JPS6356757A - Western language preparing device with correcting function - Google Patents

Abstract

PURPOSE:To improve the accuracy of a word supplied from a dictionary by replacing an erroneous character string to the character string in which pronunciation resembles and spelling is different and supplying a correct answer candidate word. CONSTITUTION:In a Western language preparing device, a main dictionary and a user dictionary are provided and it is investigated whether or not the spelling of the inputted word is present in them. When the spelling is not in the dictionaries, the word preparation is executed by one-character replacement, one-character deletion, one-character deletion, one-character addition and adjoining character inversion and it is confirmed whether the word is present in the main dictionary or the user dictionary. When the word is absent, a pronunciation approximate information table is referred to, replaced to the character string in which the pronunciation resembles and the spelling is different, a correct answer candidate word is supplied and it is confirmed whether the word is in the main dictionary or the user dictionary.

Description

[Detailed description of the invention] <Industrial application field> The present invention relates to a European language creation device with a correct function.

<Prior art> Conventionally, in computer systems that handle English word information, word processing devices, typewriter devices, and other Roman writing devices, there has been a process for checking whether there are any errors in the spelling of input English m-words. , usually equipped with a dictionary. Furthermore, in addition to simply pointing out misspelled words, systems can now also provide suggestions for correct words. The currently popular method is a combination of the following four processes, but statistically speaking, correct words can only be obtained with an accuracy of around 80%.

1) Single character replacement example 5tataon −+ 5t
ation (replace a with i) 2) Delete one character Example 5tation -5ta
tion (delete 1 character 0) 3) Add 1 character Example 5taton→5tation(
Add i between t and 0) 4) Example of reversing adjacent characters 5tatre −+ 5tate
r (reverse re to er) By the way, in the past, most English words were input using the keyboard while looking at the manuscript.
Today, data entry methods are changing, such as typing on a keyboard while listening to a tape. As a result, the breakdown of spelling errors that occur is increasing, with errors such as errors in the same pronunciation but different spellings accounting for a larger proportion of total errors than simple key input errors. Therefore, conventional one character replacement, one character deletion, one
The process of adding characters and reversing adjacent characters can only yield correct candidate words with extremely low accuracy.

<Purpose of the Invention> Therefore, the present invention provides a more accurate correct candidate word by replacing input data of incorrect characters or character strings with characters or character strings that have similar pronunciations but different spellings. The goal is to make it possible.

<Structure of the Invention> In order to achieve the above object, the European language creation device with a collect function of the present invention includes a storage device for storing character information input from an input device, and at least a main dictionary area and characters or characters constituting a word. a dictionary device having a pronunciation approximation information table storing approximate pronunciation data that is a character or character string whose pronunciation is similar to that character or character string in association with the character string; Determine whether or not the character string is in the main dictionary, and if the character string is in the main dictionary, output the character string to the output device,
If the above string is not in the main dictionary, replace one character,
Perform one of the four processes of deleting one character, adding a character, and reversing adjacent characters, or add up the combination thereof, determine whether the character string after that process is in the main dictionary, and check whether the character string after the above process is in the main dictionary. If the character string exists in the main dictionary, the processed character string is output to the output device, while if the character string after the above processing is not in the main dictionary, the character string is decomposed and the character string after decomposition is output. The present invention is characterized in that it includes a control device that replaces the character or character string with approximate pronunciation data when the character or character string matches the character or character string in the pronunciation approximation information table.

<Example> Hereinafter, the present invention will be explained in detail with reference to illustrated embodiments.

In Figure 1, l is a keyboard, a tablet device, 0
CR (optical character reader), an input device for inputting English character word information data such as a magnetic tape, 2 is a core memory, IC, etc. that stores character information input from the input device l.
3 is a storage device such as a memory or a magnetic disk; 3 is an output device such as a printer, display device, magnetic tape, or magnetic disk; 4 is a main dictionary area; A dictionary device having a user dictionary area and a pronunciation approximation information table and supplying valid information in a timely manner with respect to the total spelling information of character and word data stored in a storage device; 5 is the input device 11; It is a control device made of, for example, a computer, which controls the exchange of signals between the storage device 2, the output device 3, and the dictionary device 4, and also performs a process of collecting character strings of input words, which will be described later.

As shown in Figure 3.4.5, the pronunciation approximation information table stores the characters or character strings constituting a word in association with the characters or character strings whose pronunciation is similar, that is, approximate pronunciation data. are doing. That is, this pronunciation approximation report table does not add pronunciation information to each word, but groups character strings that are likely to have the same pronunciation based on the arrangement of character strings that make up the word.

