JP2766084B2 - Kana-Kanji conversion method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kana-kanji conversion device for converting a kana input into a kanji-kana mixed sentence, and more particularly to a kana-kanji conversion device in which the input is not broken down into clause units. The present invention relates to a kana-kanji conversion device suitable for so-called solid input, which can also convert characters. 2. Description of the Related Art Conventionally, as a method of converting kana to kanji characters in solid writing, Hitachi Review, May 1981, Vol. 63, No. 5, "HITAC L-320 / 30H, 50H Document Processing Function" Ref. (1)), the method of combining the longest matching backtracks described in Ref. (1), 1993 19th Information Processing Society of Japan 19th Annual Convention Transactions 5E-4, "Kana-Kanji Conversion System for Solid Writing Sentences" (Referred to as (2)), the two-phrase longest-matching method described in 1981, Information Processing Society of Japan, Computational Linguistics Research Group Material 25-6, "Phrase Structure Analysis Algorithm Using Table Method and Its Efficiency" The term number minimization method described in (3) is well known. The method of Document (1) cuts out an independent word from the input left end by the longest match, and then repeats a phrase conversion process of assigning a grammatically connectable attached word to the independent word until it reaches the right end. If the right end is not reached, backtracking (backward) is performed and another phrase conversion is attempted to proceed to the destination. In the method of reference (2), all possible conversion candidates are brute-force-extracted over two phrases from the left end of the input, and a candidate having the longest sum of the lengths of the two phrases is selected from among them.
This is a method in which a break is set as a break of a phrase, and the same is repeated starting from that point to determine a break for each phrase. The method of Document (3) is a method in which phrases are cut out from the left end in a brute force manner, and a combination that minimizes the number of phrases is selected from the cutouts to obtain a conversion result. [0006] In the case of solid writing, since there is a choice of where to make a break between phrases, it is generally necessary to use a conversion method in comparison with the case of phrase segmentation. Many candidate ambiguities occur. For example, there are many interpretations of the input "in the context of mathematics,""in the transfer of mathematics,""in the mathematical society,""in mathematics or archeological sites," and so on. It is possible. Therefore, there is a problem that it is difficult to obtain high conversion accuracy even by using the methods described in Documents (1), (2), and (3). An object of the present invention is to provide a means for solid writing input for efficiently extracting and holding a plurality of conversion candidates having a high likelihood of conversion certainty, and easily and quickly selecting and confirming a correct conversion result among them. To provide. [0008] To achieve the above object,
The kana-kanji conversion method of the present invention includes a plurality of phrases,
Enter a kana character string that is not separated in clause units
Character string input means and cut out multiple phrases from kana character strings
And at least one variable starting with the first character of each clause
Kana-Kanji conversion means for extracting Kana-kanji candidates and performing Kana-Kanji conversion
And a kana-kanji conversion device having display means.
Kana-Kanji conversion method for Kana-Kanji conversion
Of the conversion candidates extracted according to the input indicating
Consists of conversion candidates with the highest probability for each clause
Display candidate character strings on the display
The conversion candidate corresponding to the phrase to be displayed is displayed so as to be distinguishable from others,
Depending on the input that changes the segment break of the target phrase,
For the part of the name string after the changed clause break
Display conversion candidates obtained by re-evaluating the likelihood of conversion
This is displayed on the means. As described above, by selecting and holding a candidate having a high possibility, it becomes possible to display and select another candidate at high speed even if a conversion error occurs. Requires the least equipment. If a conversion error is corrected by selecting another candidate, not only that part but also subsequent candidates are re-evaluated. For example, if the position of the boundary between candidates changes, Subsequent candidate strings starting from the new delimiter position are prepared, and the number of subsequent corrections can be reduced. An embodiment of the present invention will be described below. FIG. 1 is a diagram showing the configuration of a kana-kanji conversion device according to the present invention, which comprises an input / output unit 1, a control unit 2, and a storage unit 3. The input / output unit 1 includes a keyboard for inputting kana input, conversion and selection instructions, and a display unit for displaying conversion results and selection candidates. The control unit 2 controls execution of kana-kanji conversion. The storage unit 3 temporarily stores an input kana character string and data used for conversion. FIG. 2 shows a method of executing conversion using the present kana-kanji conversion apparatus. The user inputs a kana character string from the input / output unit 1 and instructs conversion (10). The control unit 2 performs a conversion process (20) for creating a candidate having a high likelihood of conversion certainty and storing the candidate in the storage unit 3 with respect to the kana input. Next, the control unit 2 creates an alternative candidate to be selected corresponding to the candidate string having the highest likelihood and the first candidate from the candidates in the storage unit 3 (30), and causes the input / output unit 1 to display the alternative candidate. (40). The user selects a target candidate from the selection candidates (50).
