JP2002279350A - Word dictionary maintenance method, character recognition method, computer-readable storaging medium stored with program for making the methods executed, and word dictionary maintenance system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the increase in the number of words by providing validity of each word data in a word dictionary and appropriately deleting the words whose validity has expired. SOLUTION: This word dictionary maintenance method in a information processing device is provided with a language processing part 2, and a word dictionary part 5 for storing the words used for the language processing by the language processing part 2. A word dictionary maintenance means 6 applies the validity to the words, determines whether the validity of the words has expired, and deletes the words whose validity is determined to have expired from the word dictionary part 5. When validity of the words has expired, the word dictionary maintenance means 6 confirms to a user whether the words whose validity has expired are deleted and deletes the words whose validity has expired from the word dictionary part 5, based on the command of the user.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a word dictionary maintenance system in an information processing device such as a character recognition device.

[0002]

2. Description of the Related Art Conventionally, there has been disclosed a method of additionally registering a correction result of a user in a word dictionary (Japanese Patent Laid-Open No. 6-25120).
No. 4). In addition, when a user corrects a result of character recognition, a method of counting words and occurrence frequencies in the corrected recognition result and adding / registering words having a high appearance frequency to a word dictionary is also described. JP-A-5-225401).

In these prior arts, frequently used or important words are added to a word dictionary according to the user, and the recognition performance is improved as the user uses the character recognition device. There is a merit of going.

[0004]

However, on the other hand, there is a problem that the number of words registered in the word dictionary increases. When the number of words increases, there is a concern that the processing speed will decrease and the capacity of the word dictionary will increase. Also, when a large number of words having a low appearance frequency are registered, there is a high possibility that the processing of a word having a high appearance frequency will be adversely affected.

Therefore, the present invention has been made in view of such problems, and provides an expiration date for each word data in a word dictionary, and appropriately performs processing such as deletion of a word whose expiration date has passed. The purpose is to prevent an increase in the number of words.

[0006]

The above object is achieved by the following means of the present invention.

According to a first aspect of the present invention, there is provided a word dictionary maintenance method for an information processing apparatus having a language processing unit and a word dictionary unit storing words used for language processing by the language processing unit. So,
Providing an expiration date for the word, determining whether or not the expiration date of the word has passed, and deleting the expired word from the word dictionary unit. It is a feature.

According to a second aspect of the present invention, there is provided the word dictionary maintenance method according to the first aspect, wherein the user confirms whether or not to delete the expired word from the word dictionary section. And deleting a word whose expiration date has expired from the word dictionary unit based on an instruction from the user to delete a word whose expiration date has passed.

According to a third aspect of the present invention, there is provided the word dictionary maintenance method according to the first or second aspect, wherein the user instructs not to delete the expired word from the word dictionary section. And re-entering an expiration date for the word for which the user instructed not to delete.

According to a fourth aspect of the present invention, there is provided the word dictionary maintenance method according to the first or second aspect, wherein a user instructs not to delete an expired word from the word dictionary section. And resetting the expiration date of the word instructed not to be deleted by using a predetermined expiration date.

According to a fifth aspect of the present invention, there is provided the word dictionary maintenance method according to the first or second aspect, wherein a user instructs not to delete an expired word from the word dictionary section. Calculating the expiration date of the word instructed not to be deleted from the history of the update history section in which the update history up to the present is stored; and calculating the expiration date of the word instructed not to be deleted. Setting based on the following.

According to a sixth aspect of the present invention, there is provided a character recognition apparatus comprising: a language processing unit; and a word dictionary unit storing words used for language processing by the language processing unit. A character recognition method, comprising: a step of giving an expiration date to the word; and a step of ignoring a word whose expiration date has passed in the course of the character recognition processing.

According to a seventh aspect of the present invention, there is provided a character recognition apparatus comprising a language processing section and a word dictionary section in which words used for language processing by the language processing section are stored. A character recognition method, comprising: a step of giving an expiration date to the word; and a step of outputting a word having a long expiration date as a priority as a recognition result in performing the character recognition processing. It is assumed that.

The word dictionary maintenance method according to the present invention is a word dictionary maintenance method for an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. So,
A user registering a word in the word dictionary unit;
A user inputting an expiration date for the word registered in the word dictionary unit.

According to a ninth aspect of the present invention, there is provided a word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit storing words used for language processing by the language processing unit. So,
A user registering a word in the word dictionary unit;
Calculating the expiration date of the word registered in the word dictionary unit from the date and time when the word was registered in the word dictionary unit.

