JPH06223055A - Document input device - Google Patents

Abstract

PURPOSE:To convert an input which contains KANJI (Chinese character) partially from KANJI to KANA (Japanese syllabary) at the time of input by KANA-KANJI conversion and to prevent word dictionary capacity from increasing so much. CONSTITUTION:The document input device consists of an input device 1, a converting device 2 which converts an inputted character string, an output device 3 such as a display where a character string as a conversion result is displayed, and a controller 4 which controls the whole operation and the converting device 2 is equipped with a word dictionary 21 in which the reading, description, and part of speech of the inputted character string are registered, a word retrieval part 22 which retrieves words registered in a word dictionary 21 according to the inputted character string, a word storage part 23 which stores the words retrieved by the word retrieval part 22, a word confirmation part 24 which confirms whether or not the retrieved words match the input character string, and a word selection part 25 which selects proper words out of the retrieved word group.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a text input device, and more particularly to a text input device for converting a kana-kanji character by inputting a character string containing a mixture of kanji characters.

[0002]

2. Description of the Related Art As a popular text input device,
There is Kana-Kanji conversion that converts Kana or Romaji input into a character string containing Kanji and Kana. As input character strings for these devices, not only kana but also kanji may be directly input. For example, in the case of input by handwriting character recognition using a pen and a tablet, it is possible to mix and input kanji and kana. In addition, some kanji-kanji input methods have a kana-kanji conversion function as an auxiliary function for inputting kanji whose kanji code is unknown. A kana-kanji conversion device that can input part or all of the input in kanji is used in both devices.

An example of a conventional text input device for performing kana-kanji conversion by inputting a sentence containing mixed kanji is "Kana-kanji conversion for 2-stroke input" (Shiomi, Kita, Kawai, Oiwa: Information Processing Society of Japan Magazine, Vol.33, No.7, p
p. 920-927, 1992), or "A morphological analysis program with notation conversion and its application" (Kaneko, Torihara, Kenseki: IPSJ 45th National Convention, 4C-4,
1992).

In any of the text input devices described in any of the papers, in order to perform kana-kanji conversion of a text in which kanji are mixed,
Words in which some or all of the reading of each word is in Kanji are registered in the word dictionary. For example, with respect to the word whose reading is “Doyoubi” and whose notation is “Saturday”, eight ways shown in FIG. 2 (a) are registered in the method of the latter paper.

The kana-kanji conversion process of the above-mentioned conventional text input device is the same as that of a general kana-kanji conversion device that uses only kana as an input character string except that kanji is allowed for reading an input character string or a word dictionary. .

An example of a conventional word dictionary construction method will be given below.

As a simple method, there is a method in which word information is directly described as shown in FIG. 2, sorted with a key at the time of search, and then searched by a binary search method or the like.

In the method of sharing the common part at the beginning of the key for word search (FIG. 7), the capacity of the shared part can be reduced. In addition, it is easy to search for a word that has a common first character, which is often used in Japanese processing.

In order to further reduce the capacity, in the "natural language dictionary search technique using two tries" (Morimoto, Aoe: 45th National Convention of Information Processing Society of Japan, 5F-4, 1992), the common part at the end of words Has been proposed (Fig. 8).

For other various dictionary construction methods and search methods, see "The Art of Computer P".
programming ”(Knuth, 1973),
"Serial Series: Key Search Technique" (Aoe: Information Processing, Vo
l. 33, No. 11, pp. 1359-1366, 1
992) and the like.

[0011]

In the above-mentioned conventional text input device, all possible mixed writing methods are registered as words. However, in order to be able to handle no matter how mixed words are written by this method, it is generally necessary to register 2 ⁿ words for words containing n kanji characters, and the number of words to be registered in the word dictionary is Grows very large. Since these mixed words have a common part at every character position, the conventional text input device equipped with a word dictionary sharing only the beginning and end of the word cannot sufficiently reduce the dictionary capacity. There are challenges.

The object of the present invention is to solve the above-mentioned problems,
It is an object of the present invention to provide a sentence input device that reduces the increase in the dictionary capacity by sharing a common part with a corresponding kana written word when registering a word in which kanji is mixed in reading into a word dictionary.