Then, the control unit 2 determines the selected portion and learns the selected candidate (60). Next, the control unit 2 performs a display candidate creation process (30) again on the undetermined portion, and creates and displays a conversion result and a replacement candidate for the subsequent portion. The above display (40), selection (50), determination processing (60), and display candidate creation processing (30) are repeated until all the conversion results are determined. Hereinafter, the operation of the control unit 2 will be described in more detail. FIG. 3 is a flowchart showing the operation of the conversion process (20). FIG. 4 is a diagram showing an example of data created on the storage unit 3 by the conversion process (20).
The conversion process (20) of FIG. 3 will be specifically described using the example shown in FIG. Now, it is assumed that the input character string is "in the world". 3 and FIG.
n is a pointer indicating the position of the character counted from the beginning of the input character string. In step 201 of FIG. 3, n is set to 1 and positioned at the head of the input character string. In step 202, it is checked whether or not there is a clause end where the position of n ends. It is assumed that the beginning of the character string is specially regarded as a clause end. In step 203, phrase conversion is performed. In the phrase conversion, a conversion candidate capable of performing one phrase conversion is extracted with the position of n as the head of the phrase. Here, the phrase refers to a form in which one independent word is preceded by an optional prefix, an optional suffix can be added after the independent word, and an optional appendix is connected. Compound words are considered clauses. In the example of FIG. 4, "math", "number", "suck", "mathematical society", "mathematic world",
“Mathematical floor”, “number”, “sucks”, etc. are extracted as conversion candidates from the top. As the kana-kanji conversion in phrase units, the method shown in "Kana-Kanji Conversion by Computer" in NHK Technical Research Vol. 25, No. 5, May 1973 is well known. Next, in step 204, the likelihood of the probability of the conversion candidate is determined, and it is determined whether the conversion candidate should be stored in the storage unit 3 or discarded (referred to as skill cutting). The likelihood of certainty is basically, as described in Refs. (1), (2), and (3), that the input character string is a sequence of longer phrases, or, in other words, fewer phrases. Is determined so that the likelihood is higher when the image is decomposed into columns. However, in the same phrase "this"
Adnominals such as "no" and formal nouns such as "koto" and "mono" are often attached to other phrases, so they are not considered as independent phrases like nouns or verbs. When counting the number of clauses, the value should be smaller than 1. As described above, the number of phrases is obtained by multiplying the weight in consideration of the part of speech and the appearance frequency, and the smaller the number is, the higher the likelihood is. Specifically, nouns, verbs, adjectives, adjective verbs are given a weight of 1, formal nouns, auxiliary verbs, adverbs, etc. are given a weight of 0.1, and prefixes and suffixes are given a weight of 0.5 as semi-independent words And In pruning, the likelihood from the beginning of the character string to the end of the phrase to be determined at present is determined, and this is defined as the likelihood at the character position at the end of the phrase. If the likelihood is found in the above, the value is compared with the value, and if the value exceeds a certain allowable value, pruning is executed. In this embodiment, the allowable value is the minimum value of the number of weighted phrases at the same character position + 1. However, the present invention
Needless to say, the method is not limited by the method of determining the likelihood of the certainty and the size of the allowable value of the pruning. A feature of the present invention is that by extracting all phrase candidates within a certain range of likelihood of certainty and storing them in the storage unit 3, later selection and correction can be easily performed at high speed. In FIG. 4, all the clauses extracted from the head are stored in the storage unit 3. (Step 205).
