JPH04369763A - Kana(japanese syllabary)-kanji(chinese character) conversion device and its method - Google Patents

Abstract

PURPOSE:To accurately convert name having a homonym notation and to absorb the vibration of the difficult reading of a place name by providing a place name dictionary holding the hierarchy structure of the place name and a vibration correction table. CONSTITUTION:A dictionary part 105 is provided with the place name dictionary having information showing the hierarchy structure of the place name and information showing a connection relation between the hierarchies of the place names. A control part 103 reads an input character string obtained from an input part 101 and transfers it to a notation retrieval part 104. It refers to the vibration correction table 106 and retrieves the dictionary part 105 with the corrected vibration of reading as the object of retrieval. When the connection relation between the higher hierarchy place name and candidate notation agrees, first priority is added to the notation. When reading agrees but the connection relation with the higher hierarchy, place name, does not agree, second priority is added and it is transferred to the control part 103. It generates an analysis tree by viewing the result, and a display part 102 displays evaluation value in high order from the analysis tree.

Description

[Detailed description of the invention]

0001

[Industrial Field of Application] The present invention relates to a character processing device, for example, a kana-kanji conversion device, which has the function of converting an input reading character string into a desired character string. When converting to a place name,
The present invention relates to a highly accurate place name conversion method and device that can improve the conversion accuracy.

0002

[Prior Art] Conventional kana-kanji conversion is generally performed using a dictionary that combines readings, notations, and part-of-speech information, and analyzes words or sentences in light of Japanese grammar. be. In this Kana-Kanji conversion method, if the input and conversion target is an address, if you use general Kana-Kanji conversion, a place name such as ``Surugasisuru Goose (Suruga Town, Suruga City)'' will be changed to ``Print Starving Goose''. As a place name, it is often translated incorrectly, or incorrectly.

[0003] As for character input methods limited to inputting addresses, there are known methods that do not use kana-kanji conversion. Using a place name dictionary that maintains
There is an input method in which displayed place names are input hierarchically from the highest place name to the lowest place name.

0004

[Problem to be solved by the invention] When trying to input the kanji of a place name by converting it into kana-kanji using a general dictionary, if the place name has a homophonic notation (the same reading but different notations), the notation is incorrect. It cannot be determined in a single sense. For example, "sendai"
There are two spellings for this reading: ``Sendai'' and ``Kawauchi.'' In other words, because place names were previously treated as individual nouns, it was not possible to distinguish between homophones and convert them. As a result, there was the problem that place names such as ``Sendai City, Kagoshima Prefecture'' were converted into names that did not actually exist.

Another problem with inputting place names and addresses using the kana-kanji conversion method is that if the user misreads the place name, the correct kanji may not be obtained. There are many place names that have special pronunciations, or that can be read in many different ways, and it is easy for a user to input the pronunciation incorrectly, resulting in a mismatch with the place name dictionary. If the pronunciation of place names is narrowed down to one of the original pronunciations, users will not be able to obtain the desired spelling because they do not know the correct pronunciation of the desired place name. When inputting text, it may not be converted correctly due to small reading fluctuations.

The present invention was made in view of the above-mentioned problems, and its object is to provide a kana-kanji conversion method and device that performs conversion that distinguishes homophones of place names.

Another object of the present invention is to provide a method and apparatus for converting Kana to Kanji that absorbs fluctuations in reading and performs conversion.

[0008]

[Means for Solving the Problems] In order to solve the above problems, the present invention provides a place name dictionary that maintains a hierarchical structure in addition to the pronunciation and notation of place names, and examines the hierarchical connections of place names at the time of conversion. , made it possible to correctly convert homophonic place names.

[0009] In another aspect of the present invention, a sway correction table for converting the input pronunciation to another reading is provided, and when performing pronunciation matching with a dictionary, a match search is performed while referring to the sway correction table. This made it possible to absorb fluctuations in reading.

[0010] Furthermore, by providing means for storing the conversion results by hierarchy each time a kana-kanji conversion is performed, and by referring to the stored contents at the time of next conversion, priority is given to the place name notation that is subordinate to the recently used higher place name. made it possible to convert.

[0011] In addition, the configuration is such that a place name dictionary and a dictionary other than place names can be referenced at the same time, and when the conversion result of the input pronunciation string is converted to the lowest place name, the remaining pronunciations are converted after that. It is now possible to search dictionaries for words other than place names and create candidates.

