JP3230641B2 - String search device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string search apparatus for use in an electronic apparatus having a character reading device and for searching a database for data including a character string matching a search character string.

[0002]

2. Description of the Related Art Conventionally, a database search may be performed on a character string such as a title, a keyword, or a data body of each data in the database.
Search for such a string (string searchin
g] or string pattern matching)
Specify the search string (pattern [pattern]), and the search target string (text [text]) of the field to be searched (search key [search key]) in each data matches this search string This is to determine whether or not it contains a character string. A conventional character string search process will be described. In this search processing, as shown in FIG. 10, a search character string P is designated and a search is performed on the search target character string T of each data in the database. Here, each character is indicated by □. Also, the search string P
Sets the number of characters to m (m = 4 in the figure), the i-th character is represented by P [i], the searched character string T has n characters, and the i-th character is T [ i]. The search process checks whether or not the search character string P matches each partial character string of the number m of characters included in the search target character string T. This partial character string exists only when the search target character string T is equal to or more than the number of characters of the search character string P (n ≧ m), and there are (n−m + 1) pieces in the search target character string T. That is, the character string of T [1] to T [m] from the first character to the m-th character in the search target character string T becomes the first partial character string.
The character string of [2] to T [m + 1] becomes the second partial character string,
Thereafter, similarly, the character string of T [n-m + 1] to T [n] becomes the (n-m + 1) -th last partial character string.

As shown in FIG. 11, in a first step (hereinafter, referred to as "S") 51 of a retrieval process, first, a value of a top position top of a partial character string is initialized to "0". The value of the top position top indicates the number of the first character of the partial character string to be inspected from the first character T [1] of the searched character string T. Then, it is determined whether or not there are m characters in the number of characters after the head position top of the searched character string T having the number of characters n (S52). Are not matched, and the search process ends. At the start of the search process, the search result is determined to be inconsistent, and the process ends when the number n of characters of the searched character string T is smaller than the number m of characters of the search character string P.

If it is determined in S52 that there are m characters and a partial character string exists, the value of the counter i is initialized to "1" (S53), and the i-th character in the search character string P It is determined whether or not the character P [i] matches the i-th character T [top + i] from the top position top of the partial character string in the search target character string T (S54). If these characters match, "1" is added to the counter i (S55), and the process returns to S54 until the value of the counter i exceeds the number m of characters of the search character string P, and this processing is repeated. It repeats (S56). If it is determined in step S56 that the value of the counter i has exceeded the number of characters m, the search processing is determined to be a match and the search process ends. Therefore, when all the characters of the partial character string of m characters from the head position top in the search target character string T match all the corresponding characters at the same character position of the search character string P, the loop of S54 to S56 is performed. After repetition of m times, it is determined that the search result is a match and the process ends.

However, if even one character mismatch is detected during the loop from S54 to S56, the process exits from this loop in S54. Then, the head position top of the searched character string T is advanced by one character to set the next partial character string (S57), and the process returns to S52 to repeat the above processing. Therefore, in the loop of S52 to S57, FIG.
From the first partial character string T [1] to T [m] when the head position top = 0, and n when the head position top = nm
It is checked whether or not each of the m + 1-th partial character strings T [n-m + 1] to T [n] matches the search character string P. If found, the process is exited from the loop in S56, and the search process is terminated assuming that the search result is a match. On the other hand, if there is no match with any of the partial character strings, it is determined in S52 that there are no more partial character strings, and the process exits from the loop and ends the search processing as a mismatch.

In the search for a character string in a database, the above character string is searched for a character string to be searched in each data, and the search character string matches any of the partial character strings of the character string to be searched. If the processing ends, data having the searched character string is extracted.

In the algorithm shown in FIG. 11, there is no use in repeatedly comparing the same character. Therefore, in the actual character string search, the BM method [Boyer-Moore string pattern matc
hingalgorithm] is often used.

[0008]

Incidentally, recent electronic devices include a coordinate input device (position input device) such as a tablet.
[pointing device])
written character reader] (handwritten character recognition device [hand
Many have an optical character reader (OCR [Optical Character Reader]) using an image input device such as a written character recognition system] or an image scanner. These character reading devices perform character recognition by selecting a standard pattern that matches a feature extracted from an input pattern, and output a character code of the standard pattern. However, when a character string is input by such a character reading device, it is inevitable that some character codes cannot be completely determined. In this case, a plurality of possible candidate characters must be listed. become. In some cases, erroneous recognition of a character occurs, and a character code different from the input character is determined.

However, in the conventional character string search, the search process cannot be executed unless the character codes of both the search character string and the searched character string are determined. For this reason,
In an electronic device that inputs a search character string or a character string to be searched using a character reading device, if character recognition is incomplete, a confirmation operation must be performed to determine which candidate character is the correct input character. You need to perform a search operation. However, since this operation requires the operator to interactively select one of the listed candidate characters,
This is extremely troublesome. In addition, when a large number of character strings to be searched in the database are input from printed type documents or the like, this determination operation may be substantially impossible.

In a conventional character string search, it is impossible to determine that the search character string matches the search target character string unless the character codes completely match. Does not execute the search process correctly.

For this reason, there has been a problem that a conventional electronic device having a character reading device cannot effectively utilize a character string search.

An object of the present invention is to solve the above-mentioned conventional problem, and an object of the present invention is to provide a character string search apparatus capable of searching a character string whose character is not determined by calculating the degree of coincidence. .