According to a tenth aspect of the present invention, there is provided a word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit storing words used for language processing by the language processing unit. A step of: registering a word in the word dictionary unit by a user; and calculating an expiration date of the word registered in the word dictionary unit based on characteristics of the word registered in the word dictionary unit. It is characterized by having.

The word dictionary maintenance method according to the present invention is a word dictionary maintenance method for an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. There is a step of automatically confirming time via a communication line for confirming the expiration date of the word.

According to a twelfth aspect of the present invention, there is provided the word dictionary maintenance method according to the first aspect, wherein the information on the expiration date of the word stored in the word dictionary is communicated with the word dictionary. And deleting the expired word from the word dictionary unit based on the information on the expiration date stored in the server connected via the server. .

The invention according to claim 13 is the invention according to claims 1 to 1
2 is a computer-readable storage medium storing a program for executing the word dictionary maintenance method described in 2.

According to a fourteenth aspect of the present invention, there is provided a word dictionary maintenance system for an information processing apparatus having a language processing unit and a word dictionary unit storing words used for language processing by the language processing unit. A word maintenance unit for assigning an expiration date to a word stored in the word dictionary unit, wherein the word dictionary maintenance unit is configured such that, when the expiration date of the word has passed, the word whose expiration date has passed Is deleted from the word dictionary unit, and the validity period of a word whose validity period has expired is selected and executed.

According to the invention described above, in general, among words used in a document or the like, there are not a few words called buzzwords. These buzzwords temporarily appear at a high frequency, but after a certain period of time, the appearance frequency becomes almost zero. For such words, an appropriate expiration date is set at the time of registration, and when the expiration date has expired, the number of words is reduced by deleting the word from the word dictionary. When the number of words is reduced, there is an advantage in that the time required for language processing is reduced and the capacity of a word dictionary is reduced (reduction in the amount of memory required for processing). There is also an advantage that the processing performance for words is improved.

[0022]

Embodiments of the present invention will be specifically described below with reference to the drawings. FIG. 1 shows an apparatus configuration (1) using the word dictionary shown in the present invention for the language processing unit 3 of the character recognition apparatus. As a configuration of the character recognition device, generally, an image input unit 1 (an image input device such as a scanner or a digital camera), a character recognition unit 2 (here, line segmentation,
It consists of character extraction, one-character recognition, and processing for outputting candidate character groups), a language processing unit 3, and a result output unit 4 (display on a screen or output of a text file). The content of the word dictionary 5 is updated by the word dictionary maintenance unit 6. The word dictionary maintenance unit 6 may update the word dictionary 5 via a network as shown in the device configuration (2) of FIG. 2, or may collect information necessary for updating time and the like via the network. May be.

Since the present invention is deeply related to the language processing unit 3 therein, the language processing unit 3
Will be described in detail.

Next, the operation of the language dictionary unit in this embodiment will be described. (Embodiment 1) In Embodiment 1, the processing of the maintenance unit 6 of the word dictionary 5 will be described. The word dictionary 5 is shown in FIG.
As shown in the example (1) of the word dictionary, information necessary for linguistic processing (language processing information) such as a word, its expiration date, part of speech, and connection is stored. The expiration date “**** / ** / **” in the figure indicates that there is no expiration date. Since the language processing information is not the center of the present invention, the description is omitted here.

A word dictionary 5 initially provided by a system for character recognition or the like has an indefinite period for all words as shown in a word dictionary example (1) in FIG. It may be set, or a dictionary provided by the system may include a word whose expiration date is set. In addition, by separating words with an expiration date and those without an expiration date in the word dictionary, it is possible to eliminate the expiration date field for words without an expiration date (the capacity of the dictionary can be reduced. ).

Here, the processing of the word dictionary maintenance unit 6 when the user adds the word "revenge" will be described with reference to FIG. When adding, the user enters an expiration date along with the word. Further, as a simpler example, there is a method in which only a fixed time from the date and time of registration is set as an expiration date on the system side. For example, if the validity period is three years, and if the registration date is February 23, 2000, the validity period is February 23, 2003, as shown in the word dictionary example (2) in FIG. Register additional.