If the input word is an error (mispelling), the character string that makes up the input word is broken down, and if it is the same as the character string registered in this table, it is replaced with a candidate character string that is likely to have a similar pronunciation and a new one is created. It provides candidate words.

FIG. 2 is a flowchart of the collection process of this Latin language creation device with a collection function. First, step S5. St
checks whether the input word string exists in the main dictionary and user dictionary. The dictionary structure in this case is hash (
ITash) method, the method using the chain formula, and the dictionary construction method of the old Word Star (Word 5tar) by MicraPro (constructing the word by arranging the initial letter data horizontally in the order of Δ~Z) Any method may be used, such as a method using a two-dimensional index in which the number of characters is arranged vertically. If the character string is found in the main dictionary or user dictionary, the character string is assumed to be correct and is stored in the storage device 2 and further outputted from the output device 3. If the character string is not registered in such a main dictionary or user dictionary, step S3. S4. S.

Proceed to step S, and perform one of the above-mentioned processes. That is,
The correct spelling candidates from the components C=C,...Ci...CQ of the word length Q are: 1) One character replacement Character C at position 1.1≦i≦σ
Replace i with other C*.

2) Delete one character Character C4 at position i, 1≦1≦Q is removed.

3) Adding one character Insert a virtual character C* between position 1 and i+1.0≦i≦ρ.

4) Invert ci and ci+-1 at adjacent character inversion positions i and i+l, I≦i≦e-1.

Check whether a new word that can be newly derived by this process exists in the main dictionary or user dictionary, and if so, store the matching word in the storage device 2.

Statistical figures have shown that the correction rate of erroneous slurping can be achieved in the 80% range through the processing so far. However, an error rate of several 10% still occurs. Therefore, proceeding to step S7, in addition to the above-mentioned one process, 3.4.
．． By referring to the pronunciation approximation information table shown in FIG. 5, more accurate correct candidate words can be provided. This pronunciation approximation information table breaks down the character strings that make up a word, and when the character string registered in this pronunciation approximation information table and the input data are the same character string, the approximate pronunciation data that follows, Provides a new correct word candidate by replacing the candidate string. For example, if the pronunciation of one word is ``nee'', it is eau (beautiful).

eshiku rcad), eshic(queue), c
w (pew), 1 cu (adieu).

iew (view), ue (cue), ui (sui)
t), you(you).

There are character strings that have the same pronunciation or almost the same pronunciation as yu (yule), and these are j
li cannot be fully covered by the above processes 1) to 4).

Therefore, the above-mentioned processing is performed using the pronunciation approximation information table. Check whether the correct word candidate obtained in this way is the same as a word in the main dictionary or user dictionary, and if it matches, store it as a correct word candidate 59
i, supply to tsu2. In this way, highly accurate correct words can be supplied.

In addition, 3rd. ．． The pronunciation 8 similar information table shown in Figure 1.5 omits patterns that overlap with the four conventional processes.

<Effects of the Invention> As is clear from the above, the European language creation device with a correct function of the present invention, in addition to the conventional processing of one character replacement, one character deletion, one character addition, and adjacent character inversion, If a word is misspelled, it refers to the pronunciation approximation information table and replaces the incorrect character string with a character string that has a similar pronunciation but a different slurp, and searches for and supplies the correct candidate word. You can greatly improve the accuracy of the words you use.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a European language creation device with a correct function according to an embodiment of the present invention, FIG. 2 is a flowchart of the correct processing, and FIG. 3.4.5 is an explanatory diagram of a pronunciation approximation information table. l...input device, 2...storage device, 3...output device,
4...Dictionary device, death...control device. Patent applicant: Sharp Corporation Agent
Patent Attorney Tokide Ao AI-2fIEI Figure 1f2

Claims (1)

[Claims] (1) A storage device that stores character information input from an input device, and an approximate pronunciation that is a character or character string that has a pronunciation similar to that of at least the main dictionary area and the characters or character strings that make up the word. A dictionary device having a pronunciation approximation information table that stores data in association with each other, and determining whether or not the character string stored in the storage device is in the main dictionary, and if the character string is in the main dictionary. outputs the string to the output device, while if the string is not in the main dictionary, it replaces one character, deletes one character,
Perform any of the four processes of adding one character and reversing adjacent characters, or a combination thereof, and determine whether the character string after that process is in the main dictionary, and if the character string after the above process is in the main dictionary. In this case, the processed character string is output to the output device, while if the processed character string is not in the main dictionary, the character string is decomposed and the decomposed characters or character strings are stored in the pronunciation approximation information table. 1. A European language creation device with a correct function, comprising: a control device that replaces the character or character string with approximate pronunciation data when the character or character string matches the character or character string.