[0013]

According to the character string retrieval apparatus of the present invention, each character is composed of one or more candidate characters, and an uncertain character string in which a certainty value is associated with each candidate character. Search character string input means for inputting as a search character string, a database consisting of a set of data having the search target character string whose characters are determined, and the search character included in each search target character string in each data in the database. For each partial character string having the same number of characters as the column, if each character of the partial character string matches any of the candidate characters of the corresponding search character string at the same character position, the candidate matched for each character Matching degree calculating means for calculating the matching degree of the partial character string based on the certainty value of the character; and determining the matching degree of each partial character string calculated by the matching degree calculating means to the matching degree of another partial character string or a predetermined degree. Value and And compare the data with the search string containing selected portions string based on the comparison result and which was equipped with a data extracting means for extracting from said database, said object is achieved.

[0014] Preferably, a search character string input means for inputting a search character string whose character has been determined, and each character is composed of one or more candidate characters, and a certainty value is associated with each candidate character. And a partial character string having the same number of characters as the search character string included in each search character string in each data in the database. If any of the candidate characters of each character of the partial character string matches the character of the corresponding search character string at the same character position, the relevant part is determined based on the certainty value of the candidate character matched for each character. A degree-of-match calculating means for calculating the degree of matching of the character strings; and a degree of matching of each partial character string calculated by the degree-of-match calculating means is compared with a degree of matching of another partial character string or a predetermined value. To The data having the searched string containing selected portions string Hazuki and a data extracting means for extracting from the database.

Further, preferably, each character is 1 or 2
A search character string input means for inputting, as a search character string, an uncertain character string composed of the above candidate characters and associated with a certainty value for each candidate character; And a database consisting of a set of data having a character string to be searched, which is an uncertain character string in which a certainty value is associated with each candidate character, and a character string included in each character string to be searched in each data in the database. For each partial character string having the same number of characters as the search character string, any candidate character of each character of the partial character string matches one of the candidate characters of the corresponding search character string at the same character position. In this case, a matching score calculating means for calculating the matching score of the partial character string based on the likelihood value of both candidate characters matched for each character; and Data extracting means for comparing the matching degree with a matching degree of another partial character string or a predetermined value, and extracting data having a searched character string including the partial character string selected based on the comparison result from the database; .

Further, preferably, the search character string input means identifies each character of the input character string by a character reader, selects one or more candidate characters for each character, and selects each selected character. A certainty value indicating the accuracy of recognition is added to each candidate character.

Preferably, each character string to be searched for each data in the database identifies each character of the input character string by a character reading device, and selects one or more candidate characters for each character. In addition, a certainty value indicating the accuracy of recognition is added to each selected candidate character and input.

[0018]

[0019]

According to the above arrangement, the degree-of-coincidence calculating means can calculate, as the degree of coincidence, the degree of matching of each partial character string of the character string to be searched with the search character string composed of the uncertain character string. Moreover, even if a character that does not match any of the candidate characters is detected, the processing of the matching degree calculating means is continued up to the last character, so that the matching degree calculation is performed even on the partial character string including the mismatching character. Can be calculated. Further, the data extracting means detects the maximum matching degree of each partial character string by mutually comparing the matching degree of each partial character string with the matching degree of another partial character string or by comparing it with a predetermined value. , Sort each substring in descending order of matching [sorti
ng] or selecting a partial character string having a matching degree equal to or more than a predetermined value, and extracting data desired by the operator from the database.

Therefore, even if the search character string is an uncertain character string in which each character has not been determined, it is possible to perform a search on the search target character string of each data in the database without any change. It is possible to extract the highest data or to sort and extract data having a matching degree equal to or higher than a predetermined value in descending order of the matching degree. In addition, even when a part of the character includes a non-character, if the other character matches and a high degree of matching is obtained, the data can be extracted.

The candidate characters usually have a specific character code, but it is also possible to use a character representing a specific character type or a metacharacter that matches all characters in the character string to be searched. It is possible. The database is
The term is used in a broad sense, including not only a set of data defined by the database schema but also a file composed of text data. In this case, each data can be, for example, each line in the text data.

The certainty value is explicitly associated with each candidate character, or may be associated so that it is automatically determined according to the total number of candidate characters for each character or the order of the candidate character. Good. The certainty value may be a score that is higher as the possibility that the corresponding candidate character is actually the character is higher, or may be a numerical value such as a probability indicating that the character is actually the character. For the degree of coincidence of each partial character string, an appropriate initial value is given first, and each time a character of the partial character string matches a candidate character of the search character string, the degree of coincidence calculating means calculates based on the probability value. Is applied. If the certainty value is a score, for example, if the initial value of the matching degree is set to 0 and an operation of sequentially adding the certainty values is performed, the partial character string matches the more likely candidate character of the search character string. The higher the score, the higher the degree of coincidence. In addition, for non-matching characters, the addition of the certainty value is not performed, so that a high matching score can be suppressed. Also, the accuracy value is 0
In the case of the probability of having a value of 1 or more and 1 or less, for example, if the initial value of the matching degree is set to 1 and an operation of sequentially multiplying the respective probability values is performed, the probability that the partial character string matches the search character string is high. The degree of coincidence closer to 1 can be obtained. However, in this case, even when there is a mismatched character, it is necessary to perform an operation of multiplying the degree of match by a predetermined sufficiently low probability.

When the data extracting means compares the degree of coincidence of each partial character string with the degree of coincidence of another partial character string, the same as when comparing the degree of coincidence of another partial character string in the same searched character string. There may be a case where the comparison is made with the matching degree of a partial character string in another searched character string in data, and a case where the comparison is made with the matching degree of a partial character string in another searched character string in other data. When comparing with the matching degree of the same string to be searched in the same data or a partial character string of another string to be searched, the maximum matching degree is generally obtained. The comparison between them is for performing sorting such as arranging in descending order of the degree of coincidence.

Further, with the above configuration, the matching degree calculating means can calculate, as the matching degree, the degree of matching between each partial character string of the searched character string composed of the uncertain character string and the search character string. Therefore, even if the search target character string of each data in the database is an undetermined character string in which each character has not been determined, it is possible to perform a search using the search character string as it is.