0012

[Operation] The operation of the present invention will be explained using the block diagram shown in FIG.

The pronunciation of the place name inputted from the input section 101 is passed to the control section 103 . The control unit 103 cuts out one or more characters from the beginning and passes them from the control unit 103 to the notation search unit 104. The notation search unit 104 searches the dictionary unit 10 for the notation for the reading passed from the control unit 103.
5 and returns the matching expression and its evaluation value to the control unit 103. The evaluation value will be described later. The control unit 103 grows an analytic tree based on the matching notation and evaluation value passed from the notation search unit 104, compares the evaluation of each route, and displays the routes as conversion candidates in the order of the highest evaluation value on the display unit 102. pass it on to

The evaluation values mentioned above will now be described. The notation search unit 104 searches the dictionary unit 105 for notations corresponding to the pronunciations, and also determines an evaluation value for each notation based on the connection relationship with the previous clause. Control unit 103
A comparison between the analytic tree formed in 1 and the hierarchical data for each notation obtained from the dictionary section 105 is performed as follows. (1) If the connection relationship (hierarchical relationship) between the upper layer place name and the candidate notation matches, the control unit 10 assigns the first priority to the notation.
Pass it to 3. (2) If the pronunciations match but the connection relationship with the upper layer place name does not match, the second priority is given and the name is passed to the control unit.

When the notation search unit 104 performs a notation search in the dictionary unit 105 , it refers to the fluctuation correction table 106 held to absorb reading fluctuations, and determines whether the passed reading is the subject of the fluctuation correction table 106 . If the dictionary contains a pronunciation, the dictionary section 105 is searched for the pronunciation with the fluctuation corrected as a dictionary search target.

[0016]

[Embodiment] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 2 is a diagram showing the hardware configuration of an embodiment of the kana-kanji conversion device of the present invention. In the same figure, 20
1 is a keyboard for inputting readings, 202 is a display for displaying conversion results, 203 is a CPU for controlling ephemeral-kanji conversion, and 204 is a memory for storing data created by the CPU 203 and place names for each layer of conversion results. , 205 and 206 are dictionary units that calculate kanji notation based on the readings input from the keyboard, 204 is a ROM dictionary (common dictionary), and 206 is a floppy dictionary (purpose-specific dictionary).
It is. The floppy dictionary 206 consists of a place name dictionary 207 and a person name dictionary 208, each of which has a user registration area 209,
I have 210. However, place name dictionary and person name dictionary are FD.
It is not limited to a dictionary, and it is also possible to have it as a ROM.

In the basic block diagram of FIG. 1, the present invention is roughly divided into three steps. The user inputs the pronunciation of the address through the input unit 101 . The control unit 103 reads the input character string obtained from the input unit 101 and passes it to the notation search unit 104 (first means). The notation search unit 104 searches the dictionary unit 105 while referring to the shake correction table 106 (second means). The control unit 103 creates an analysis tree while looking at the results (third means). As a result, the display unit 102 displays the analysis tree in descending order of evaluation value.

FIGS. 3 and 4 regarding the first means mentioned above.
The explanation will be based on. FIG. 3 is a flowchart illustrating character string extraction for matching search from input pronunciation, and FIG. 4 is an analysis example for explanation.

In order to extract a pronunciation string to be matched with the dictionary from the input character string 401, one pointer P (402 in FIG. 4) and one variable N (403 in FIG. 4) are set. The pointer P moves over the input character string and points to the number of first characters to be cut out. Variable N holds the number of characters cut out from the pointer. Initially, pointer P points to the beginning of the input character string. The number of characters N is kept at 1 (step 301). Pass N characters from pointer P to the notation search unit (step 302).

[0021] At this time, the notation retrieval unit that has received the search pronunciation sequence compares the received pronunciation with the shake correction table shown in FIG. The structure of the shake correction table is that one record is made up of two fields: a first reading string for searching, and one or more second reading strings that are likely to cause reading shake, and shake correction is performed. A table is a collection of records. For example, the search pronunciation for "Sawa" is "Zawa", and conversely, the search pronunciation for "Zawa" is "Sawa", "
``Machi'' corresponds to ``chō'', ``chō'' corresponds to ``machi'', and ``hara'' corresponds to ``bara'' and ``wara''. Compare the received reading sequence with the shake correction table. In other words, the first reading sequence of the shake correction table and the dictionary search reading sequence are matched sequentially, and if they match, the place name dictionary is searched for a match, including the rearranged place name in the second reading sequence. . For example, if the input pronunciation sequence is "Yoshi Ostrich", the last pronunciation sequence 803 matches the first pronunciation sequence 803 of the shake correction table. Therefore, the place name dictionary is also searched for "Yoshidamachi", which is a rearrangement of the second pronunciation sequence 802 "Machi" corresponding to "Chou". In other words, by inputting "Yoshi Ostrich", it is possible to perform a dictionary search for both "Yoshi Ostrich" and "Yoshidamachi".

