JPS641821B2 - - Google Patents

Description

[Detailed description of the invention]

INDUSTRIAL APPLICATION FIELD The present invention relates to a Japanese input device that performs kana-kanji conversion. Configuration of conventional examples and their problems In recent years, the Kana-Kanji conversion method is often used in Japanese input devices. However, with the current kana-kanji conversion, the desired kanji notation may not be obtained, so a certain amount of kanji notation selection, usually called homophone selection, is required after conversion. Therefore, it is important to minimize the chances of having to select kanji notation in the future through effective learning regarding kanji notation selection. A conventional Japanese input device will be explained below. FIG. 1 shows a block diagram of a conventional Japanese input device. In FIG. 1, 1 is an input code determination unit, which transfers the input code to the input character storage unit 2, which will be described later, if the input code is a character code, and if it is a replacement code, provides replacement instructions, which will be described later. Part 8
If the code is a selection code, the homophone learning information updating unit 9, which will be described later, is started. 2 is an input character storage unit which stores input characters. 3 is a word string recognition section, which activates the word search section 4 as needed to perform a word search, reads words from the obtained word group, extracts them in order of length, and certifies the word string; The plurality of recognized word strings are transferred to the word string storage section 6, and the first word string is instructed to the output section 7 to start it. Reference numeral 4 denotes a word search unit, which searches the word dictionary 5 for a group of words having a pronunciation that matches the character string after the specified position of the input character string in the input character storage unit 2. The word dictionary 5 has readings of words, kanji notation, and information indicating the inclusive relationship of readings between words. The word string storage section 6 stores a plurality of word strings certified by the word string certification section 3. Reference numeral 7 denotes an output section, which takes out the word string instructed by the word string recognition section 3 and replacement instruction section 8 from the word string storage section 6 and outputs it as a kanji notation. The replacement instruction unit 8 instructs the output unit 7 which word string should be output next to the word string currently being output. Reference numeral 9 denotes a homophone learning information updating unit, which updates the homophone learning information in the word dictionary 5 based on information related to word string homophones output at that time. The operation of the conventional Japanese input device configured as described above will be briefly described below. First, the input code determination unit 1 determines the input code, and if the code is a character code, it is transferred to the input character storage unit 2, and if it is a conversion activation code, it activates the word string recognition unit 3. The word string recognition section 3 searches the word dictionary 5 for a word group having a pronunciation that partially matches the character string in the input character storage section 2 by activating the word search section 4 as necessary. , qualify the word string. A case where the input character string is "Shinkai Hatsu" will be explained as an example. Tables 1 and 2 are examples of word search results and certified word strings of conventional Japanese input devices. Table 1 shows the search results for words with pronunciations that match the position of "shi" and later in the input character string "Shinkai Hatsu", including pronunciation, notation, inclusion relations of pronunciations, homophone relations, and homophone learning. Consists of information. Here, the inclusive relationship of reading refers to a relationship between words in which the reading of one word is part of the reading of another word from the beginning. The arrow in the term ``inclusion relationship'' indicates a relationship in which the pronunciation of the word at the tip of the arrow is from the beginning of the pronunciation of the word at the source of the arrow, and the black circle indicates the final word in the above relationship. Further, an arrow in a term related to homophones indicates that the word at the tip of the arrow is a homophone of the word at the source of the arrow, and a black circle indicates the end of the homophone. Furthermore, the number in the homophone learning information section indicates the number of the previously selected word among the words in a homophone relationship, counting from the first word, counting the first word as number 0. , written at the beginning of the homophone.