Further, as shown in FIG. 5, there is a method in which the expiration date is made variable by the characteristics of the word to be added. The characteristics of a word include characteristics such as part of speech, word length, and character type. As a simple example, an example of calculation based on the character type of the characters constituting the word will be described. Examples of the character type and the validity period are shown in the validity period example (1) of FIG. As shown in FIG. 5, it is effective for two years if the word is hiragana, one year for katakana, three years for kanji, one year for numbers, and two years for mixed words. The deadline can be variable depending on the characteristics of the word. In this example, since “Revenge” is katakana, the validity period is one year. If the registration date and time is February 23, 2000, the validity period is 2001/2.
As month 23, additional registration is performed as shown in a word dictionary example (3) in FIG.

Hereinafter, the processing when the expiration date has passed will be described using the case of the word dictionary example (2) in FIG.
For words whose validity period has passed, the system may automatically delete the word whose validity period has passed, or may ask the system user for confirmation and delete it. When asking the user for confirmation, there is a method of displaying a message as shown in FIG. 7, and when the user inputs Yes, deleting the corresponding word. If the user does not agree with the deletion, the expiration date may be input again. FIG. 8 shows a message example in that case.
A confirmation screen example (5) is shown.

On the other hand, if the user is not asked for confirmation,
What is necessary is just to determine the default validity period at the time of resetting, to obtain the value, and to perform resetting. For example, if the default validity period at the time of resetting is 2 years, the term of validity of “Revenge” is extended by 2 years from February 23, 2003 to the word dictionary after resetting, and becomes February 23, 2005, The word dictionary is as shown in a word dictionary example (4) in FIG. Alternatively, if the user's consent is not obtained, the word is still frequently used after a certain period of time, so the corresponding word is determined to be not a temporary buzzword it can. Therefore, when the user's consent cannot be obtained, there is a method of setting the expiration date to be indefinite.

If the user's consent cannot be obtained with respect to "revenge", assuming that the expiration date is indefinite, the word dictionary is updated as shown in a word dictionary example (5) in FIG. Further, there is a method of storing an update history in a word dictionary and calculating the expiration date of the word to be updated based on the update history. An example in which information on the number of updates is stored as an update history in the case of the word dictionary example (2) in FIG. 4 is shown in a word dictionary example (6) in FIG.

When the word dictionary example (6) in FIG. 11 is used,
If the expiration date has expired and the user has not approved the deletion,
The validity period thereafter is determined as shown in the validity period example (2) in FIG. 12 according to the number of updates. As shown in FIG. 11, when the expiration date of the word has expired and the user has not approved the deletion, the update count is 0 for 3 years, 1 time for 5 years, and 2 times. The expiration date is renewed indefinitely for three times for seven years. For example, 200
When the expiration date of “Revenge” expires on February 23, 3rd, and the user does not approve the deletion, the number of updates at that time is 0. February 23, 2006. Figure 1 shows the results.
3 is shown in a word dictionary example (7).

Further, on February 23, 2006,
If the expiration date of “Revenge” comes again and the user does not approve the deletion, the number of updates at that time is one, so the expiration date is five years and the expiration date is 2011
March 23. The result is shown in FIG. 14 (8).
Shown in Here, the simplest example is shown, but there are many other methods for setting the expiration date using the update history. Second Embodiment In the first embodiment, the processing of the word dictionary maintenance unit 6 has been described. In the second embodiment, the first embodiment
2 shows an example of a character recognition process using the word dictionary shown in FIG.
An example of the character recognition result processed here is shown in a character recognition result example (1) of FIG. A word dictionary record used in the processing is shown in a word dictionary example (9) in FIG. Since the details of the character recognition processing are not the gist of the present invention, the description is omitted.

Here, it is assumed that three words of "human face", "heaven", and "large country" are obtained when a word that matches the combination of candidates is searched from the word dictionary from the top. In a general character recognition device, a candidate word is selected based on a candidate order, a word occurrence frequency, and the like. In the case of this example, if a word is selected from only the candidate ranks, “person” (average candidate rank 1.
5) The word "person" is selected, which is "large country" (2.0) and "heaven" (3.0), and has a small average candidate rank. Although the average candidate rank is not shown in the figure, the average candidate rank is set to 1.5 for "person", 2.0 for the great power, and 3.0 for heaven.

In this embodiment, an expiration date is used when selecting this word. In the case of the above example, as shown in FIG. 16, there is no expiration date for “large power” and “heaven”, and for “person”, the expiration date is March 18, 2002. Choose "Great Power" or "Heaven". Here, since the average candidate ranking is smaller for "large country" (not shown), "large country" is finally selected. Similarly, for the fourth and fifth characters, “face” and “white day” are compared, and the expiration dates are the same, so that “face” having a smaller average candidate rank is selected. Although not shown, "white eyes"
It is assumed that the “face” has a smaller average candidate rank.
Therefore, finally, a candidate of "the face of a great power" is selected and output.