[0013]

In order to solve the above-mentioned problems, a sentence input device of the present invention comprises an input device for inputting an input character string, a first character string before replacement for each word, and a replacement after replacement. Of the second character string and part of speech, a word search unit for searching the word dictionary for a word in which the first character string matches a part of the input character string, and the word search unit. For each word, compare the first character string and the second character string to delete the word not registered in the word dictionary, and from the words remaining after deleting the word by the word confirmation. The word dictionary includes a word selection unit that creates an output character string, and an output device that displays the output character string, and the word dictionary is the same as the word when the word including Chinese characters is registered in the first character string. Said second character string and the same item The have to share a common portion of the first string between the words without the Chinese character in the first string.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a text input device of the present invention.

In FIG. 1, the embodiment of the present invention is such that an input device 1 such as an input tablet or a keyboard, a conversion device 2 for converting an input character string, and an output such as a display for displaying the conversion result character string. It is composed of a device 3 and a control device 4 for controlling the overall operation.

The conversion device 2 includes a word dictionary 21 in which readings, notations, and parts of speech of words are registered, and a word search unit 2 for searching a word registered in the word dictionary 21 from an input character string.
2, a word storage unit 23 for storing the words retrieved by the word retrieval unit 22, a word confirmation unit 24 for confirming whether or not the retrieved word matches the input character string, and a retrieved word group. The word selection part 25 which selects a suitable word is provided.

The word dictionary 21 is composed of an index part formed based on the reading of words and other information such as notations and parts of speech. In the text input device of the present invention, the term "yomi" is used, including the case where part or all of the pronunciation is replaced with kanji. The index unit of the word dictionary 21 has a tree structure including a table (n-th character table) in which characters appearing at the n-th character in reading the registered word are collected for each character having the same n-1 character. . Each n-th character table is composed of a combination of an n-th character, a pointer to information on a word composed up to the n-th character, and a pointer to an (n + 1) -th character table that appears after the character.

As an example, FIG. 4 shows an index portion of a word dictionary in which three words "upper and lower lines", "up and down lines" and "castle town" are registered. In FIG. 4, reference numeral 41 is a table in which the character (“ji”) appearing as the first character in reading the word is registered. A pointer (indicated by a solid arrow in FIG. 4) is provided from the character appearing as the first character to the second character table 42 in which the character appearing as the second character (“yo”) is registered. In the example of FIG. 4, the character that appears in the fourth character is "ka" or "ke", so the fourth character table 44 includes two characters. Furthermore, the fifth character table following each character of the fourth character table is 45,
46.

From the end of each word, a pointer (indicated by a dashed arrow in FIG. 4) that points to information such as the notation and part of speech of the word or a storage location of the information (indicated by a double circle in FIG. 4) is displayed. It is registered.

The word dictionary 21 shares a common part with the corresponding kana writing word when the post-mix writing is registered.
For example, mixed words of "upper and lower lines" and "castle town" (Fig. 2
When registering (b)), the word dictionary of the conventional text input device requires the index unit having the configuration shown in FIG. On the other hand, the index part of the word dictionary 21 of the text input device according to the present invention is shown in FIG.
The configuration is shown in.

In FIG. 3, when a word whose reading is "Asensen" is registered, since the common portion "Gensen" with the corresponding Kana writing word "Jogensen" is shared, the first character table 3
"Up" of 1 is the same as "u" of the third character table 33 4
It has a pointer to the character table 34. In addition, when registering a word whose reading is "josen", the common part "sen" is shared, so "bottom" of the fourth character table 34 is registered.
Has a pointer to the fifth character table 35, which is the same as "ge". Since “below” in the fourth character table 34 is also used to search for “joshitamachi”, it also has a pointer to the same fifth character table 36 as “ka” in “jokamachi”.

The word search unit 22 searches the word dictionary 21 from each character position of the input character string. The operation of the word search unit 22 will be described with reference to FIG. In the following description and FIG. 5, the input character string is s, and the jth character of the input character string is s [j].
Express.