【table】

[Table] Therefore, the word "Shinka" is pronounced "evolution",
However, the word ``Shin'' indicates that ``Shin'' was selected last time. Continuing on, the word string recognition unit 3 causes the word search unit 4 to search for a group of words that have a pronunciation that matches the character string after the position of “ha” following “Shinkai” with the longest pronunciation length, and searches for the word “Shinkai” with the longest pronunciation length. ” is obtained, and a word string consisting of “deep sea” and “first” is recognized and stored in the word string storage unit 6. Below, in order of reading length, words are not searched after the ``i'' position following ``shinka'', and the word ``development'' is searched after the ``ka'' position following ``shin''. In this case, a word string consisting of "new" or "true" and "development" is added to the word string storage section 6, and the three word strings shown in Table 2 are obtained. Next, the output unit 7 is activated by the word string recognition unit 3, and
The first word string in the word string storage section 6, that is, the word string A in Table 2 is output. The input code determination unit 1 activates the replacement instruction unit 8 when determining that the input code is a replacement code. In order to sequentially output the word strings stored in the word string storage section 6 as A, B, and C, the replacement instruction section 8 instructs the output section 7 about the word string to be output next and activates it. . Further, the input code determination unit 1 activates the homophone learning information update unit 9 when determining that the input code is the selected code. The homophone learning information update unit 9 is
The homophone learning information in the word dictionary 5 is updated based on the word string output at that time and information related to homophones. Assuming that the word string being output at that point is C in Table 2, we can say that "shin" has been selected among the homophones of the reading "shin".
Learning is performed by setting the homophone learning information section of the word "new" in the first table in the word dictionary 5 to the value 1. However, with the above configuration and operation,
Since learning remains within the range of words that have homophone relationships, even if the character string "Shinkai Hatsu" is input again, the first output of the word string will be "Shinkai" and "Hatsu" in Table 2 A. Therefore, it is necessary to select the notation, and there are problems in that the operability is poor and the Japanese input speed is generally slow. Purpose of the invention The present invention solves the above-mentioned conventional problems.
The purpose of the present invention is to provide a Japanese input device that performs powerful learning on notation selection, has excellent operability, and has a fast Japanese input speed as a whole. Structure of the Invention The present invention provides means for determining the type of input code, means for storing input characters, and at least information indicating the inclusion relationship of pronunciations between words, as well as the pronunciation and kanji notation of words, and information included in the pronunciations. a word dictionary having learning information regarding selections among related word groups;
means for searching the word group having a pronunciation that partially matches the input character string; means for recognizing the word string; means for storing the word string; and learning the output order of the plurality of word strings. A means for determining based on the information and a means for updating the learning information when a selection code is input are provided, and learning regarding selection in a word group having an inclusive relationship in pronunciation is performed,
By determining the output order of multiple word strings based on this learning information and homophone learning information, it is possible to achieve excellent operability when inputting Japanese and increase the overall input speed. It is. DESCRIPTION OF EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 2 shows a block diagram of a Japanese language input device according to an embodiment of the present invention. In FIG. 2, 11 is an input code determination section, which corresponds to the input code determination section 1 in FIG. 1
2 is an input character storage section, which corresponds to the input character storage section 2 in FIG. Reference numeral 13 denotes a word string recognition section, which performs a word search by activating a word search section 14 (described later) as needed, extracts words from the obtained word group in order of pronunciation length, and generates a word string. The plurality of certified word strings are transferred to the word string storage section 6, and the word string output order determining section 17 is activated. 14 is a word search section, which corresponds to the word search section 4 in FIG. Reference numeral 15 denotes a word dictionary, which includes readings of words, kanji notation, information indicating the inclusion relationship of readings between words, and information regarding selection among word groups whose readings have inclusion relationships. Reference numeral 16 denotes a word string storage section, which temporarily stores the word string certified by the word string recognition section 13, and then stores the word string certified by the word string recognition section 13.
Store the resulting word string. Reference numeral 17 denotes a word string output order determining unit, which rearranges the word strings in the order in which they should be output based on the information on the word strings in the word string storage unit 16 and the learning information in the word search results and transfers them to the word string storage unit 16. At the same time, the first word string is instructed to the output unit 18 to start it. 18 is an output unit, which includes a word string output order determining unit 17 and a replacement instruction unit 19
The word string specified by is retrieved from the word string storage section 16 and output as Kanji notation. 19 is a replacement instruction section, which corresponds to the replacement instruction section 8 in FIG.
Reference numeral 20 denotes a word group learning information update unit, which updates learning information regarding the selection of a word group that has an inclusive relationship to the pronunciation in the word dictionary 15 and homophone learning information based on the word string that is output at that time. Update. The operation of the Japanese language input device of this embodiment configured as described above will be explained below, focusing on the points that differ from the conventional example. Similar to the description of the conventional example, the case where the input character string is "Shinkai Hatsu" will be described as an example. The process up to the word search is the same as the conventional example. Table 3 shows the results of searching for words with pronunciations that match the position of "shi" in the input character string.
In Table 3, in addition to the items in Table 1, there is an item of learning information regarding selection in a word group that has an inclusive relationship in reading. The number in this section is the number of words that were previously selected from among the word groups that have inclusionary pronunciations, such as "deep sea,""shinka," and "new," or their homophones, counting from the beginning with the beginning being number 0. It is attached to the first word to indicate which word it is. Therefore, in the example of Table 3, since the corresponding numerical value of "deep sea" is 2, the word selected last time in the word group is "new" or "true".

