores a pointer to the word dictionary 3a. The FW stores a code similar to the kana reading code that constitutes the reading field of the word dictionary 3a for the attached word string of the homophone. The PHID stores the homophone ID of the subsequent pair homophone that is a pair of examples together with the homophone. When PHID is “−1”, it means that there is no subsequent pair homophone. The APYR stores the example applied between the subsequent pair homophones when determining the first candidate as a pointer to the head of the corresponding example in the example dictionary 3b.
[0031]
FIG. 8 is a diagram illustrating an example of the homonym pool 3g of the character processing device according to the present embodiment. The homophone pool 3g stores all possible word candidates as homophone candidates for the homophone buffer that stores only the information of the first candidate. In the figure, which homophone is a candidate for HID is stored as the homophone ID of the corresponding homophone buffer DBF. JWP stores independent part words that can be used as homophones of the corresponding phrase as pointers to the word dictionary 3a. The FWP stores an attached word string for the independent part word JWP. For example, a phrase having a homophone ID “10” has two homophone candidates “flower” and “nose”.
[0032]
Next, the operation in the present embodiment will be described with reference to the flowchart shown in FIG.
[0033]
First, in step S101, a key is pressed from the keyboard 5 to wait for an interrupt to occur. When a key is input, the process proceeds to step S102, where the input key is determined, and the process branches to one of steps S103 to S108 depending on the type of the input key.
[0034]
Step S103 is processing when the reading input key (YOMI) 5a is pressed, and stores the pressed reading code in the input reading buffer memory (YBUF) 3c. Step S104 is processing when the conversion key (CON) 5b is pressed. The character string to be converted into kana-kanji converted and input in step S103 is converted into kanji and output to an output buffer (not shown). To do. Step S105 is processing when the next candidate conversion key (NXT) 5c is pressed, and displays another candidate for the homophone in the homophone word pool corresponding to the homophone buffer output in step S104.
[0035]
Step S106 is processing when the selection key (SEL) 5d is pressed, confirms the homophone in the output buffer displayed on the screen, and outputs the confirmed character string in the document. Further, it is determined whether or not the selected word matches the example applied to determine the first candidate, and if necessary, the applied example is stored as a suppression pattern in association with the field information. Step S107 is processing when the biological field designation key (BIO) 5e is pressed, and the value of the bit of the priority conversion field designation flag (BFLG) 3h corresponding to the designated field is set to “1”. Step S108 is processing when a key other than YOMI, CON, NXT, SEL, BIO, for example, a key used for document editing such as a cursor movement key is pressed, and is generally performed in the same type of character processing apparatus. This process is well known and will not be described in particular.
[0036]
Next, the details of the “conversion process” in step S104 will be described with reference to the flowchart shown in FIG.
[0037]
First, in step S201, a phrase candidate creation process (detailed later) is performed to create a phrase candidate table (BCT) 3e. Next, in step S202, a first candidate determination process, which will be described later in detail, is performed to narrow down candidates suitable as first conversion candidates from the phrase candidates stored in the phrase candidate table 3e. In step S203, based on the first candidate determined in step S202, the conversion result is created and output to the homophone buffer (DBF) 3f. At the same time, other candidates for the same reading start position and phrase length that did not become the first candidate are stored in the homophone pool (DBP) 3g.
[0038]
Details of the “sentence candidate creation process” in step S201 will be described below with reference to the flowchart shown in FIG.
[0039]
First, in step S301, the phrase candidate creation start index i on the input reading buffer (YBUF) 3c and the index j of the phrase candidate table (BCT) 3e are initialized to “0”. In step S302, the word dictionary (WDIC) 3a is searched based on the reading in the reading buffer 3c indicated by i to obtain a word candidate. Next, in step S303, a phrase connection verification process for analyzing an attached word string to be connected to the found word candidate is performed to obtain a phrase candidate. In step S304, the obtained phrase candidates are stored in the phrase candidate table 3e, and the independent word words and the attached word string information obtained as a result of word search and phrase connection test are stored in the BCJW and BCFW fields, respectively, SLNK, Other phrase candidates that can have this phrase candidate in DLNK are also set. After all the information is stored in the phrase candidate table 3e, the index j of the phrase candidate table 3e is counted up in step S305.
[0040]
Next, in step S306, it is determined whether or not all words having the same phrase start position, i.e., words having the same reading start position, have been searched. Returning to S302, the search is continued. If all the words at the same reading start position have been searched, the process advances to step S307 to find the phrase end position indicated by the attached post-reading sequence described in the phrase candidate table 3e as the reading start position of the subsequent phrase. Then, the found phrase end position is set to i as the phrase candidate start position. At this time, if all the phrase candidates are terminated, that is, if the DLNK value of the phrase candidate table 3e is “−1”, the end of the input reading has been reached. finish. However, if not terminated, the process returns to step S302 with i set, and the subsequent phrase candidate creation process is continued.