Further, according to the above configuration, the matching degree calculating means sets the degree of matching of each partial character string of the searched character string consisting of the uncertain character string with the search character string consisting of the uncertain character string as the matching degree. Can be calculated. Therefore, even if the search character string is an uncertain character string in which each character has not been determined, and the searched character string of each data in the database is the same uncertain character string, the search is performed as it is. It becomes possible.

Further, according to the above configuration, a search character string composed of an uncertain character string can be input by a character reading device. Therefore, for example, in an electronic device equipped with a handwritten character reading device, when a search character string is designated by handwriting, even when a plurality of candidate characters are selected, the work of confirming the characters is not performed, and erroneous recognition is not performed. Even when there is an error, the search can be executed immediately without correcting the character.

Further, according to the above configuration, it is possible to input a character string to be searched consisting of an uncertain character string by a character reading device. Therefore, for example, in an electronic device equipped with an optical character reading device, it is possible to mechanically read a printed type document or the like to obtain a character string to be searched for each data in a database without performing a character fixing operation or a correction operation. In this case, the search can be performed immediately.

Further, according to the above configuration, it is possible to search only when the number of characters of the searched character string matches the number of characters of the searched character string.

[0029]

Embodiments of the present invention will be described below.

FIGS. 1 to 5 show a first embodiment of the present invention. FIG. 1 is a block diagram showing a configuration of a character string search device. FIG. FIG. 3 is a block diagram showing a configuration, FIG. 3 is a flowchart showing an operation of a search process in the character string search device, FIG.
FIG. 4 is a diagram showing a state where candidate characters of a search character string match characters of a partial character string.

In this embodiment, an electronic apparatus including a character string search device for searching a database based on a search character string of an uncertain character string will be described. This electronic device is shown in FIG.
As shown in FIG. 1, an arithmetic unit 1 for controlling the entire apparatus and performing various calculations is provided. The arithmetic unit 1 is connected to a main storage device 2 composed of a semiconductor memory or the like. The main storage device 2 loads a program such as a character string search device for execution, and also secures a work area for storing a search character string during execution of the program and for storing a score of the degree of coincidence. Used for etc. The arithmetic unit 1 includes an auxiliary storage device 3 such as a hard disk device or a floppy disk device, and a CRT [Cathode Ray Tube].
A display and a display device 4 such as an LCD [Liquid Crystal Display] are connected. The auxiliary storage device 3 is used for storing programs such as a character string search device and a database. The display device 4 is used for displaying an input search character string and displaying search results during execution of a program of the character string search device. Further, a tablet 5 and an image scanner 6 are connected to the arithmetic device 1 as input devices. The tablet 5 is a coordinate input device for inputting a pattern such as a stroke of a handwritten character using the input pen 7, and executes the handwritten character reading device program stored in the auxiliary storage device 3. Can be recognized as a character code and stored in the main storage device 2 as a search character string. The image scanner 6 is an image input device for optically reading a printed type document or the like of a printed matter and inputting it as an image pattern. The image scanner 6 executes a program of the optical character reading device stored in the auxiliary storage device 3 to execute the type printing. Such an image pattern can be recognized as a character code and stored in the auxiliary storage device 3 as a searched character string of each data in the database. Note that the tablet 5 may be integrated with the display device 4. Further, instead of the tablet 5, another coordinate input device such as a mouse can be used.

FIG. 1 shows the configuration of a character string search device that executes a database search process in an electronic device having the above hardware configuration. This character string search device is a tablet 5
The pattern of each character handwritten using the input pen 7 is recognized by the handwritten character reading device 11 and input to the search processing unit 12 as a search character string. However, in this embodiment,
Even when the handwritten character reading device 11 cannot determine the handwritten input character, the handwritten character is input as an undetermined character string without performing the determination operation. In the uncertain character string, each character is 1 or 2
It is a character string composed of the above candidate characters and associated with a certainty value for each candidate character. When a handwritten input character is determined, the character is composed of only one candidate character, and a score of a perfect score is associated as a certainty value. If it is not possible to determine which candidate character the handwritten character corresponds to, a plurality of candidate characters are enumerated, and the higher the possibility that the handwritten character corresponds to the handwritten character, the closer the score to the full score. Are associated. However, not only when there are multiple candidate characters but also when there is only one character,
Erroneous recognition in which the actual handwritten input character does not correspond to any of the candidate characters may occur.

The indeterminate character string can be realized by an arbitrary data structure. That is, for example, when the maximum number of candidate characters for each character is determined, it can be realized by a two-dimensional array. In this case, the element of each character is specified by the subscript of the first dimension, and the element of each candidate character is specified by the subscript of the second dimension. The likelihood values can also be associated by storing them in a numerical two-dimensional array having the same number of elements. If the certainty value can be represented by the same data type as the character type, the uncertain character string may be a three-dimensional array, and the elements of the certainty value may be designated by all the subscripts of the last dimension. Further, the accuracy value is, for example, 1 candidate character.
If there are only characters, the score is a perfect score. If two characters, the first candidate character has a score of 60% of the full score and the subsequent candidate character has a score of 40% of the full score. Can be automatically determined in accordance with the total number of candidate characters, the order of the candidate characters, and the like.
Therefore, the certainty value in this case is not attached to the uncertain character string as data, but is associated as an algorithm.

For example, one or more candidate characters of each character are arranged in order, a special delimiter is determined, and the delimiter is inserted between the characters so that a character delimiter can be determined. Then, the uncertain character string can be handled as sequential data such as a one-dimensional array, and the number of candidate characters for each character can be unlimited. Also in this case, the accuracy value can be arranged immediately after each candidate character, stored in another sequential data, or associated as an algorithm. In addition, this uncertain character string has a tree structure [tree st
ructure].