Consider a case where similar processing is performed on March 20, 2002. According to the first embodiment, as shown in FIG. 16, the dictionary record of “person” should have been deleted on March 20, 2002. Here, the dictionary record of the first embodiment is deleted. Processing when no part is used will be described. In this case, the effect of preventing an increase in the dictionary capacity cannot be expected, but by using the following method, it is possible to prevent a word that is not frequently used from adversely affecting the processing.

As in the previous example, when a word that matches the combination of candidates from the beginning is searched in the word dictionary, the expired word is ignored. In this case, FIG.
As shown in, "Human face" has already expired at this point and is ignored,
"Heaven" and "Superpower" will be obtained. Since the subsequent processing is the same, the description is omitted, but by ignoring the expired words in this way, even if the expired words are not deleted from the word dictionary, the occurrence frequency is low. It is possible to prevent a word from adversely affecting processing. (Third Embodiment) In a third embodiment, a method of maintaining a word dictionary via the network 8 will be described using an example shown in the device configuration (1) of FIG. The word dictionary maintenance unit 6 periodically obtains accurate time information using the time correction server 9 via the network 8. The time correction server 9 has a site on the Internet that provides time using an atomic clock or the like.
You can use them. Other processes are the same as those in the first embodiment, and a description thereof will not be repeated. (Embodiment 4) Next, in Embodiment 4, the apparatus configuration (2) shown in FIG.
A method of maintaining a word dictionary via a network will be described using an example shown in FIG. In the fourth embodiment, the word dictionary is divided into a part required for processing and a part required for maintenance, and information necessary for the part required for maintenance is stored in the word dictionary maintenance unit 6.
Accesses the word dictionary 5 via the network. The word dictionary 5 in this case is replaced with the word dictionary example (10) in FIG.
Further, an example of the maintenance data 10 is shown in a maintenance data example (1) of FIG. The process of adding a word is the same as the process except that additional information is sent via a network, and thus the description is omitted.

Here, a method of deleting a word whose expiration date has expired will be described. The word dictionary maintenance unit 6 manages the expiration date in the maintenance data 10 and sorts the expiration dates in a simple example and stores the closest expiration date. , Delete the corresponding word through the network.

In the case of the maintenance data (1), 200
Since the word “revenge” expires on February 23, 2001, the word dictionary maintenance unit 6 accesses the word dictionary 5 via the network 7 and reads the word dictionary in the word dictionary example (10) in FIG. Search for the word "revenge"
delete. Although the simplest method has been described here, a general-purpose method of DB management, such as associating data in the maintenance data 10 with words in the word dictionary using information such as a word record ID, is also used here. Available. When the deletion is performed, the result is as shown in the word dictionary (11) in FIG. 19, and the number of dictionary records is reduced, and it is possible to prevent a word having a low appearance frequency from affecting the processing.

According to the invention described above, in general, among words used in a document or the like, there are not a few words called buzzwords. These buzzwords temporarily appear at a high frequency, but after a certain period of time, the appearance frequency becomes almost zero. For such words, an appropriate expiration date is set at the time of registration, and when the expiration date has expired, the number of words can be reduced by deleting the word from the word dictionary. Reducing the number of words has the advantages of reducing the time required for language processing and reducing the size of the word dictionary (reducing the amount of memory required for processing), and deleting words that appear less frequently.
There is also a merit that processing performance for a word having a high appearance frequency is improved.

Although the preferred embodiments of the present invention have been described in detail above, the present invention is not limited to the specific embodiments, and various modifications may be made within the scope of the present invention described in the appended claims. Can be modified and changed.

[0041]

As is clear from the above description, according to the invention described in the claims, an appropriate expiration date is set when registering a word, and when the expiration date expires, the word The number of words can be reduced by deleting from the dictionary. In addition, when the number of words decreases, the time required for language processing can be reduced, and the capacity of the word dictionary can be reduced (reduction in the amount of memory required for processing). Processing performance with respect to words having a high score can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a device configuration example (1) according to the present embodiment.

FIG. 2 is a diagram showing a device configuration example (2) in the present embodiment.

FIG. 3 is a diagram showing a word dictionary example (1).