[0041]
Details of the “first candidate determination process” in step S202 will be described below with reference to the flowchart shown in FIG.
[0042]
First, in step S401, a counter memory max serving as an index value for determining the first candidate is initialized to “0”. This max indicates the maximum likelihood among all the phrase candidates, and the likelihood 0 is the minimum value. In step S402, one phrase candidate string is extracted from the phrase candidate table 3e. Next, in step S403, the phrase pair likelihood of the extracted phrase candidate string is calculated. The phrase pair likelihood is calculated as follows.
[0043]
(Phrase pair likelihood) = {sum of (word likelihood)} + {sum of (inter-phrase likelihood)}
+ {Sum of (word field likelihood)}
Here, the value of the FQ field of the word dictionary 3a is used for the word likelihood. The phrase likelihood is a value indicating the likelihood of connection between phrases, and is fixed at “0” in the present embodiment. The field likelihood of a word is given as a field likelihood CF × 8 when the field information in the CT field of the word dictionary 3a matches the field designation flag (BFLG) 3h indicating the field at the time of conversion.
[0044]
Next, in step S404, an example applicable to the extracted phrase candidate string is searched, and the example application likelihood is added to the phrase pair likelihood. This example dictionary search process will be further described later with reference to FIG. In step S405, it is determined whether the calculated phrase pair likelihood is greater than the maximum likelihood max. If it is larger, the process proceeds to step S406, and the maximum likelihood is updated with the calculated phrase pair likelihood. In step S407, the phrase candidate string corresponding to the maximum likelihood is stored in the work memory did. The work memory did has the same configuration as the phrase candidate table 3e, except that SLNK and DLNK values are set to be unique so as to link the phrase candidates corresponding to the phrase candidate string having the maximum likelihood.
[0045]
In step S408, it is determined whether another phrase candidate string can be extracted from the phrase candidate table 3e. If it can be extracted, the process returns to step S402 to continue the search. When all the phrase candidate strings have been processed and no more phrase candidate strings can be extracted, the process returns.
[0046]
Details of the “example dictionary search process” in step S404 will be described below with reference to the flowchart shown in FIG.
[0047]
First, in step S501, an example search attention phrase is extracted from the phrase candidate string extracted in step S402 of FIG. 12, and the example dictionary (YRDIC) 3b is searched for whether there is a pair phrase candidate constituting the example. In the next step S502, it is determined whether there is a pair clause candidate constituting the example, and if not found, the process returns. However, if it is found, the processing after step S503 is performed.
[0048]
In step S503, the type of the found example is checked. If the example is the example of “suppression type”, the application field of the suppression example is further checked in step S504, and the application field and field designation flag ( If the field information indicated by (BFLG) 3h matches, the process proceeds to step S505, and the phrase pair likelihood value is set to “0” to suppress the phrase from becoming the first candidate.
[0049]
On the other hand, if it is determined in step S503 that the example is “normal type”, the fixed value 1012 is added to the phrase pair likelihood as the example application likelihood, and the process returns.
[0050]
Returning to FIG. 9, the details of the “next candidate display process” in step S105 will be described below according to the flowchart shown in FIG.
[0051]
First, in step S601, a homophone ID of a homophone for which the next candidate is to be displayed is acquired. In the next step S602, the next candidate having the same homophone ID in the homophone pool (DBP) 3g is extracted from the independent word JWP of the homophone pool and the attached rear row FWP, and the corresponding JW and FW in the homophone buffer (DBF) 3f are extracted. Store it in the field and display it.
[0052]
Details of the “selection process” in step S106 will be described below with reference to the flowchart shown in FIG.
[0053]
First, in step S701, the homophone ID of the homophone to be selected is extracted, and in the subsequent step S702, the pair homophone ID in the homophone buffer 3f having the extracted homophone ID is extracted. Then, in step S703, the independent word of the selection target homophone and the independent word of the pair homophone are extracted, and in step S704, this independent word pair is stored in the APYR field in the homophone buffer 3f. It is determined whether it matches the application example at the time of determination. The example shown by this APYR is acquired from the example dictionary 3b, and it is determined whether or not it matches with the independent word pair constituting this example. If the example shown in APYR is described by semantic classification, it is checked whether or not they match by comparing the semantic classification of independent words JW in the homophone buffer.