The search processing unit 12 searches for a character string to be searched in each data in the database stored in the auxiliary storage device 3 based on the search character string of the uncertain character string. When searching this database, a search target column in each data is designated as a search key. The search key field is a data title, a keyword, or the data itself. In the case of the title field, the character string of the title name is the character string to be searched, and in the case of the keyword field, the character of each listed keyword is displayed. Each column is a searched character string, and in the case of a data body, each character string constituting this data body is a searched character string. The search processing of the search processing unit 12 first calculates a matching degree, which will be described later, based on the likelihood value of the matching candidate character for each partial character string having the same number of characters as the search character string in the search target character string. The maximum score of the matching degree of each partial character string in the searched character string is obtained as the maximum matching degree. Then, the data with the highest maximum matching degree is displayed with priority on the search result list of the display device 4.

The operation of the retrieval processing unit 12 having the above configuration will be described with reference to the flowcharts of FIGS. The search processing unit 12 sequentially reads the searched character strings of each data from the database stored in the auxiliary storage device 3 and executes a search process for each searched character string. Here, the number of characters of the search character string P is m, and the number of characters of the search target character string T is n. In the search process, as shown in FIG. 3, first, the score of the maximum matching score is initialized to "0" (S1), and the value of the top position top of the partial character string is also initialized to "0" (S2). . The value of the top position top indicates the number of the first character of the partial character string to be inspected from the first character of the searched character string T. And
It is determined whether or not there are m characters in number after the start position top of the searched character string T having n characters (S3). If there are not enough m characters and there is no partial character string, the search process ends. I do. The number of characters n of the search target character string T is the search character string P
If the number of characters is less than m, the process is immediately terminated at the start of the search process, and the score of the maximum matching degree remains at the initial value.

If it is determined in step S3 that there are m characters and a partial character string exists, m
Calculation processing of the degree of coincidence is performed on the partial character strings for the characters (S4). Then, the degree of coincidence of the partial character string calculated in S4 is compared with the maximum degree of coincidence (S5). If the score of the maximum degree of coincidence is lower, the score of the maximum degree of coincidence is calculated in S4. (S
6) If the score of the maximum matching score is higher or the score is the same, the head position top is moved forward by one character, the next partial character string is set (S7), and the process returns to S3. This processing is repeated until the processing of all the partial character strings of the searched character string T is completed. Therefore, in this search process,
A partial character string of m characters is sequentially cut out from the head of the search target character string T, and the degree of coincidence with the search character string P is calculated. At the end of the search processing, the maximum score of the matching degree of each partial character string in the searched character string T is stored in the maximum matching degree.

The details of the process of calculating the degree of coincidence in S4 will be described with reference to FIG. Here, the j-th candidate character of the i-th character in the search character string P is P
[i, j], and the i-th character in the searched character string T is represented by T [i]. At the start of the coincidence calculation process, first, the score of the coincidence is initialized to "0" (S11), the value of the counter i is initialized to "1" (S12), and the value of the counter j is also set to "1". (S
13). Then, the j-th candidate character P [i, j] of the i-th character in the search character string P is the i-th character T [top from the top position top of the partial character string in the search target character string T.
+ i] is determined (S14).

If it is determined in S14 that the candidate character P [i, j] does not match the character T [top + i], "1" is added to the counter j (S15), and the next candidate character P [i, j] is added. P [i,
j] is determined (S16), and if the next candidate character P [i, j] exists, the process returns to S14 to repeat this process. Therefore, each character T [top + i] of the partial character string is sequentially compared with one or more candidate characters P [i, j] corresponding to the same character position in the search character string P here. The presence or absence of the next candidate character P [i, j] in S16 can be detected by a method according to the data structure of the uncertain character string. For example, when using a two-dimensional array,
If a special empty symbol is stored in an element having no candidate character, the presence or absence of the next candidate character can be detected by checking whether the candidate character P [i, j] is this empty symbol.
In the case of sequential data using a delimiter, the presence or absence of the next candidate character can be detected by checking whether the character next to the immediately preceding candidate character is this delimiter.

In S14, when it is determined that the candidate character P [i, j] matches the character T [top + i], the comparison with the remaining candidate characters is terminated, and the candidate character P [i, j] is terminated. j] is added to the score of the degree of coincidence (S1
7). When the coincidence is added in S17,
Alternatively, if it is determined in S16 that the comparison with all the candidate characters of the character has been completed, "1" is added to the counter i and the process proceeds to the next character (S18), and the value of the counter i is searched. Until the number m of characters of the character string P is exceeded, S1
3, the process is repeated after the counter j is reinitialized (S19). If it is determined in step S19 that the value of the counter i has exceeded the number m of characters, the process of calculating the degree of coincidence is terminated. Therefore, when each character of the partial character string matches any of the corresponding candidate characters at the same character position in the search character string P, the certainty value of the candidate character is sequentially added to the score of the degree of coincidence. In addition, even if no match is found with any of the candidate characters, the score is not added, but the processing is continued for each subsequent character. For this reason,
The score of the matching degree of the partial character string in which no character matched each candidate character of the search character string P remains at the initial value “0”, and generally, the score of the matching degree increases as the number of matching characters increases. Become. Also, even when the characters at the same character position in the partial character string match, the degree of matching changes depending on which candidate character matches, and the higher the possibility of matching the candidate character, the higher the score. .