FIG. 4 is a diagram showing a word dictionary example (2).

FIG. 5 is a diagram illustrating an example of a valid period (1).

FIG. 6 is a diagram showing a word dictionary example (3).

FIG. 7 is a diagram showing a confirmation screen example (4).

FIG. 8 is a diagram showing a confirmation screen example (5).

FIG. 9 is a diagram showing a word dictionary example (4).

FIG. 10 is a diagram showing a word dictionary example (5).

FIG. 11 is a diagram showing a word dictionary example (6).

FIG. 12 is a diagram showing an expiration date example (2).

FIG. 13 is a diagram showing a word dictionary example (7).

FIG. 14 is a diagram showing a word dictionary example (8).

FIG. 15 is a diagram illustrating a character recognition result example (1).

FIG. 16 is a diagram showing a word dictionary example (9).

FIG. 17 is a diagram showing a word dictionary example (10).

FIG. 18 is a diagram illustrating an example of maintenance data.

FIG. 19 is a diagram showing a word dictionary example (11).

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Image input part 2 Character recognition part 3 Language processing part 4 Result output part 5 Word dictionary part 6 Word dictionary maintenance part 9 Time correction server

Claims (14)

[Claims] 1. A word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored, wherein a term of validity is set for the word. A word dictionary maintenance method, comprising: a step of assigning; a step of determining whether or not an expiration date of a word has passed; and a step of deleting a word whose expiration date has passed from the word dictionary unit. 2. The word dictionary maintenance method according to claim 1, further comprising: asking the user whether to delete the expired word from the word dictionary unit; Deleting a word whose expiration date has passed from the word dictionary unit based on an instruction to delete the word. 3. The word dictionary maintenance method according to claim 1, wherein when the user instructs not to delete the expired word from the word dictionary unit, the user instructs the word not to be deleted. A word dictionary maintenance method comprising the step of re-entering an expiration date. 4. The word dictionary maintenance method according to claim 1, wherein when the user instructs not to delete the expired word from the word dictionary unit, the expiration date of the word not instructed to be deleted is provided. Re-setting using a predetermined validity period. 5. The word dictionary maintenance method according to claim 1, wherein when the user instructs not to delete the expired word from the word dictionary section, the expiration date of the word not instructed to be deleted. Calculating from the history of the update history part in which the update history up to the present is stored, and setting the expiration date of the word instructed not to be deleted based on the calculated expiration date. A word dictionary maintenance method characterized by the following. 6. A language processing unit comprising: a language processing unit; and a word dictionary unit in which words used for language processing by the language processing unit are stored.
A character recognition method in a character recognition device that performs character recognition processing, comprising: a step of giving an expiration date to the word; and a step of ignoring a word whose expiration date has passed in the course of the character recognition processing. A character recognition method characterized in that: 7. A language processing unit comprising: a language processing unit; and a word dictionary unit in which words used for language processing by the language processing unit are stored.
What is claimed is: 1. A character recognition method for a character recognition device performing a character recognition process, comprising: providing an expiration date to the word; And a character recognition method. 8. A word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. Registering the
A user inputting an expiration date for the word registered in the word dictionary unit. 9. A word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. Registering the
Calculating the expiration date of the word registered in the word dictionary unit from the date and time when the word was registered in the word dictionary unit. 10. A word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. Registering the
Calculating the expiration date of the word registered in the word dictionary unit based on the characteristics of the word registered in the word dictionary unit. 11. A word dictionary maintenance method in an information processing apparatus having a language processing unit and a word dictionary unit in which words used for language processing by the language processing unit are stored. A step of automatically checking the time via a communication line. 12. The word dictionary maintenance method according to claim 1, wherein information on an expiration date of a word stored in the word dictionary unit is stored in a server connected to the word dictionary unit via a communication line, A word dictionary maintenance method, comprising: deleting a word whose expiration date has passed from the word dictionary unit based on information about the expiration date stored in the server. 13. A computer-readable storage medium in which a program for executing the word dictionary maintenance method according to claim 1 is stored. 14. A word dictionary maintenance method in an information processing apparatus having a language processing section and a word dictionary section in which words used for language processing by the language processing section are stored, wherein the words stored in the word dictionary section A word maintenance means for assigning an expiration date to the word dictionary, wherein the word dictionary maintenance means deletes the expired word from the word dictionary unit when the expiration date of the word has passed, and A word dictionary maintenance method, which selects and executes any one of updating the expiration date of an expired word.