JPH0460871A - Controlling system for symbol string collating device - Google Patents

Abstract

PURPOSE:To collate plural collating symbol strings shorter than some length and plural symbol strings to be collated with each other in parallel by changing a row from which a collated result is read according to the length of the collating symbol string. CONSTITUTION:As for collation in the case that the length of the collating symbol string is an optional positive integer L below N, symbol string collation is executed by executing initialization by setting the storage data of a cell f11 to '1' after making all the cell values of a cell array into '0', and simultaneously, giving the symbol string to be collated successively by every one symbol. Namely, by searching the cell whose storage data is '1' and whose line number is the smallest from among M-pieces of the cells of the (L+1)-th row of the cell array, distance between the collating symbol string and the symbol string to be collated is obtained as the result of the collation. Thus, the collation is executed for the collating symbol string of the optional length smaller than the number of the rows of the cell array.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to the components of an information processing system, and more specifically, the present invention relates to components of an information processing system, and more specifically, the present invention relates to components of an information processing system, and more specifically, the present invention relates to the components of an information processing system. The present invention relates to a control method for a symbol string matching device that searches for a matching symbol string.

[Conventional technology]

Symbol string matching devices are used for searching text databases, matching feature sequences in button recognition systems, keyword searches from documents created with word processors, supporting machine translation, email address filtering, etc. It is indispensable in information processing systems.

It is desired that this symbol string matching device be able to search not only for a matched symbol string that completely matches a matching symbol string, but also for a matching symbol string that has a certain similarity to the matching symbol string. This is because, for example, in text database searches, this function is necessary when the text contains misspellings or when performing ambiguous keyword searches, and when matching feature sequences in pattern recognition, it is necessary to use this function completely. This is because it is rare to find a matching symbol string, and it is necessary to select the matching symbol string most similar to the matching symbol string from among multiple matching symbol strings.

The concept of distance is used to measure the similarity between the above-mentioned matching symbol string and the to-be-matched symbol string. The distance mentioned here is 1
Considering the removal, substitution, and insertion of symbols as unit operations, we consider how many operations it takes to move from one symbol string to another, and calculate the distance between these two symbol strings by calculating the minimum number of operations. Regarding this concept of distance, for example, 198
Computing 5urveys, Volume 12, Issue 4, Page 381, published in 2013, titled Approximate String Matching.
e StringMatching), author Patri
ck, Hall, Geoff, dowling.

is described in detail.

A symbol string matching device capable of searching for a symbol string to be matched within an arbitrary distance from a matching symbol string is disclosed in Japanese Patent Application No. 63-269746.
There is a symbol string matching device and its control method (hereinafter referred to as the prior invention).

The symbol string matching device of the prior invention has a length of N (N is a positive integer).
a cell array in which cells for storing symbol string comparison results are arranged in M rows (M is a positive integer) and N+1 columns for the matching symbol string;
Only when the same symbol to be matched as the jth symbol (j is any positive integer equal to or less than N) in the collation symbol string is given, the i-th row (i is any positive integer equal to or less than M) column j of the cell array. The first transfer means transfers the stored data of cell fi1 to cell fij+1, and only when a collated symbol different from the j-th symbol of the collation symbol string is given, the h row (
(h is any positive integer less than M) A second transfer means that transfers the stored data of cell f□, which is the jth column, to cell fh++++++; Regardless of what happens, the hth row and column (
k is any positive integer equal to or less than N+1), the data stored in the cell f hk is transferred to the cell f bulk, and at least one 1 is transferred to the cell fik by the first to third transfer means. When the data is transferred, the stored data of the cell fik is set to 1, and if no 1 is transferred to the cell fik by the first to third transfer means, the data stored in the cell fik is set to 1.
an input means for setting the stored data of cell fhl to 0; and when the stored data of cell fhl becomes 1, the input means sets cell fh++ to 0;
Regardless of the input given to +++, the cell f h+l j
In this configuration, by setting the memory data of cell fl+ to 1 after setting the memory data of all cells of the cell array to 0, Performing initial settings; Executing symbol string matching by sequentially giving symbol strings to be matched one symbol at a time; and
A control method characterized in that the distance between the verification symbol string and the to-be-matched symbol string is obtained as a verification result by searching for a cell with stored data of 1 and the smallest row number among the cells. is used.