[0022] Check whether there is a matching character string (
Step 303), and registration is performed based on the evaluation value (Step 304). Registration using evaluation values will be described later. Next, the variable N
1 is added to (step 305), and if the number of characters is not exceeded (step 306), the reading string to be searched is increased by one character and the dictionary search is performed again. If the number of characters is exceeded,
Set pointer P to the next cutting position of the analysis path (
Step 307). Here, the next extraction position is the sum of the character strings of place names in a certain analysis path that have been analyzed, and is the analysis end position of the path at that point. If the set position exceeds the input string, the first candidate is displayed and the process ends.

Next, the above-mentioned second means will be explained based on FIG. 5. FIG. 5 is a flowchart showing the processing procedure for notation search. The reading character string obtained from the control unit is searched for a match with the place name dictionary (step 502). If there is no match here, an evaluation value of "no candidate" is given and the notation search ends (step 503). If there is a matching place name, it is extracted as a candidate place name. Next, check whether the pointer P is at the beginning of the input string (step 504).
), if it is the first, the hierarchical structure cannot be used, so an evaluation value of "priority 2" is given, and all candidate place names are handed over to the analysis unit in the order of registration. (Step 505). If it is not the first place name, the immediately preceding place name is referred to as a higher place name to check whether or not the candidate place names can be combined. The possibility of joining is checked using parameters indicating hierarchical information, which will be described later.As for the parameter of the immediately preceding place name, the parameter received and placed at the same time as the reading is received in step 501 is taken out (step 506). Furthermore, the hierarchical information of the candidate place name is obtained (step 507), and the hierarchical information of the immediately preceding place name is compared with the hierarchical information of the candidate place name (step 508). A method for comparing the hierarchical information between the immediately preceding place name and the candidate place name will be described later. As a result of the comparison, if an inclusion relationship is established between the immediately preceding place name and the candidate place name, then
Give an evaluation value of "Priority 1". An evaluation value of "priority 3" is given to candidates whose pronunciations match although the inclusion relationship does not hold. Hierarchical information is compared for all conversion candidates with matching pronunciations, a priority evaluation value is attached, and the information is returned to the control unit.

Next, a method of comparing the hierarchical information of the immediately preceding place name and candidate place name will be explained with reference to FIGS. 6 and 7. FIG. 6 shows the place name dictionary data structure in the place name dictionary 207, and FIG. 7 is a processing flowchart of hierarchical information.

[0025] Hierarchical information is a data structure using coordinate parameters and range parameters. In this embodiment, a case will be explained in which there are two coordinate parameters and one range parameter. In Figure 6, the coordinate parameters are 601 to 6.
These are the two parameters enclosed in parentheses in 08,
The range parameter is one parameter to the right of the parentheses 601 to 608. The range parameter is set according to the hierarchical level of the place name, giving a larger value to the higher the hierarchy. At the lowest level, a range is not necessary and it is sufficient to indicate that it is the lowest level, so 0 is given as the range parameter. The coordinate parameters are set so that they fall within the range of "coordinate parameters ± range parameters" of the next higher place name. These three parameters are used as hierarchical information of place names. At this time, by adjusting the coordinate parameters and range parameters, it is also possible to set the range parameters so that the hierarchy of administrative divisions can be identified. For example, the range parameter "10" is set as a prefecture, and the range parameter "3" is set as a group of cities.

For example, after "Miyagi Prefecture", "Sendai"
is input, and the conversion candidates 603 "Sendai" and 604 "Kawauchi" exist in the dictionary, the hierarchy information of Miyagi prefecture "
(100, 50), range from 10'' to (90-110, 4
0 to 60) are suitable as lower place names. Therefore, the conformance description in this case is “
It turns out that it is Sendai.

The above hierarchical information comparison process will be explained with reference to FIG.

