JP3452353B2 - Recording medium recording dictionary data structure, dictionary lookup method, phrase acquisition method, dictionary lookup apparatus, phrase acquisition apparatus, recording medium recording program - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a recording medium recording a dictionary data structure, a dictionary lookup method, and a word acquisition method suitable for realizing an encrypted dictionary and a high-speed dictionary lookup using the encrypted dictionary. The present invention relates to a dictionary lookup device, a phrase acquisition device, and a recording medium recording a program.

[0002]

2. Description of the Related Art Conventionally, in Japanese Kana-Kanji conversion used in a word processor or the like, a dictionary is recorded as data of a trie structure in order to quickly determine whether or not a character sequence is registered in the dictionary. A method of recording on a medium has been proposed. FIG. 1 is a schematic diagram showing an outline of a trie structure that realizes a part of a kana-kanji conversion dictionary.

As shown in the figure, in this trie structure, the spelling "wa" and "story" are read for the reading "wa", and the spelling "cotton""wata" and the like for the reading "wata". , ・ The spelling "I" etc. is recorded for the reading "Watakushi", and the spelling "I""passing""I" etc. is recorded for the reading "I".

As shown in the figure, the trie structure is generally represented by a tree structure, and the node 104 is moved in order from the start node 101 along the arrow 103 to which the character 102 in the kana character string to be examined is assigned. Further, the end character (#) 105 in the figure indicates that there is the spelling data 106 registered by reading up to that point.

On the other hand, even in a word processor, morphological analysis is often performed because of the necessity of proofreading, summarizing, searching, etc. of sentences. The purpose of morphological analysis is to divide spelling and obtain word information such as readings. In morphological analysis, a trie structure dictionary is also used.

FIG. 2 is a schematic diagram showing an outline of a trie structure for realizing a part of the morphological analysis dictionary. Note that, in this figure, elements having the same functions as those in the above-mentioned figures are denoted by the same reference numerals.

As shown in the figure, in this trie structure, the spelling "I" is read for "I""Wakushi"
Morphological information such as "shi" is read for spelling "private" Morphological information such as "Let's" is read for spelling "wata" Morphological information such as "wata" is spelled "watakushi" Morphological information such as reading "Watakushi" is recorded for the spelling "I", and morphological information such as reading "I" is recorded for the spelling "I".

Start node 101, character 102, arrow 10
The trie structure is realized by using the node No. 3, the node 104, and the end character (#) 105 as in the case of the above-described figure, but the reading data 1 is used instead of the spelling data 106.
07 is recorded.

Further, these dictionaries can be fused into one trie structure. FIG. 3 is a schematic diagram showing an outline of a trie structure in which a part of the above-mentioned dictionary is fused. In addition,
In this figure, the elements having the same functions as those in the above-mentioned figures are designated by the same reference numerals. Note that the end character distinguishes whether the corresponding information is spelling information or reading information,
In the figure, the former is shown as # 1 and the latter as # 2.

A method for realizing such a trie-structured dictionary by using an array is disclosed in the following document, for example.・ Junichi Aoe "High-speed digital search algorithm using double array" (IEICE Transactions, D, Vol. 171-D,
No.9, pp.1592-1600, September 1988) ・ Junichi Aoe “High-speed digital search algorithm using search double array of natural language dictionary” (bit, Vol.21, No.6,
pp.776-784, May 1989)

[0011]

In the conventional method, such information is stored as it is in the spelling data 106 and the reading data 107. A problem in creating and distributing such a dictionary is plagiarism of data recorded in the dictionary, particularly plagiarism of registered word spelling and reading data. Therefore, there is a demand for a method for encrypting word data while maintaining the advantage of dictionary lookup by the trie structure.

An object of the present invention is to realize a method of preventing the reading and spelling of word data from being stolen by using an encrypted trie-structured dictionary and restoring the dictionary, reading and spelling at high speed using the dictionary. To do.

[0013]

To achieve the above object, a computer-readable recording medium in which a dictionary data structure according to a first aspect of the present invention is recorded is a phrase of an arbitrary length, and a phrase of the phrase. A dictionary data structure that holds information data and data in association with each other is recorded and configured as follows.

That is, the dictionary data structure stores the start node n ₁ .

Here, the dictionary data structure holds a phrase expressed by a character string c ₁ , c ₂ , ..., C _s having a length s and information data d of the phrase in association with each other. If you do
(A) A pair of the start node n ₁ and the character c ₁ (n ₁ , c ₁ )
And node n ₂ is stored in a one-to-one correspondence, and (b) an integer
For i (2 ≦ i ≦ s) , pairs of the character c _i and node n _i (n
_i , c _i ) and the nodes n _{i + 1} are stored in a one-to-one correspondence,
(C) The information data d is stored in one-to-one correspondence with the node n _{s + 1} .

In the present invention, the pair for integer i (1≤i≤s)
Since (n _i , c _i ) and the next node n _{i + 1} are stored in a one-to-one correspondence, they are stored in the information data d as character strings c ₁ , c ₂ , …, It is not necessary to store c _s as it is. Therefore, the spelling and reading data can be encrypted.