FIG. 4 is a diagram for explaining the operating principle of the prior invention. The figure shows, as an example, a case where a cell array with 4 rows and 5 columns is used, and "'ABCD'" is used as the collation symbol string. As will be explained in detail below, with this configuration, it is possible to search for any symbol string to be matched within a distance of 3 from "A B CD".

In the figure, cells are indicated by circles, the first transfer means is a solid arrow in the row direction, the second transfer means is a solid arrow in the column direction, and the third transfer means is a solid arrow in the diagonal direction. , the setting means are indicated by diagonal double-lined arrows. The first and second transfer means transfer the stored data of each cell to the next cell according to the arrow only when a symbol to be verified written next to the arrow is given. When a symbol to be verified is given, the third transfer means transfers the stored data of each cell to the next cell in accordance with the arrow, regardless of the given symbol to be verified. Although not clearly shown in the figure, the cell has input means for receiving and receiving data transferred from these transfer means, and when at least one 1 is transferred, the memory gater of the cell becomes 1. , if no 1 is transferred, the stored data in the cell becomes 0.

The setting means has the function of setting the next cell stored data to 1 according to the double line arrow, regardless of the data inputted from the input means, when the stored data of a certain cell becomes 1. In other words, for example, if cell f+2 is 1, at the same time f
23+ f34+ f4s also becomes 1, and cell f2□ becomes 1
At the same time, f33. fN also becomes 1. In such a configuration, the verification result can be obtained by reading the stored data of the cell in the fifth column.

The above will be explained in more detail based on FIG. 5 and FIGS. 6(a) to (h).

5 and 6(a) to 6(h) each show the same configuration as FIG. 4, but some symbols are omitted for simplicity. For symbols used below that are not explicitly shown in these figures, the symbols for the corresponding parts in FIG. 4 are used. In addition, hatched cells in the figure indicate that the stored data is 1, and white cells indicate that the stored data is 0.

FIG. 5 is a diagram showing the initial state of the symbol string matching device according to the prior invention before starting matching. In the initial state, cell f
Set the memory data of i1 to 1. Then, the storage data of f2□r f33+ f<< is also set to 1 by the setting means. All data stored in cells other than these are set to 0.

Figures 6(a)-(d) are collation symbol strings "A B CD
”, the symbol string to be matched “'ABCD” is sequentially A and B.

The operation of the symbol string matching device according to the present invention when inputting C and D is shown for each symbol input. After each collated symbol is input, the cell array contains the collated symbol strings and collated symbol strings that have been input so far.'' A B CD
” is shown.

To know this verification result, it is sufficient to read the data stored in the cell in the fifth column. Specifically, among the cells whose storage data is 1 in the fifth column, search for the one with the smallest issue number, and if it is f15, the distance between the symbol string to be matched and the symbol string to be matched is 0. ．． If f25, distance is 1. For f35, distance is 2. If f4s, the distance is 3°; if there is no corresponding cell, the distance is 4
That's all.

Each figure will be described in detail below in the order of the figures.

First, when the symbol A to be verified is input from the initial state, the stored data in each cell changes from the initial state shown in FIG. 10 as shown in FIG. 6(a). As mentioned above, this figure shows the result of matching the symbol string to be matched "A" against the symbol string to be matched "ABCD" as part of the process of matching the symbol string to be matched "A B CD ". The data stored in cell f45 is 1, which indicates that A'' is at a distance of 3 from "'ABCD".

Next, when the symbol B to be verified is input, the stored data in each cell changes from the state shown in FIG. 6(a) to the state shown in FIG. 6(b). This figure shows the result of matching the symbol string to be matched "'AB'" against the matching symbol string 11AB CD.The stored data in cell f35 is 1, which means that "'A B" is "ABCD". This shows that it is at a distance of 2 from .

When the symbol C to be verified is subsequently input, the stored data in each cell changes from the state shown in FIG. 6(b) to the state shown in FIG. 6(c). This figure shows the result of matching the symbol string to be matched "'ABC" against the matching symbol string "'ABCD"'. The data stored in cell f25 is 1, which indicates that "AB'" is at a distance of 1 from "A B CD ".

When the symbol to be collated is finally input, the stored data in each cell changes from the state shown in FIG. 6(c) to the state shown in FIG. 6(d). This figure shows the collation symbol string "A B CD
” shows the final matching result of the symbol string to be matched “’ABCD”. The stored data in cell f+5 is 1, which means that “’AB CD ” is “’ABCD”.
This shows that the distance is 0 from .