[0054]
If the example applied at the time of determining the first candidate is also established at the time of selection, it is determined that the registration process of the suppression example is unnecessary, and the process returns. If the examples do not match, the processing from step S705 onward is performed to register the application example as a suppression type.
[0055]
In step S705, the currently designated field information is extracted from the field designation flag (BFLG) 3h. This is to extract all fields in which the BFLG bit is “1”. Next, in step S706, the application example APYR of the target homonym buffer is registered in the example dictionary 3b together with the application field information extracted in step S705 as an example of “suppression type”. It is not applied when determining one candidate.
[0056]
<Modification>
In the above-described embodiments, the kana-kanji conversion device in which the input target sentence is limited to only kana characters is taken as an example of the character processing device that performs the phrase structure analysis, but the present invention is not limited to this. The phrase structure may be analyzed as long as it does not depart from the gist of the present invention, which performs processing for excluding counterexamples from analysis candidates based on counterexamples in which relations between phrases are registered. For example, as a character processing device that analyzes the phrase structure of a Japanese sentence, this is a document proofreading device that analyzes kanji structure, checks the semantic consistency of sentences, points out errors, and uses kanji-kana mixed sentences as input target sentences. It is also possible to carry out the invention.
[0057]
In the present embodiment, the example dictionary has two word (or semantic classification) pairs as a combination of words appearing in pairs. However, there are three word pairs or generally n word pairs. Even a pair (where n is an integer of 2 or more) can be processed in the same manner. Furthermore, the example dictionary for storing examples and the counterexample dictionary for storing counterexamples are the same dictionary, and it is determined whether the application type is an example or a counterexample. It is the same.
[0058]
Furthermore, in the present embodiment, the counterexamples already exist. However, the counterexamples are not prepared in advance, and may only be gradually accumulated by the operation of the operator. The counterexample is expressed by a pointer to the word dictionary of the word constituting the counterexample, but if it can identify the word constituting the counterexample instead of the pointer, it can be read, written, part of speech, etc. It may be stored with the attribute specifying the word, or a unique serial number is assigned to each word stored in the word dictionary, and the word constituting the counterexample is specified by the serial number. May be.
[0059]
Furthermore, by coexisting with an example to be applied instead of a counterexample, the analysis accuracy of the first candidate can be further improved.
[0060]
Moreover, although examples of eight fields such as “medical field” and “biological field” have been presented as field information that can be specified, the present invention is not limited to these eight fields. As the field information, not only the semantic group in which words, examples, and counterexamples are used, but also text forms such as “greeting text”, “article”, and “prose” may be specified. Even if it is set for each user such as the affiliation, the same processing can be performed.
[0061]
The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. It goes without saying that the present invention can also be applied to a case where the object is achieved by supplying a program to a system or apparatus.
[0062]
【The invention's effect】
As explained above, according to the present invention, a specific word combination is Can be given priority as a phrase candidate, Can be suppressed as a phrase candidate in a specific field, and not as a phrase candidate outside a specific field. Or can be suppressed regardless of the field Therefore, the conversion accuracy of the Japanese sentence to be input can be increased flexibly according to the field.
[0063]
In addition, according to the present invention, it is possible to minimize the memory capacity by registering unsuitable word pairs in association with information on fields to be suppressed without comprehensively registering all examples suitable as phrase structures. Therefore, it is possible to realize a character processing device with high sentence structure analysis accuracy of a Japanese sentence at a low cost.
[0064]
Furthermore, according to the present invention, word pairs that are not suitable as phrase structures can be registered in the semantic field, so that it is possible to improve the completeness of counterexamples that are not suitable and to further reduce the memory capacity.
[0065]
[Brief description of the drawings]
FIG. 1 is a block diagram showing an overall configuration of a character processing apparatus according to an embodiment.
FIG. 2 is a diagram illustrating an execution example of kana-kanji conversion according to the present embodiment.
FIG. 3 is a diagram showing a configuration of a word dictionary according to the present embodiment.
FIG. 4 is a diagram showing a configuration of an example dictionary of the present embodiment.
FIG. 5 is a diagram showing a configuration of a field designation flag according to the present embodiment.
FIG. 6 is a diagram illustrating a configuration of a sentence candidate table according to the embodiment.
FIG. 7 is a diagram showing a configuration of a homophone buffer of the present embodiment.
FIG. 8 is a diagram showing a configuration of a homonym pool of the present example.
FIG. 9 is a flowchart showing a processing procedure of the present embodiment.
FIG. 10 is a flowchart illustrating conversion processing according to the present exemplary embodiment.