In the case of storing, by arranging data in a network form as shown in FIG. 4, the occupied storage capacity can be reduced. For example, “number” and “suck” can share the right end of a phrase, thereby unifying subsequent phrases. A specific method of expressing such data is well known as list processing. In step 206 of FIG. 3, 1 is added to n to move the pointer to the next input character position. At step 207, it is determined whether or not there is any input at the character position of the pointer n. When n is 2, since there is no phrase right end at this character position, the process goes to step 206, and the pointer n is immediately moved to the next character position. When n is 3, the end of the phrase "number" or "suck" is possible, so the next phrase is extracted from this position in step 203, and "forehead", "easy", "forehead",
A phrase such as "forehead meeting" or "forehead department" can be obtained. Step 2
In 04, “mathematical” has already been obtained as a candidate at the position where n is 4, and the number of weighted phrases is one. "amount",
In the phrase candidate “easy”, the number of weighted phrases at the position where n is 4 is 2 by adding 1 of “number” and 1 of “amount” or “easy”. From the pruning conditions, “forehead” and “easy” fall within the allowable range, and are registered in the network in step 205. As described above, by repeating steps 202 to 207 until n becomes 14, the complete network shown in FIG. 4 is obtained. However, here, "meeting, world, floor, family, formation, upper, place, state" is a suffix. In step 208, a pointer P indicating the head position of the undetermined part of the conversion result is initialized to 1, and
We show that there is no final part yet. As described above, the conversion processing (2
After the operation of (0) is completed, the process proceeds to the operation of the display candidate creation process (30). FIG. 5 is a flowchart showing the operation of the display candidate creation process (30). In step 301, a candidate string having the highest likelihood among the candidate strings from the point P to the end of the input character string is set as a conversion result in the display buffer. If there are a plurality of candidate strings having the same likelihood, a learning table in the storage unit 3 which holds a compound word or an independent word with an affix from the already determined character string is referred to. Is selected as the conversion result.
If the above is not uniquely determined, the candidate string having the largest sum of the independent word lengths is selected. In addition, as a method of selecting a conversion result, there is a method of using the frequency of a word or the like, and another selection method can be created by arbitrarily combining the factors described above. In step 302, another candidate group starting from point P is set in the display buffer so that the user can select it. In the example shown in FIG. 4, when P is 1, "on the mathematical analysis" is selected as the conversion result, and "mathematical", "number", and "suck" are extracted as candidate groups for selection. Set in the display buffer. Further, the conversion result and the corresponding portion of the input character string are highlighted to clearly indicate the character string portion currently being selected. (Display (4
0)). FIG. 7 shows screen information displayed on the input / output unit 1 immediately after the input (10) is completed, and FIG. 8 shows screen information displayed after the display candidate creation processing (30) is completed. In this screen, the bottom is an input line, the top is a conversion result output line, and the bottom right is a selection candidate display area. The user looks at the conversion result on the screen and selects (5)
Perform 0). The selection is performed by using a keyboard to select an object from the selection candidate group. If the conversion result highlighted at the top of the screen is correct, it can be selected by pressing a specific key. is there. When the selection (50) is performed by the user, the control unit 2 next executes the operation of the confirmation processing (60). FIG. 6 is a flowchart showing the operation of the determination process (60). At step 601, a mark is given to the network data in the storage unit 3 corresponding to the candidate designated by the user, and the length of the selected candidate is added to P, and the head of the subsequent undetermined part is added. Indicates the position. In step 602, the selected candidate is read and stored in the learning table on the storage unit 3. The control unit 2 starts from the point P and the subsequent conversion result obtained by re-evaluating the likelihood of the network of the candidate selected in the determination processing 60 by the display candidate creation processing 30 and the candidate after the point P. Set another candidate group in the display buffer. FIG. 9 shows screen display information after the user selects "math". Highlighting is "Analysis" after "Mathematics"
Has moved to the part. "Analysis" as a selection candidate group,
"Meeting", "Kai", "Floor", "Department", "Kake", and "Ka" are displayed. When the user selects "meeting" instead of "analysis" from the candidates, as shown in FIG. 10, the conversion result after "meeting" is obtained from the network shown in FIG. Are automatically changed to "Choice", "Seki", and "Cough" are displayed as a selection candidate group. Since the adjacency between the candidates is held in this way, the segment break is selected (5
In the case where the display is changed according to (0), the subsequent display can be corrected at the same time, and the number of steps for subsequent correction can be reduced. When the user selects "meeting",
Although the learning table is also updated in the confirmation process (60), in this case, since "mathematics" has already been registered and "kai" is a suffix, "mathematics" has been added to the previously registered "mathematics". Association is added and re-registered in the form of a "mathematical society".