In this way, by sequentially inputting the symbol string to be matched "A B CD " one by one, the progress and final result of matching for this symbol string to be matched can be sequentially known. The matching result obtained in this way clearly gives the correct distance between the symbol string to be matched and the symbol string to be verified. This is because the symbol string to be matched “l A I+” is “AB
"CD" was subjected to the unit operation of removing the upper symbol three times; similarly, "A B" was performed twice, ABC" was performed once,
``ABCD I+ has been done 0 times.

FIGS. 6(e) to 6(h) show the case where the symbol string to be matched is "'ACXD" under the same conditions as FIGS. 6(a) to (d). To explain briefly, Fig. 6 (e
), the symbol string to be matched “A” is the symbol string to be matched” A B CD
6(f), (g
) and (h) are the symbol strings to be matched “'AC””“
A CX “” A CX D” is the collation symbol string “ABC”
It is clear that these results give the correct distance between each symbol string.

In this way, with the configuration shown in FIG. 4, it is possible to determine whether a certain symbol string to be verified is within a distance of 3 from the symbol string to be verified "'ABCD". Therefore, by inputting a plurality of symbol strings to be matched one after another, it is possible to search among them all symbol strings to be matched within a distance of 3 from the symbol string to be matched.

[Problem to be solved by the invention]

As described above, the symbol string matching device of the prior invention can search for a symbol string to be matched within an arbitrary distance from a matching symbol string. However, in the prior invention, the length of the symbol string that can be matched is fixed depending on the size of the cell array of the symbol string matching device, so a different symbol string matching device is required for matching symbol strings of different lengths. The drawback was that it had to be prepared.

In addition, since the symbol string matching device of the prior invention can only match one matching symbol string and the to-be-matched symbol string, a matching operation that searches for the symbol string closest to the to-be-matched symbol string from among multiple matching symbol strings cannot be performed. There wasn't. The purpose of the present invention is to solve such problems, and to provide a simple control method that is capable of collating multiple collation symbol strings of length or less and a symbol string to be collated in parallel.
The object of the present invention is to provide a control method for a symbol string matching device.

[Means to solve the problem]

In order to achieve the above object, a control method for a symbol string matching device according to the first aspect of the present invention includes a cell array in which cells for storing symbol string comparison results are arranged in M rows (M is an integer) and N+1 columns, and a cell array for storing symbol string comparison results. Only when the same input symbol as the j-th symbol (j is any positive integer equal to or less than N) is given, the cell fi1 in the i-th row (i is any positive integer equal to or less than M) and the j-th column of the cell array
The first transfer means transfers the stored data of ``fhj+1'' to the cell fhj+1, and only when an input symbol different from the j-th symbol of the collation symbol string is given, the memory data of hth row (h is any positive integer less than M) of the cell array is transmitted. ) second transfer means for transferring the stored data of the cell fhl in the jth column to the cell fh+1j+1; (k is any positive integer equal to or smaller than N+1) of cell fhk, and a third transfer means for transferring the stored data of cell fhk to cell fh+1b; When at least one ``1'' is transferred to cell fik, the stored data of cell fik is set to 1, and if no 1 is transferred to cell fik by the first to third transfer means, the stored data of cell fik is set to 0. and a setting means for setting the stored data of the cell fh+++++ to 1 when the stored data of the cell fhj becomes 1, regardless of the input given to the cell fh+lj+1 by the input means. Ori,
In such a configuration, when the length of the verification symbol string is an arbitrary positive integer equal to or less than N, verification is performed at least by setting all the cell values of the cell array to 0, and then using the cell f.
performing initialization by setting the stored data of 1l to 1; performing symbol string matching by sequentially giving symbol strings to be matched one symbol at a time; and Memory data is 1 from
The present invention is characterized in that it includes obtaining the distance between the matching symbol string and the matching symbol string as a result of matching by searching for the cell with the smallest row number.

The second control force formula of the symbol string matching device of the present invention uses a plurality of the symbol string matching devices to simultaneously execute symbol string matching between a plurality of matching symbol strings and a symbol string to be matched, and the result of the matching is The method is characterized in that the matching symbol string having the smallest distance from the matching symbol string is identified as , and the value of that distance is obtained.

[Effect]

In the present invention, by making it possible to read the matching results from all columns of the cell array used in the prior invention,
Verification can be performed on a verification symbol string of any length smaller than the number of columns in the cell array.