[0028] The positional parameters (x, y) and range parameter h of the upper place name, and the positional parameters (X, Y) of the candidate place name.
) is obtained (steps 701, 702). Therefore, the condition ``(
x-h≦X≦x+h) and (y-h≦Y≦y+h)” (
When step 703) is true, an evaluation value of priority 1 is given as an inclusion relationship holds between the immediately preceding place name and the candidate place name (step 705), and an evaluation value of priority 3 is given when the condition does not hold (step 704). , completes the comparison of hierarchical information.

Next, regarding the third means mentioned above, FIG.
The explanation will be based on.

Based on the result of the notation search, the control section creates an analysis tree. The searched expressions are 404 to 408 along with the evaluation value.
It is stored as a tree structure as shown in . Hereinafter, how to create an analytic tree will be specifically explained based on FIG. 4.

In the case of the character string 401 in FIG. 4, "chi" is passed to the notation search section first, but in this case it does not exist in the dictionary, and in the case of the character string 401, "chiba" is passed. In the end, the string 401
In this case, "Chibaken" has a priority of 2, so it is registered as a path in the parse tree. When n increases until it exceeds the number of input character strings, move the pointer to the next character that has been analyzed (in the case of character string 401, up to the 4th character has been analyzed, so the 5th character), and similarly increase N by 1. Perform a notation search while incrementing the number one by one. After matching "Chibaken", the pointer points to "na" after "Chibaken". After that, a notation search is performed from "na" and "Narashi" and "Narashinoshi" are obtained, but "Narashi" is not hierarchically compatible and has a priority of 3, while "Narashinoshi" is hierarchically compatible and has a priority of 1. . If the pointer reaches the end of the string, priority 1
and priority 2 are converted into points and displayed in order of score.

[0032] Use the pronunciations of “ken” and “shi” to cut out words.
Since using ``ken'' or ``shi'' is an effective means of extraction, it is also possible to perform an efficient notation search by using ``ken'' or ``shi'' as delimiters for place names.

[0033] As in the case of the character string 401 in Figure 4, even after the conversion to the lowest place name (Oaza/town level) has been completed using the place name dictionary, if the input pronunciation remains, the lowest place name After the place name, it is often an apartment name or company name, so it is not appropriate to convert it using a place name dictionary. If the lowest place name ("Okubo" in the case of character string 401) is registered in the tree by conversion using the place name dictionary, if the input pronunciation remains (in the case of character string 401 "Ikeda Apart"), it will be automatically changed. Performs processing to switch the dictionary used for conversion. After switching the dictionary, standard conversion is performed using the built-in ROM dictionary (205) to convert sentences other than place names into kanji and kana mixed sentences. At this time, candidates that have undergone built-in standard conversion are given a priority of 1 as an evaluation value.

When displaying the conversion results, an analytic tree and evaluation values assigned based on the relationship between place names are used. The evaluation value is 3 points for "Priority 1", 2 points for "Priority 2", and 2 points for "Priority 3".
'' is converted into one point, and the evaluation value is accumulated for each path from the root to the leaf of the analysis tree, and the path with the highest cumulative evaluation value is selected as the first candidate. In the example in Figure 4, two analysis paths are generated, but "Nara City, Chiba Prefecture" has a priority of 2 (2
points) and priority 3 (1 point), the cumulative evaluation value is 3 points. Additionally, the cumulative evaluation value of "Okubo Ikeda Apartment, Narashino City, Chiba Prefecture" is 9 points. After calculating the cumulative evaluation value for all paths, the path with the largest cumulative evaluation value is selected as the first candidate. In the case of the example in FIG. 4, "Okubo Ikeda Apartment, Narashino City, Chiba Prefecture" is the first candidate.

[0035] If the user desires a place name in kanji other than the first candidate, the next candidates are displayed in sequence by instructing to display the next candidate using the input means, and when the desired candidate is displayed, the user issues a confirmation instruction. The conversion result can be obtained.