[Table] After this, the process is the same as in the conventional example until the word string recognition unit 13 stores the word string. The word string recognition unit 13 stores the word string in the word string storage unit 16.
When the word string output order determination unit 17 is stored in the word string output order determination unit 17, the word string output order determining unit 17 is activated. The word string output determining unit 17 outputs the order of the word strings in the word string storage unit 16 based on the learning information regarding the selection in the word group whose pronunciation has an inclusive relationship among the word search results and the homophone learning information. Sort them in the order they should be. In other words, the word string including the pronunciation "Shin", which is indicated by the value 2 of the learning information regarding the selection in the word group that has an inclusive relationship with the pronunciation of the longest matching word "Deep Sea" in Table 3, "Shin", "Development", etc. "True" and "Development" are ranked high,
Other word strings such as ``deep sea'' and ``first'' are ranked lower. Then, the word string ``shin'' and ``development'' containing the word ``shin'', which is indicated by the value 0 in the homophone learning information for ``shin'' in Table 3, is placed at the top, and the homophone of ``shin'', ``shin'', is placed at the top. The word strings containing "true" and "development" are in the following order.

[Table] Table 4 shows the contents of the word string storage section 16 determined by the word string output order determining section 17. The operations of the output section 18 and replacement instruction section 19 are similar to those of the conventional example. The word group information update unit 20 is the input code determination unit 1
1 is activated when the input code is determined to be a selection code, and 2 in the word dictionary 15 is activated based on the word string output at that time and information on the inclusion relationship and homophone relationship of the word pronunciation. Update learning information for an item. In other words, if F in Table 4 had been output at that point, the pronunciation of the selected word would be "Shinkai" instead of "Shinka" or "Shin".
This is learned by setting the value of the learning information regarding the selection of the longest matching word "deep sea" in a group of words whose pronunciation has an inclusion relation to 0, which indicates the word "deep sea" itself. Next, read “Shinkai”
In this case, there is no homophone for ``deep sea,'' so nothing is done. If there is a homophone, the value of the homophone learning information for the first word ``deep sea'' in the pronunciation ``shinkai'' is set to 0, which indicates the word ``deep sea'' itself. As described above, according to this embodiment, the word dictionary includes learning information regarding the selection of word groups that have inclusion relations in pronunciation, and also includes learning information on homophone learning information.
By updating, the learning function related to kanji notation selection will be strengthened. For example, the first output of the conversion result for the subsequent input "Shinkai Hatsu" will be "New" and "Development", which will reduce the frequency of subsequent operator operations, as can be seen from the fact that replacement and selection operations are no longer required. At the same time, the overall input speed can be improved. In addition, in this example, regarding the learning information and homophone learning information regarding the selection in a word group that has an inclusionary relationship in the pronunciation, it is assumed that only the most recently selected one is learned. It goes without saying that the information may be learned as information indicating the history. In addition, word strings are recognized in the order of word search results, starting with the word with the longest pronunciation. It goes without saying that the word string recognition may be terminated midway by dynamically providing conditions for terminating the word string recognition. Effects of the Invention As described above, the Japanese input device of the present invention has a means for determining the type of input code, a means for storing input characters, and at least an inclusive relationship between readings in addition to word readings and kanji notation. and means for searching for a word group having a pronunciation that partially matches an input character string;
means for recognizing the word string; means for storing the word string; means for determining the output order of the plurality of word strings based on the learning information; and means for updating the learning information when a selection code is input. By providing this, it is possible to perform powerful learning and use of notation selection, simplify operations for the operator when inputting Japanese, and speed up the input speed, which has great effects.

[Brief explanation of drawings]

Figure 1 is a block diagram of a conventional Japanese input device.
FIG. 2 is a block diagram of a Japanese language input device according to an embodiment of the present invention. 11... Input code determination section, 12... Input character storage section, 13... Word string recognition section, 14... Word search section, 15... Word dictionary, 16... Word string storage section, 17... Word string Output order determining unit, 20...word group learning information updating unit.

Claims (1)

[Claims] 1. Input code determination means for determining the type of input code, input character storage means for storing input characters, and at least information indicating the inclusive relationship of pronunciations between words, and information included in the pronunciations, in addition to the pronunciations and kanji representations of words. a word dictionary having learning information regarding selection among related word groups; a search means for searching the word group having a pronunciation that partially matches an input character string; a certification means for recognizing a word string; and a word string. a word string storage means for storing a plurality of word strings, a means for determining an output order of the plurality of word strings based on the learning information, and an updating means for updating the learning information when a selection code is input. Input device.