For example, using the cell array shown in FIG.
In order to match against B'”, you can read the value of the cell in the third column. In this way,
By changing the column from which the verification result is read depending on the length of the verification symbol string, a symbol string verification device with a cell array of N+1 columns can process any verification symbol string of length N or less.

Furthermore, by arranging multiple symbol string matching devices, registering different matching symbol strings in each, and simultaneously inputting the symbol string to be matched one symbol at a time to each symbol string matching device, multiple matching symbol strings can be simultaneously processed. You can make it possible to match.

〔Example〕

FIG. 1 is a block diagram showing an embodiment of a symbol string matching device according to the present invention. The figure will be explained below.

First, the configuration shown in FIG. 1 will be explained. In the figure, the symbol string matching device has an initial set terminal 1l0, N matching signal input terminals 1201 to 120-N (N is a positive integer), and N
Tree verification symbol string length setting terminals 125-1 to N and M rows (M
is a positive integer) are arranged in N+1 columns, the D terminal inputs data, the Q terminal outputs data, and the S terminal sets data to 1.
Input to the Q terminal and input terminal 120-j of register 130 with ET terminal 7, register f++130 in row i (i is any positive integer less than or equal to M) and column j (j is any positive integer less than or equal to N) If the terminals are connected and i = 1, the register f on the right side,, +
, 130 are connected to the first AND gate 140, and the input is connected to the Q terminal of the register fh+130 (h is any positive integer less than M), and the input terminal 12
Second AND gate 1 whose inverting input terminal is connected to 0-j
50 and a data transfer line 16 whose input is connected to the Q terminal of register fhj130 (k is any positive integer equal to or less than N+1)
0, and as inputs the output of the first AND gate 140 connected to the Q terminal of register f, +1□130, and the second AND gate 1 connected to the Q terminal of register fy130.
50 output and register f hl+1l. A first OR gate 170 receives the output of the data transfer line 160 connected to the Q terminal of 30 and outputs it to the D terminal of register f, +, , +, 130, and the output of the D terminal of register fhj130 is sent to register f, +, , +, 130. A set line 180 input to the SET terminal of the register 130, a third AND gate 185 connected to the Q terminal of the register 130 and the collation symbol string length setting terminal 125, and a second OR gate 186 receiving the output of the third AND gate. , M output terminals 190-1 to 190-M that output the output of the second OR gate 186 to the outside.
It is equipped with

In FIG. 1, the input terminal 120-j receives 1. if the j-th symbol of the collation symbol string matches the symbol to be collated. If there is no match, O is given. Then, the first AND gate 140 transfers the data stored in the register fh+130 to the register fh+++ 130 only when the j-th symbol of the collation symbol string and the same symbol to be collated is given, and otherwise transfers 0. Conversely, when the second AND gate 150 is given a symbol to be matched that is different from the j-th symbol in the string of symbols to be matched, the data stored in the register fh1l30 is transferred to the register fh++++++.
130, and if they are the same, transfer 0.

FIG. 2 shows an embodiment of an input device for providing input signals to the symbol string matching device of FIG. The figure shows a case where each signal is composed of 1 bit.

First, from the verification symbol string registration terminals 210-1 to 210-N,
The verification symbol string is registered in the register 240.

Next, if the length of the verification symbol string is L, the value 1 is given only to the verification symbol string length setting terminal 220-L, “0” is given to the other verification symbol string length setting terminals, and the value is registered in the register 250. do.

The symbol string to be verified is input one symbol at a time from the symbol string input terminal 230 to be verified, and is compared with each symbol of the symbol string to be verified by the comparator 260. As a result of comparison, if both symbols match, 1.
If they do not match, 0 is the match signal output terminal 270-1~
270-N. Match signal output terminal 270-
By inputting the output of j to the input terminal 120-j in FIG. 1, the input described in the explanation of FIG. 1 can be obtained. Further, the value registered in the register 250 is output as is from the verification symbol string length output terminal 280.

The output of the collation symbol string length output terminal 280 in FIG. 2 is applied to the symbol string length setting terminal 120 in FIG.

In other words, if the cloudiness of the verification symbol string is L, 1 is given only to 120L. As a result, only the output of the Q terminal of the register 130 in the 2nd+1st column is sent to the second OR gate 186 by the action of the third AND gate.

What is applied to the output terminal 190 is the output of the register 130 in the L+1 column, that is, the matching result for the matching symbol string of length L.