Furthermore, each time the user confirms the result of the kana-kanji conversion, the place name resulting from the conversion is stored in the memory 204 in a hierarchical manner. As a result, the next time there is an input without a higher-level place name, that is, if a higher-level place name is omitted, the higher-level place name stored in memory will be referred to and the place name belonging to that place name will be given priority. It can be output. For example, when "Ishinomaki City, Miyagi Prefecture" is determined by the user after kana-kanji conversion, place name information (coordinate parameters and range parameters) for each hierarchy is stored in the memory as shown in FIG. Next, when the pronunciation of "Sendaishi" with the place name omitted in the prefecture hierarchy is input, by referring to the coordinate parameters of 901 "Miyagi Prefecture" stored in memory, the hierarchy information mentioned above is Using a comparison algorithm, "Sendai City" (Miyagi Prefecture) is given priority over "Kawauchi City" (Kagoshima Prefecture) as a candidate.

[0037] This is particularly effective when addresses within a relatively small area are input consecutively, for example, when inputting addresses that do not require high-level place names, such as in a directory of a neighborhood association.

[0038]

As described above, according to the present invention, when a user inputs an address using kana-kanji conversion, even place names with homophones can be converted into the correct spelling.

Furthermore, when inputting a place name in kana whose correct pronunciation is unknown, the kana-kanji conversion system absorbs the slight fluctuation in the pronunciation, making it possible to input the name without making the user aware of it.

Furthermore, even if the input of the upper place name is omitted, the input operability is improved by implicitly referencing the upper place name. Furthermore, if there is an input reading sequence remaining after the lowest place name, conversion will be performed using a dictionary other than place names, so enter the apartment name, company name, school name, etc. after the address and convert at once. This improves user operability.

[Brief explanation of drawings]

[Figure 1] Basic block diagram,

[Fig. 2] Hardware configuration diagram of one embodiment,

[Figure 3] Character string extraction processing flowchart,

[Figure 4]
Examples of place name cutting and analysis trees,

[Figure 5] Processing flowchart of notation search,

[Figure 6] Example of place name dictionary data structure,

[Fig. 7] Flowchart of comparison processing of hierarchical information,

[Figure 8]
A diagram showing a shake correction table,

FIG. 9 is a diagram showing a conversion result storage area.

[Explanation of symbols]

101...Character input device, 102...Display device, 103...Control unit, 104...Notation search unit, 105...Dictionary unit, 106...
Shake correction table.

Claims (12)