FIG. 11 is a flowchart illustrating phrase candidate creation processing according to the embodiment.
FIG. 12 is a flowchart illustrating first candidate determination processing according to the embodiment.
FIG. 13 is a flowchart showing an example dictionary search process according to the present embodiment.
FIG. 14 is a flowchart showing a next candidate display process according to the embodiment.
FIG. 15 is a flowchart illustrating selection processing according to the embodiment.
[Explanation of symbols]
1 CPU
2 ROM
3 RAM
4 External memory
5 Keyboard
6 Cursor register
7 Display buffer
8 CRT controller
9 Display device
10 Character generator

Claims (3)

Word dictionary means for storing word reading and notation in association with each other;
The notation of the first and second words constituting the example, an example type indicating whether the example is to be prioritized or suppressed, and the specific field to which the example is to be applied are indicated. An example storage means for storing an example application field indicating that the field should be generally applied if there is no field to be applied;
An input means for inputting a kana character string;
Field designation information receiving means for receiving field designation information for designating a field to be converted with priority given to the character string input by the input means ;
Field designation information storage means for storing field designation information received by the field designation information reception means ;
Referring to the word dictionary means, dividing the character string input by the input means into phrase candidates, and phrase candidate creating means;
Search means for searching for a combination of words corresponding to the example stored in the example storage means among the phrase candidates created by the phrase candidate creation means;
An example type determination means for determining whether the combination of words found by the search means corresponds to an example to be prioritized or to be suppressed;
The application field of the example to which the combination of words found by the search means corresponds should be generally applied, or is a specific field specified by the field designation information stored in the field designation information storage means Field judgment means for judging whether or not the field matches,
When the example type determination unit determines that the example type is an example to be prioritized, the combination of words corresponding to the example is given priority as a phrase candidate, and the example type determination unit suppresses the type of the example. If it is determined to be a usage example, it should be generally applied by the field determination means, or if it is determined to match the specified field, a combination of words corresponding to the example is selected as a phrase candidate. And a phrase candidate determining unit that determines a phrase candidate without suppressing a combination of words corresponding to the example as a phrase candidate when it is determined that the specified field does not match. Character processing device characterized by.   The word dictionary means stores the word reading, the notation, and the semantic classification in association with each other, and the example storage means further includes a first word notation constituting the example, a second word semantic classification, , An example type that indicates whether the example should be preferred or suppressed, and the specific field to which the example should be applied, that field should be indicated; The character processing apparatus according to claim 1, wherein an example application field indicating the effect is stored in association with each other. Word dictionary means for storing word readings and notations in association with each other, notation of first and second words constituting the example, and an example type indicating whether the example should be prioritized or suppressed , An example storage means for indicating a specific field to which the example is to be applied and indicating the field, and indicating that it should be generally applied if there is no field to be applied; Input means for inputting a character string, field designation information storage means for storing field designation information for designating a field to be converted with priority given to a character string input from the input means, and a phrase for storing phrase candidates a candidate storing means, and processing means for executing various processes based on programs, that the program has a stored program storage means, and cooperating with said processing means and said program storing means Phrase candidate creating means implemented Te, search means, a processing method of a character processing apparatus comprising example type determination means, the field determining means, and the phrase candidate determination means,
A phrase candidate creating step in which the phrase candidate creating means divides the character string input from the input means into phrase candidates with reference to the word dictionary means, and stores the created phrase candidates in the phrase candidate storage means; ,
Said search means, a search step wherein among the stored phrase candidate phrase candidate storage unit, to search for combinations of words corresponding to the examples stored in the example storage means,
The example type determining means refers to the example type of the example to which the combination of words found by the search step corresponds from the example storage means, and determines whether the example type is an example to be prioritized or suppressed. Example type determination process;
The field determination means refers to the example application field of the example to which the combination of words found in the search step corresponds from the example storage unit, and the example application field should generally be applied or specified A field determination step for determining whether the field matches the field specified by the field specification information stored in the field specification information storage means;
When the phrase candidate determining means determines that the example type of the example is to be prioritized by the example type determining step, the phrase candidate determining unit prioritizes a combination of words corresponding to the example as a phrase candidate, and the example type determining step If the example type of the example is determined to be an example to be suppressed, the field determination step should generally apply, or if the example type is determined to match the specified field, the example If the combination of words corresponding to is suppressed as a phrase candidate and it is determined that it does not match the specified field, the phrase combination is determined without suppressing the word combination corresponding to the example as a phrase candidate. And a phrase candidate determination step for performing the processing.

Images