FIG. 5 shows an example of how the candidate character of each character of the search character string P and each character of the partial character string included in the search target character string T match in the above-described matching degree calculation processing.
Here, each character or candidate character is indicated by □, and a matching character or candidate character is indicated by a black circle inside □. In addition, the search character string P has four characters (m = 4),
The first character has up to the fifth candidate character, and the second character is 3
It is assumed that there are up to the fourth candidate character, the third character has the fourth candidate character, and the fourth character has the fifth candidate character. In the first loop of the counter i = 1, when the counter j = 2, the second candidate character P [1,2] of the first character of the search character string P is replaced with the first character T [of the partial character string. to
p + 1], the likelihood value of the candidate character P [1,2] is added to the score of the matching degree in S17. In the loop of the counter i = 2, when the counter j = 1, the first candidate character P [2,1] of the second character of the search character string P is the partial character string 2
Since the character T matches the character T [top + 2], the certainty value of the candidate character P [2,1] is added to the score of the matching degree. In the loop of the counter i = 3, when the counter j = 2, the second candidate character P [3,2] of the third character of the search character string P is replaced with the third character T [top + 3], the certainty value of the candidate character P [3,2] is added to the score of the matching degree. Then, in the loop of the counter i = 4, the counter j
= 3, the fourth candidate character P [4,3] of the fourth character of the search character string P matches the fourth character T [top + 4] of the partial character string. The certainty value of P [4,3] is added to the score of the coincidence. Therefore, the degree of coincidence calculated here is a certainty value of candidate character P [1,2], candidate character P [2,1], candidate character P [3,2], and candidate character P [4,3]. Is the sum of

When the search processing unit 12 completes the search process for the searched character string, the search unit 12 refers to the score of the maximum matching degree and determines the score of the data. Further, when there are a plurality of searched character strings in the same data, the largest one of the scores of the maximum matching degree is set as the score of the data. Then, for example, only data whose score is equal to or more than a predetermined value are extracted from the database, and these data are displayed on the search result list of the display device 4 in descending order of the score. In FIG. 1, two data are extracted from the database stored in the auxiliary storage device 3 by the search processing of the search processing unit 12, and the highest score data 2 and the next highest score data are displayed in the search result list of the display device 4. 5 shows the displayed state.

As described above, according to the character string search device for electronic equipment of this embodiment, the tablet 5 and the input pen 7 are used.
When a search character string is input by the handwritten character reading device 11 using, even if an unconfirmed character or an erroneously recognized character is included in the search character string, the character is determined or corrected. Search can be performed without any need. That is, even if there is an undetermined character consisting of a plurality of candidate characters in the search character string, if the corresponding character of the search target character string matches any of the candidate characters, the matching score is high, so Data having a search character string can be reliably extracted. In addition, even when the handwritten character reading device 11 erroneously recognizes a part of the handwritten character, a relatively high score can be obtained if the remaining other characters are surely matched. It is more likely that data having a searched character string that matches the column can be extracted.

In the process of calculating the degree of coincidence shown in FIG. 4, it is normal that most of the partial character strings are actually completely different from the search character string. In addition, if a character that does not match any of the candidate characters is detected by a certain number or more, even if the calculation is continued up to the last character, the matching score cannot be high. Therefore, in order to speed up the processing, when a predetermined number or more of unmatched characters are detected, the calculation processing of the degree of matching of the subsequent characters may be terminated for the partial character string.

When performing a search for uniformly extracting all the data of the searched character string in which the highest matching score is equal to or more than the predetermined value, regardless of the score, the search processing shown in FIG. When a partial character string having a degree of coincidence equal to or greater than the predetermined value is detected, it is determined that the data has already been extracted, and the calculation processing of the degree of coincidence in S4 for the subsequent partial character string is terminated. You can also do so.

FIGS. 6 and 7 show a second embodiment of the present invention. FIG. 6 is a flowchart showing the operation of the calculation processing of the degree of coincidence in the search processing. FIG. It is a figure showing the state where a candidate character of a character string matches. Note that components having the same functions as those of the first embodiment shown in FIGS. 1 to 5 are denoted by the same reference numerals.

In the present embodiment, an electronic apparatus having a character string search device for searching a character string to be searched for an undefined character string in a database based on a search character string in which each character is determined will be described. The hardware configuration of this electronic device is the same as that of the first embodiment shown in FIG. The configuration of the character string search device may be the same as that of the first embodiment shown in FIG. However, in this embodiment, the searched character string in each data of the database stored in the auxiliary storage device 3 is constituted by the same uncertain character string as shown in the first embodiment. Further, in the present embodiment, when the handwritten character reading device 11 cannot determine the handwritten input character at the time of inputting the search character string by the tablet 5 and the input pen 7, the determination operation is performed to determine all the characters. Then, it is input to the search processing unit 12. Therefore, the search character string may be input from a character input device such as a keyboard (not shown). The character string to be searched in the database is, for example, a character type document of a printed matter read as an image pattern by the image scanner 6 shown in FIG. And The input may be input by the handwritten character reading device 11 using the tablet 5 and the input pen 7.

The search processing unit 12 according to the present embodiment searches for a character string to be searched consisting of an uncertain character string in each data in the database stored in the auxiliary storage device 3 based on the search character string in which each character is determined. I do. The search processing of the search processing unit 12 is the same as that of the first embodiment shown in FIG. 3, but the content of the process of calculating the degree of coincidence in S4 is different from that of the first embodiment shown in FIG. different.

The details of the process of calculating the degree of coincidence in this embodiment will be described with reference to FIG. Here, the i-th character in the search character string P is represented by P [i], and the j-th candidate character of the i-th character in the search target character string T is T [i, j]. Shall be represented. At the start of the process of calculating the coincidence, the score of the coincidence is initialized to "0" (S21), and the value of the counter i and the value of the counter j are initialized to "1" (S22, S23). And
The i-th character P [i] in the search character string P matches the j-th candidate character T [top + i, j] of the i-th character from the top position top of the partial character string in the search target character string T It is determined whether or not to perform (S24).