FIG. 11 shows the contents of the learning table after "Mathematical Society" is registered. The control unit 2 repeatedly executes the display candidate creation processing (30) and the determination processing (60) until there is no undetermined part. When all the conversion results have been determined, the process ends. When the above-described series of processing is completed, the network data in the storage unit 3 is erased, but the data in the learning table is retained. If the same character string is subsequently input, the conversion result is given priority. Therefore, the conversion accuracy can be increased according to use. In the past, the learning function associated with conversion was used, but even if the prefix and suffix were learned by themselves, the same objection of the same voice often appeared in the same sentence, but rather reduced the conversion accuracy There were many problems. In addition, when a homonym appears, such as "news reporter is on a train", it was not possible to distinguish between "reporter" and "train" because it was learned in the unit of reading "kisha". . According to the present invention,
Prefixes and suffixes are learned together with the independent words to which they are attached, and compound words are learned in long units without being decomposed into independent words, so that there is an effect that the accuracy of learning can be improved more. According to the present invention, since candidates having a high likelihood of conversion certainty are held, another candidate can be immediately selected even if the conversion result has an error. This has the effect of speeding up the correction. By performing the processing for creating and holding the candidates in parallel during the input operation of the user, it is possible to further reduce the conversion waiting time of the user.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a configuration diagram of a kana-kanji conversion device according to the present invention. FIG. 2 is a diagram illustrating a method of performing conversion. FIG. 3 is a flowchart illustrating an operation of a conversion process. FIG. 4 is a diagram showing an example of data. FIG. 5 is a flowchart illustrating an operation of a display candidate creation process. FIG. 6 is a flowchart illustrating an operation of a determination process. FIG. 7 is a diagram showing screen display information. FIG. 8 is a diagram showing screen display information. FIG. 9 is a diagram showing screen display information. FIG. 10 is a diagram showing screen display information. FIG. 11 is a diagram illustrating an example of contents of a learning table. [Description of Signs] 1 ... Input / output unit 2 ... Control unit 3 ... Storage unit

Continuation of front page    (56) References JP-A-56-38663 (JP, A)                 JP-A-58-115528 (JP, A)                 JP-A-53-7132 (JP, A)

Claims (1)

(57) [the claims] 1. A character string input means for inputting a kana character string that includes a plurality of phrases and is not separated in units of phrases; In the kana-kanji conversion method of performing kana-kanji conversion by a kana-kanji conversion device having a kana-kanji conversion device for extracting and performing kana-kanji conversion, and a kana-kanji conversion device having display means, Among the conversion candidates, the candidate character string composed of the conversion candidate with the highest probability for each phrase is displayed on the display means, and the conversion candidate corresponding to the phrase of interest is displayed so as to be distinguishable from the others. In response to an input that changes the segment break of the phrase of interest, the likelihood of conversion of the portion of the kana character string that follows the changed segment break is re-examined. Kanji conversion method and displaying the conversion candidates obtained by valence on the display means.