FIG. 3 shows an embodiment showing a control system when K (K is any positive integer) of this symbol string matching device are arranged. In the figure, each input device 350 and each symbol string matching device 360 are connected as described above, and X number of them are arranged in the row direction. Each input device 350 includes a verification symbol string input terminal 31
Verification symbol strings starting from 0 and having a length of N or less are registered respectively. Further, an input corresponding to the length of each verification symbol string is registered in each input device 350 from the verification symbol string length setting terminal 320.

After initializing each symbol string matching device 360 from the initial setting terminal 340, inputting the symbols to be matched one by one into the symbol string input terminal 330 to perform matching according to these registrations.
This is done by simultaneously applying it to the input terminal 230 of the symbol string to be collated of each input device 350. The output terminal 190 of each symbol string matching device 360 receives each matching symbol string as a matching result,
The distance of the matched symbol string can be obtained at the same time. This output is provided to priority encoder 370.

Since the priority encoder is well-known, a detailed explanation will be omitted, but it encodes the input given to the input terminal 327 and uses the symbol string matching device 360 in which the matching symbol string having the shortest distance from the symbol string to be matched is registered. It has a function to output the address and the distance value at that time. This makes it possible to search for the symbol string closest to the to-be-matched symbol string from a plurality of matching symbol strings, and to obtain the distance at that time.

〔Effect of the invention〕

According to the control method of the symbol string matching device according to the present invention, as explained above, by simply adding simple logic gates to the structure of the symbol string matching device of the prior invention and arranging them,
It becomes possible to perform matching on a wave train against any matching symbol string of a certain length or less.

[BRIEF DESCRIPTION OF THE DRAWINGS] Fig. 1 is a block diagram showing an embodiment of the symbol string matching device of the present invention, Fig. 2 is a block diagram of an input device, and Fig. 3 is a symbol for simultaneously performing multiple symbol string matching. Configuration diagram of the column matching device, Fig. 4,
FIGS. 5 and 6 are principle diagrams for explaining the prior art. 130... Register, 240... Collation symbol register, 250... Symbol string length register, 260... Comparator, 350... Input device, 360... Symbol string matching device, 370... Priority encoder, 380...
- Matching address output device, fij, etc...cell.

Claims (1)

[Claims] 1. A cell array in which cells for storing symbol string comparison results are arranged in M rows (M is a positive integer) and N+1 columns, and a j-th column of matching symbol strings (j is any positive integer equal to or less than N). a first transfer means that transfers the stored data of the cell f_i_j in the i-th row (i is any positive integer equal to or less than M) and the j-th column of the cell array to the cell f_i_j_+_1 only when an input signal having the same symbol as is given; Then, only when an input signal different from the j-th symbol of the collation symbol string is given, the cell f_h in the h-th row (h is any positive integer less than M) of the j-th column of the cell array
a second transfer means for transferring the stored data of _j to cell f_h_+_1_j_+_1; a third transfer means for transferring the stored data of the cell f_h_k, which is a positive integer), to the cell f_h_+_1_k; and the stored data of the cell f_i_k, at least one 1 being transferred to the cell f_i_k by the first to third transfer means. is set to 1, and if no 1 is transferred to the cell f_i_k by the first to third transfer means, the cell f
an input means for setting the stored data of the cell f_h_j to 0; and a setting means for setting the stored data of the cell f_h_+_1_j_+_1 to 1 when the stored data of the cell f_h_j becomes 1, regardless of the input given to the cell f_h_+_1_j_+_1 by the input means. In the symbol string matching device equipped with the above-mentioned symbol string matching device, matching when the length of the matching symbol string is an arbitrary positive integer L less than or equal to N is performed at least by setting the values of all cells in the cell array to 0, and then performing the matching in the cell f_1_l. Memory data 1
means for performing initial setting by setting a symbol string to be compared; means for performing symbol string matching by sequentially giving a symbol string to be matched one symbol at a time; is 1 and has the smallest row number, and includes means for obtaining the distance between the collated symbol string and the to-be-verified symbol string as a result of the collation. . 2. Using a plurality of the above-mentioned symbol string matching devices, simultaneously execute symbol string matching between a plurality of symbol strings to be matched and a symbol string to be matched, and as a result of the matching, select a matching symbol string that has the smallest distance from the symbol string to be matched. 2. The control method for a symbol string matching device according to claim 1, wherein the method specifies the symbol string matching device and obtains a value of the distance.