In S24, the character P [i] is replaced with the candidate character T
If it is determined that they do not match [top + i, j], “1” is added to the counter j (S25), and the next candidate character T [top
+ i, j] is determined (S26), and if the next candidate character T [top + i, j] exists, the process returns to S24 to repeat this process. Therefore, each character P [i] of the search character string P is sequentially compared with one or more corresponding candidate characters T [top + i, j] at the same character position in the partial character string.

In S24, the character P [i] is replaced with the candidate character T
If it is determined that the character string matches [top + i, j], the comparison with the remaining candidate characters is terminated, and the candidate character T [top + i, j] is canceled.
j]] is added to the score of the degree of coincidence (S27). If the degree of coincidence is added in S27, or if it is determined in S26 that the comparison with all the candidate characters in the character has been completed, “1” is added to the counter i and the next character is added. (S28), and until the value of the counter i exceeds the number m of characters of the search character string P,
Returning to S23, the counter j is reinitialized, and this process is repeated (S29). If it is determined in step S29 that the value of the counter i has exceeded the number of characters m, the process of calculating the degree of coincidence is terminated. Therefore, when each character of the search character string P matches any of the corresponding candidate characters at the same character position in the partial character string, the certainty value of the candidate character is sequentially added to the score of the degree of coincidence. In addition, even if no match is found with any of the candidate characters, the score is not added, but the processing is continued for each subsequent character. Therefore, as in the case of the first embodiment, the higher the degree of matching with the search character string P, the higher the score of each partial character string.

FIG. 7 shows an example in which each character of the search character string P matches a candidate character of each character of the partial character string included in the search target character string T in the above-described matching degree calculation processing.
Here, each character or candidate character is indicated by □, and the matching character or candidate character is indicated by a black circle inside □.
Further, the number of characters of the search character string P is set to 4 characters (m = 4),
The first character of the substring to be searched T is the seventh candidate character, the second character is the fifth candidate character, the third character is the second candidate character, and the fourth character is the fourth character. The character is assumed to have up to the third candidate character. Counter i
In the first loop of = 1, when the counter j = 2, the first character P [1] of the search character string P is changed to the second candidate character T [top + 1, 2]
In S27, the certainty value of the candidate character T [top + 1,2] is added to the score of the matching degree. In the loop of the counter i = 2, when the counter j = 3, the second character P of the search character string P
[2] is the third candidate character T [top +
2,3], the certainty value of the candidate character T [top + 2,3] is added to the score of the matching degree. In the loop of the counter i = 3, when the counter j = 1, the third character P [3] of the search character string P becomes the first candidate character T [top + 3, 1], the candidate character T [to
p + 3,1] is added to the score of the coincidence. And
In the loop of the counter i = 4, when the counter j = 3, the fourth character P [4] of the search character string P becomes the third candidate character T [top + 4, 3], the likelihood value of the candidate character T [top + 4,3] is added to the score of the matching degree. Therefore, the degree of matching calculated here is that the candidate character T [top + 1,2], the candidate character T [top + 2,3], the candidate character T [top + 3,1], and the candidate character T [top + 4,3].

When the search processing unit 12 completes the search process for the searched character string, the score of the maximum matching score is used as the score of the data, as in the case of the first embodiment. For example, this score is equal to or higher than a predetermined value. Only data is extracted from the database, and these data are displayed on the search result list of the display device 4 in descending order of the score.

As described above, according to the character string search device of the electronic apparatus of the present embodiment, the character string to be searched is input into each data of the database by the optical character reading device using the image scanner 6. In such a case, even if an unconfirmed character or an erroneously recognized character is included in the searched character string, the search can be executed without performing the character fixing operation or the correction operation. Therefore, it is possible to perform a search using a search character string even when a large number of printed type documents or the like are mechanically read and stored in a database.

The calculation processing of the degree of coincidence shown in FIG. 6 and the search processing shown in FIG. 3 are performed in the same manner as in the first embodiment, when a predetermined number of unmatched characters are detected or when the degree of coincidence is detected. When the score becomes equal to or more than a predetermined value, the processing can be appropriately terminated midway for speeding up.

FIGS. 8 and 9 show a third embodiment of the present invention. FIG. 8 is a flowchart showing the operation of the calculation of the degree of coincidence in the search process. FIG. It is a figure showing the state where candidate character of a partial character string matches. Note that components having the same functions as those of the first embodiment shown in FIGS. 1 to 5 are denoted by the same reference numerals.

In this embodiment, an electronic apparatus equipped with a character string search device for searching a character string to be searched for an uncertain character string in a database based on a search character string of an uncertain character string will be described. The hardware configuration of this electronic device is the same as that of the first embodiment shown in FIG. The configuration of the character string search device is the same as that of the first embodiment shown in FIG. 1, and the search character string input by the handwritten character reading device 11 using the tablet 5 and the input pen 7 is an uncertain character string. It is configured. However, in the present embodiment, similarly to the case of the second embodiment, the character string to be searched in each data of the database stored in the auxiliary storage device 3 is also constituted by the uncertain character string.

The search processing unit 12 according to the present embodiment searches for a character string to be searched consisting of an uncertain character string in each data in a database stored in the auxiliary storage device 3 based on the uncertain character string search character string. I do. The search processing of the search processing unit 12 is the same as that of the first embodiment shown in FIG. 3, but the content of the process of calculating the degree of coincidence in S4 is the same as that of the first embodiment shown in FIG. This is different from the case of the second embodiment shown.