[Claims] Claim 1: An input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing the kana-kanji conversion, a kana-kanji conversion means for converting the reading character string to kana-kanji, and the reading character string. and an output means for outputting a candidate string that is a result of converting the reading character string to kana-kanji, the kana-kanji conversion device includes: first information indicating a hierarchical structure of place names; and first information indicating a joining relationship between hierarchies of place names. 2 information for performing kana-kanji conversion of place names.
A kana-kanji conversion device characterized by being equipped with a dictionary. [Claim 2] Determining that among the candidate strings converted into kana-kanji by the kana-kanji conversion means, a candidate string whose relationship between the candidates matches the second information of the second dictionary is the first candidate string to be output. Claim 1 characterized in that it comprises means.
The Kana-Kanji conversion device described. 3. The second dictionary has one or more coordinate parameters representing a position, and one or more coordinate parameters representing a certain range based on the position represented by the coordinate parameter. 2. The kana-kanji conversion device according to claim 1, wherein the range parameter represents a hierarchical structure of place names. Claim 4: An input means for inputting a reading character string to be subjected to kana-kanji conversion; a first dictionary for performing kana-kanji conversion; first information indicating a hierarchical structure of place names; a second dictionary for performing kana-kanji conversion of a place name, which has second information indicating a conjunctive relationship; a kana-kanji conversion means for converting a reading character string into kana-kanji using the dictionary; means for determining, as a first candidate string to be output, a candidate string whose relationship between the candidates matches the second information of the second dictionary among the converted candidate strings; This kana-kanji conversion device is equipped with an output means for outputting a candidate string as a result of kana-kanji conversion, and is characterized by comprising a storage means for storing place name candidates selected by an operator in each hierarchy. 5. If the candidate string converted into kana-kanji based on the first dictionary and the second dictionary is a candidate string that cannot be determined by the determining means, the relationship with the place name candidate stored in the storage means is 5. The kana-kanji conversion device according to claim 4, further comprising second determining means for determining a candidate string that matches the second information of the second dictionary as the first candidate string to be output. [Claim 6] An input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing kana-kanji conversion, and a connection between first information indicating a hierarchical structure of place names and the hierarchy of place names. a second dictionary of place names having second information indicating a relationship; a kana-kanji conversion means for converting the reading character string to kana-kanji; and a candidate string that is the result of converting the reading character string and the reading character string to kana-kanji. A kana-kanji conversion method, characterized in that the kana-kanji conversion device is equipped with an output means for outputting, and converts the reading character string into kana-kanji using the first dictionary and the second dictionary at the same time. Claim 7: Input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing kana-kanji conversion, and first information indicating a hierarchical structure of place names and a connection between the place name hierarchy. a second dictionary for performing kana-kanji conversion of place names having second information indicating a relationship; a kana-kanji conversion means for converting the reading character string to kana-kanji using the first dictionary and the second dictionary; In a kana-kanji conversion device comprising: a determining means for determining the output order of a candidate string that is a result of converting a reading character string to kana-kanji; and an output means for outputting the reading character string and the candidate string, a first kana-kanji conversion means. Kana-Kanji conversion characterized by determining a candidate string whose relationship between candidates matches the second information of the second dictionary from among the candidate strings converted to Kana-Kanji by Kana-Kanji conversion, and setting the candidate string as the first candidate string to be output. Method. [Claim 8] An input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing kana-kanji conversion, and a connection between first information indicating a hierarchical structure of place names and the hierarchy of place names. a second dictionary of place names having second information indicating a relationship; means for converting a reading character string into kana-kanji using the first dictionary and the second dictionary; and a candidate string converted into kana-kanji by the kana-kanji conversion means. Of these, the relationship between the candidates is the second
determining means for determining a candidate that matches the second information of the dictionary and setting it as a first candidate string to be output; and an output for outputting the reading character string and a candidate string that is the result of converting the reading character string into kana-kanji. 1. A kana-kanji conversion method, characterized in that the kana-kanji conversion device comprises a storage means for storing place names for each hierarchy, and a kana-kanji conversion method, characterized in that place name candidates selected by an operator are stored in the storage means for each hierarchy. Claim 9: Input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing kana-kanji conversion, and first information indicating a hierarchical structure of place names and a connection between the place name hierarchy. a second dictionary of place names having second information indicating a relationship; means for converting the reading character string into kana-kanji using the first dictionary and the second dictionary; and kana-kanji conversion by the kana-kanji conversion means. a first determining means that determines a candidate among the candidate sequences whose relationship matches the second information of the second dictionary and sets the candidate as a first candidate sequence to be output; and a place name selected by an operator. A storage means for storing place names in each layer of candidates, and a first candidate string to output a candidate string whose relationship with the place name candidates stored in the storage means matches second information of a second dictionary. In the kana-kanji conversion device, the kana-kanji conversion device includes a second determination means for determining the character string as determined by the first determination means; A kana-kanji conversion method characterized in that, in the case of a candidate string that cannot be used, a second determining means is used. Claim 10: An input means for inputting a reading character string to be subjected to kana-kanji conversion, a first dictionary for performing kana-kanji conversion, and first information indicating a hierarchical structure of place names and a connection between the place name hierarchy. a second dictionary for performing kana-kanji conversion of place names having second information indicating a relationship;
means for converting a reading string into kana-kanji using a dictionary;
In a kana-kanji conversion device comprising the above-mentioned reading character string and an output means for outputting a candidate string that is a result of converting the reading character string into kana-kanji, the reading character string is A kana-kanji conversion method characterized in that when a string is converted to the lowest place name hierarchy, the next consecutive reading character string is converted using a first dictionary. 11. An input means for inputting a reading character string to be subjected to kana-kanji conversion, a dictionary for performing the kana-kanji conversion, a kana-kanji conversion means for converting the reading character string to kana-kanji using the dictionary, and a reading character string. and an output means for outputting a candidate string that is a result of converting the reading character string into kana-kanji, in a kana-kanji conversion device, converting a first reading character or character string into a second reading character or character string. A kana-kanji conversion device characterized by having a shake correction table. 12. Input means for inputting a reading character string to be subjected to kana-kanji conversion; and a first reading character or character string;
a swing correction table for converting into a second reading character or character string; a dictionary for performing kana-kanji conversion; a kana-kanji conversion means for converting the reading character string to kana-kanji using the dictionary; and the reading character string and the reading character string. and an output means for outputting a candidate string that is a result of converting the inputted first reading character or character string into a different second reading character or character string using the above-mentioned fluctuation correction table. A kana-kanji conversion method comprising converting the converted second reading character or character string into a desired character using a kana-kanji conversion means.