(A) i ← 1 (search from the word that matches the first and subsequent characters of s). (Step 51) (b) When the word search is performed for all the characters s, the process ends.
(Step 52) (c) j ← 1 (search is started from the first character table of the word dictionary 21). (Step 53) (d) It is checked whether or not s [i] is in the i-th character table. If there is, proceed to (e), and if not, proceed to (h).
(Step 54, Step 55) (e) It is checked whether there is a word corresponding to the reading of s [i], s [i + 1], ..., S [i + j-1], and if there is, the word information is stored in the word storage unit 23. Write. (Step 56) (f) Check the pointer to the j + 1th character table. If there are multiple pointers, one is left and the other is held on the stack together with the values of i and j. (Step 57)
(G) Increase j by 1 and proceed to (d). (Step 58) (h) It is checked whether or not there is an unprocessed pointer on the stack. If there is an unprocessed pointer, the process proceeds to (i), and if not, the process proceeds to (j). (Step 59) (i) Take out an unprocessed pointer from the stack (d)
Go to. (Step 510) (j) Increase i by 1 and proceed to (b). (Step 51
1) The word confirming unit 24 confirms whether or not the reading used in the search and the notation of the searched word can be associated with each other among the words written in the word storage unit 23 that include Chinese characters in the reading. If the word cannot be associated, the word is deleted from the word storage unit 23. The word search unit 22 may successfully search for a word that is not actually registered. For example, if the word dictionary 21 of FIG. 3 is searched with the reading “upper and lower towns”, information (37 in FIG. 3) of the word “castle town” is retrieved. However, "upper and lower towns"
Is a word registered in the word dictionary 21 of FIG.
Not (b)). Therefore, the word checking unit 24 checks whether the word in the word storage unit 23 is actually registered in the word dictionary 21. The word confirmation unit 24 compares the reading “upper and lower towns” used in the search with the type and order of the kanji included in the searched word notation “castle town”, and if they do not match, deletes them from the word storage unit 23.

An operation example of the word confirmation unit 24 will be described with reference to FIG. In the following description and FIG. 5, the i-th reading character is expressed as y [i], the j-th character writing is expressed as h [j], and the reading length is expressed as L _y .

(A) i ← 1, j ← 1. (Step 61) (b) If i> L _y , the association is successful and the process ends. (Step 62) (c) Search for kanji after y [i]. If not found, matching is successful and the process ends. If found, p ← let found character position. (Steps 63 and 64) (d) Search y [p] after h [j]. If not found, matching fails and ends. If found, set q ← found character position. (Steps 65 and 66) (e) As i ← p + 1 and j ← q + 1, proceed to (c).
(Step 67) The word selection unit 25 is the same as that used in the conventional text input device. For example, "Kana-to-Kana conversion system for solid writing and its homophone processing" (Makino, Kizawa: IPSJ Journal, Vol. 22, No. 1, pp. 59-6.
7 (1981)).

[0027]

As described above, the text input device according to the present invention shares a common part with a corresponding kana written word when registering a word in which kanji is mixed in reading into a word dictionary. The effect is that the dictionary capacity does not increase much.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a text input device of the present invention.

FIG. 2 is a block diagram showing an example of mixed writing of words.

FIG. 3 is a conceptual diagram of a word dictionary of the text input device of the present invention.

FIG. 4 is a conceptual diagram of a word dictionary in which only kana written words are registered.

FIG. 5 is a flowchart showing the operation of the word search unit.

FIG. 6 is a flowchart showing the operation of the word confirmation unit.

FIG. 7 is a conceptual diagram of a word dictionary of a conventional text input device.

FIG. 8 is a conceptual diagram of a word dictionary in which the beginning and end of a word are shared.

[Explanation of symbols]

 1 Input Device 2 Conversion Device 3 Output Device 4 Control Device 21 Word Dictionary 22 Word Search Section 23 Word Storage Section 24 Word Confirmation Section 25 Word Selection Section

Claims (1)

[Claims] 1. A text input device for replacing an input character string with another character string to form an output character string, an input device for inputting the input character string, and a first character string / replacement before replacement for each word. A word dictionary for registering a subsequent second character string and a part of speech, a word search unit for searching the word dictionary for a word in which the first character string matches a part of the input character string, and the word search unit. For each word searched for, after comparing the first character string and the second character string and deleting the word not registered in the word dictionary, after deleting the word by the word confirmation unit A word selection unit that creates an output character string from the remaining words, and an output device that displays the output character string, and the word dictionary, when registering a word containing Kanji in the first character string, The second letter identical to the word A sentence input device, wherein a common part of the first character string is shared with a word having a column and the same part of speech, and the first character string does not include kanji.