Details of the process of calculating the degree of coincidence in this embodiment will be described with reference to FIG. Here, the j-th candidate character of the i-th character in the search character string P is represented by P [i, j].
The k-th candidate character of the character is represented by T [i, k]. At the start of the process of calculating the coincidence, the score of the coincidence is initialized to "0" (S31), and the value of the counter i, the value of the counter j, and the value of the counter k are each initialized to "1". (S32-S34). Then, the j-th candidate character P of the i-th character in the search character string P
[i, j] is the k-th candidate character T [top + of the i-th character from the head position top of the partial character string in the search target character string T
i, k] (S35).

In S35, when it is determined that the candidate character P [i, j] does not match the candidate character T [top + i, k],
"1" is added to the counter k (S36), and it is determined whether or not the next candidate character T [top + i, k] exists (S37).
If the next candidate character T [top + i, k] exists, the process returns to S35 and repeats this process. Also, the next candidate character T [top
+ i, k] does not exist, the counter j
Is added (S38) to determine whether or not the next candidate character P [i, j] exists (S39).
If [i, j] exists, the process returns to S34 to reinitialize the counter k, and then repeats this process. Therefore, here, one or more candidate characters P [i, j] of each character of the search character string P and one or more candidate characters T [top + i, k] are brute force compared in order.

If it is determined in S35 that the candidate character P [i, j] matches the candidate character T [top + i, k], the comparison with the remaining candidate characters is terminated, and the candidate character P [ i, j]
And the respective probabilities associated with the candidate characters T [top + i, k] are added to the score of the degree of coincidence (S40). Then, if the degree of coincidence is added in S40, or if it is determined in S39 that the comparison of all candidate characters in the character has been completed, “1” is added to the counter i and the next character is added. Proceeding (S41), the process returns to S33 and re-initializes the counters j and k until the value of the counter i exceeds the number m of characters of the search character string P, and then repeats this process (S41).
42). If it is determined in S42 that the value of the counter i has exceeded the number m of characters, the process of calculating the degree of coincidence is terminated. Therefore, if any of the candidate characters of each character of the search character string P matches any of the corresponding candidate characters at the same character position in the partial character string, the likelihood value of both of the matched candidate characters is Are sequentially added to the score. In addition, even if no match is found with any of the candidate characters, the score is not added, but the processing is continued for each subsequent character. Therefore, as in the case of the first and second embodiments, the higher the degree of matching with the search character string P, the higher the score of each partial character string.

FIG. 9 shows an example of how the candidate character of each character of the search character string P and the candidate character of each character of the partial character string included in the search target character string T match in the above-described matching degree calculation processing. . Here, each character or candidate character is indicated by □, and the matching character or candidate character is indicated by a black circle inside □. In addition, the search character string P has four characters (m
= 4), and each character has the same number of candidate characters as in the case of the first embodiment shown in FIG. Also,
It is assumed that each character of the partial character string has the same number of candidate characters as in the second embodiment shown in FIG. In the first loop of the counter i = 1, the counters j = 2, k =
When 2, the second candidate character P [1,2] of the first character of the search character string P matches the second candidate character T [top + 1,2] of the first character of the partial character string Therefore, in S40, the certainty values of these candidate characters P [1,2] and candidate characters T [top + 1,2] are added to the score of the matching degree. In the loop of the counter i = 2, when the counters j = 1 and k = 3, the first candidate character P [2,1] of the second character of the search character string P is the third character of the second character of the partial character string.
Since the candidate character T [top + 2,3] matches, the certainty value of these candidate characters P [2,1] and candidate character T [top + 2,3] is added to the score of the degree of coincidence. . In the loop with the counter i = 3,
When the counters j = 2 and k = 1, the second candidate character P [3,2] of the third character of the search character string P becomes the first candidate character T [top + of the third character of the partial character string. 3, [1], the likelihood values of these candidate characters P [3,2] and T [top + 3,1] are added to the score of the matching degree. And the counter i = 4
When the counters j = 3 and k = 3, the fourth candidate character P [4,3] of the fourth character of the search character string P becomes the third candidate character of the fourth character of the partial character string. T [top + 4,3], these candidate characters P [4,3] and candidate character T [to
p + 4,3] is added to the score of the coincidence. Therefore, the degree of coincidence calculated here is a certainty value of candidate character P [1,2], candidate character P [2,1], candidate character P [3,2], and candidate character P [4,3]. And the candidate characters T [top + 1,2] and the candidate characters T [to
p + 2,3], the sum of the accuracy values of the candidate character T [top + 3,1] and the candidate character T [top + 4,3].

When the search processing unit 12 completes the search process for the search target character string, the score of the maximum matching score is used as the score of the data, as in the first and second embodiments. Is extracted from the database, and these data are displayed on the search result list of the display device 4 in descending order of score.

As described above, according to the character string search device of the electronic apparatus of the present embodiment, when a search character string or a character string to be searched is input by a character reading device, these character strings are determined. Even if a character that has not been performed or a character that has been misrecognized is included, the search can be performed without performing a character fixing operation or a correction operation.

In the calculation processing of the degree of coincidence shown in FIG. 8 and the search processing shown in FIG. 3, as in the first and second embodiments, a predetermined number or more of unmatched characters are detected. In such a case, or when the score of the degree of coincidence is equal to or more than a predetermined value, the processing can be appropriately terminated midway for speeding up.

In the search processing in the first to third embodiments, even if a character that does not match any of the candidate characters is detected, the comparison of the subsequent characters is continued. No algorithm can be used. However, for example, when the comparison of the subsequent characters is terminated when a predetermined number or more of mismatched characters are detected, B
It is also possible to use a high-speed algorithm applying the M method or the like.

Further, in the first and third embodiments, a search is performed by using a search character string for an uncertain character string. The uncertain character string is combined by selecting [union] each candidate character for each character. In addition, since each character can be represented as a regular expression [regular expression] combined by concatenation, a finite automaton [finite automaton] formed by this regular expression is used instead of the above search processing.
maton] can also be used. However, in this case, in addition to the state of each candidate character for each character, a state of accepting all characters other than these candidate characters is added, and a probability value is associated with the state of each candidate character. In addition, it is necessary to perform a process of sequentially adding the certainty values of the state that has actually transitioned by receiving a character from the partial character string.

In the search processing in the first to third embodiments, all the partial character strings included in the searched character string are searched, but the number of characters in the searched character string is reduced to the number of characters in the searched character string. Only when there is a match, the search can be performed with the entire searched character string as a partial character string.

[0069]

As described above, according to the character string search apparatus of the present invention, the matching degree between the search character string and each partial character string included in the searched character string is calculated based on the certainty value of the candidate character. ,
The higher the likelihood that the matched candidate character is a correct character, the more it can be extracted by comparing the degree of matching. Therefore, the uncertain character string of each character that has been input by a character reader or the like is searched for. The search can be performed as a column or a character string to be searched. At this time, either the search character string or the searched character string may be an indeterminate character string, or both may be indeterminate character strings. Moreover, even if some characters do not match all the candidate characters of the corresponding characters in the uncertain character string, a relatively high degree of matching can be obtained if the other characters match exactly. Also, the search can be performed even when character recognition cannot be performed at all or erroneous recognition has occurred. Therefore, for example, even when a search character string is input by a handwritten character reading device or a search target character string input by an optical character reading device, a search can be performed without performing a character fixing operation or a correction operation. become able to.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a character string search device according to a first embodiment of the present invention.

FIG. 2, showing the first embodiment of the present invention, is a block diagram illustrating a configuration of an electronic apparatus including a character string search device.

FIG. 3, showing the first embodiment of the present invention, is a flowchart showing an operation of a search process in a character string search device.

FIG. 4 is a flowchart illustrating the operation of calculating the degree of coincidence in the search processing according to the first embodiment of the present invention.

FIG. 5 illustrates the first embodiment of the present invention, and illustrates a state where candidate characters of a search character string match characters of a partial character string.

FIG. 6 is a flow chart showing a second embodiment of the present invention and showing an operation of a process of calculating a degree of coincidence in a search process.

FIG. 7 illustrates the second embodiment of the present invention, and illustrates a state in which characters of a search character string match candidate characters of a partial character string.

FIG. 8 is a flow chart showing a third embodiment of the present invention and showing an operation of a process of calculating a degree of coincidence in a search process.

FIG. 9 illustrates the third embodiment of the present invention, and illustrates a state where candidate characters of a search character string match candidate characters of a partial character string.

FIG. 10 illustrates a conventional example, and is a diagram illustrating a relationship between a search character string and each partial character string of a search target character string.

FIG. 11 shows a conventional example, and is a flowchart showing an operation of a character string search process.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 arithmetic unit 3 auxiliary storage device 5 tablet 6 image scanner 7 input pen 11 handwritten character reading device 12 search processing unit

Claims (5)

(57) [Claims] A search character string input means for inputting, as a search character string, an uncertain character string in which each character is composed of one or two or more candidate characters and a probability value is associated with each candidate character; A database consisting of a set of data having a character string to be searched for which characters are fixed; and for each partial character string having the same number of characters as the search character string included in each character string to be searched in each data in the database, If each character in the string matches any of the candidate characters in the corresponding search string at the same character position, the matching of the substring based on the likelihood value of the candidate character that matched for each character Means for calculating the degree of match, and the degree of match of each partial character string calculated by the degree of match calculation means is compared with the degree of match of another partial character string or a predetermined value, and selected based on the comparison result. Department A character string search device comprising: data extraction means for extracting data having a character string to be searched including a minute character string from the database. 2. A search character string input means for inputting a search character string in which a character is fixed, and a character string comprising one or more candidate characters, and a probability value associated with each candidate character. A database including a set of data having a searched character string that is a fixed character string; and a partial character string having the same number of characters as the search character string included in each searched character string in each data in the database. If any candidate character of each character in the partial character string matches the character of the corresponding search character string at the same character position, based on the certainty value of the candidate character matched for each character, Matching degree calculating means for calculating the matching degree; comparing the matching degree of each partial character string calculated by the matching degree calculating means with the matching degree of another partial character string or a predetermined value, and selecting based on the comparison result Part A character string search device comprising: data extraction means for extracting data having a character string to be searched including a minute character string from the database. 3. Search character string input means for inputting, as a search character string, an uncertain character string in which each character is composed of one or more candidate characters, and a probability value is associated with each candidate character, A database consisting of a set of data each having one or more candidate characters, and having a character string to be searched, which is an uncertain character string associated with a certainty value for each candidate character; For each partial character string having the same number of characters as the search character string included in each searched character string in each data, a search character string in which any candidate character of each character of the partial character string corresponds at the same character position A matching degree calculating means for calculating a matching degree of the partial character string based on the certainty value of both candidate characters matched for each character when the matching character matches any of the candidate characters; Every time The degree of coincidence of each partial character string calculated by the arithmetic means is compared with the degree of coincidence of another partial character string or a predetermined value, and data having a searched character string including the partial character string selected based on the comparison result is obtained. A character string search device comprising: data extraction means for extracting data from a database. 4. The search character string input means identifies each character of the input character string by a character reading device, and selects one or more candidate characters for each character.
4. The character string search device according to claim 1, wherein a certainty value indicating recognition accuracy is added to each of the selected candidate characters. 5. Each character string to be searched for each data in the database identifies each character of the input character string by a character reading device, and selects one or more candidate characters for each character. 3. A method according to claim 2, wherein a certainty value indicating the accuracy of recognition is added to each of said candidate characters.
5. The character string search device according to any one of-to 4.