On the other hand, since there is a one-to-one correspondence, the high speed search method of the trie structure can be applied as it is.

A dictionary lookup method according to a second aspect of the present invention is a word / phrase expressed by a character string e ₁ , e ₂ , ... E _s of length s from a recording medium storing the above-mentioned dictionary data structure. The information data held in association with the above is subtracted, and a head acquisition step, a sequential acquisition step, and a data output step are provided.

Here, in the head acquisition step, the node n ₂ accumulated in a one-to-one correspondence with the pair (n ₁ , e ₁ ) of the start node n ₁ and the character e ₁ is acquired.

On the other hand, in the sequential acquisition process, the integer i (2 ≦ i ≦
node n obtained earlier for (s) _iAnd the letter e_iWith
Pair (n_i, e_i), A node n stored in a one-to-one correspondence
_{i + 1}Are sequentially acquired.

Further, in the data output step, the information data d accumulated in a one-to-one correspondence with the obtained node n _{s + 1} is obtained and outputted.

According to the present invention, information data of a phrase expressed by a character string can be obtained at high speed from the recording medium.

Further, the dictionary lookup method of the present invention can be configured to further include a reporting step.

Here, in the reporting step, if the nodes accumulated in a one-to-one correspondence in the head acquisition step or the sequential acquisition step cannot be acquired, the information data held in correspondence with the word / phrase is held. Report that there is no.

According to the present invention, when the word / phrase expressed by the character string is not registered in the dictionary, a report to that effect can be obtained at high speed.

A word / phrase acquisition method according to a third aspect of the present invention obtains a word / phrase held in association with information data d held in a recording medium storing the above-mentioned dictionary data structure. Therefore, a node acquisition process, a head acquisition process, a sequential acquisition process, and a character string output process are provided.

Here, in the node acquisition step, information data
The node m ₀ stored in one-to-one correspondence with d or by being referred to by the information data d is acquired.

On the other hand, in the head acquisition step, the pair (m ₁ , e ₁ ) stored in one-to-one correspondence with the acquired node m ₀ is acquired.

Furthermore, in the sequential acquisition step, the integer j (1 ≦
For j), the node m obtained earlier _jOne-to-one correspondence with
The accumulated pairs (m_{j + 1}, e_{j + 1}) Are sequentially acquired.

Then, in the character string output step, if the start node n ₁ matches any one of the leading elements m _k of the sequentially obtained pairs, the sequence e _k of the last elements of the sequentially obtained pairs,
e _k−1 , ... E ₂ , e ₁ are output as a string of characters expressing a word held in association with the information data d.

According to the present invention, it is possible to restore the reading or spelling character string of the word from the word data d of the encrypted dictionary.

According to a fourth aspect of the present invention, there is provided a dictionary lookup device, wherein a phrase expressed by a character string e ₁ , e ₂ , ... E _{s having} a length s from a recording medium storing the above dictionary data structure. The information data held in association with
A head acquisition unit, a sequential acquisition unit, and a data output unit are provided.

In this case, the head acquisition unit is the start node n ₁
And a node n ₂ accumulated in a one-to-one correspondence with the pair (n ₁ , e ₁ ) of the character e ₁ and the character e ₁ .

On the other hand, the sequential acquisition unit, for an integer i (2 ≦ i ≦ s), pairs (n) of the node n _i acquired previously and the character e _i.
i , e _i ) and the nodes n _{i + 1} stored in a one-to-one correspondence are sequentially acquired.

Furthermore, the data output unit is configured to store the information data stored in one-to-one correspondence with the acquired node n _{s + 1.}
Get d and output it.

The dictionary lookup apparatus of the present invention can be configured to further include a reporting unit.

Here, when the report part cannot be acquired in the head acquisition part or the sequential acquisition part because there is no node accumulated in a one-to-one correspondence, the report part is held in association with the relevant phrase. Report that there is no information data available.

A word / phrase acquisition device according to a fifth aspect of the present invention obtains a word / phrase held in association with information data d held in a recording medium storing the above-mentioned dictionary data structure. Therefore, the node acquisition unit, the head acquisition unit, the sequential acquisition unit, and the character string output unit are provided.

Here, the node acquisition unit acquires the node m ₀ accumulated in one-to-one correspondence with the information data d, or by being referred to by the information data d.

On the other hand, the head acquisition unit acquires the pair (m ₁ , e ₁ ) stored in one-to-one correspondence with the acquired node m ₀ .

Furthermore, the sequential acquisition unit, for the integer j (1≤j), stores the pair (m _{j + 1} , e _{j + 1} ) stored in one-to-one correspondence with the node m _j acquired previously. Are sequentially acquired.

Then, the character string output unit is the start node
If n ₁ matches the first element m _k of any of the sequentially obtained pairs, the last element sequence e _k , of the sequentially obtained pair
e _k−1 , ... E ₂ , e ₁ are output as a string of characters expressing a word held in association with the information data d.

A computer-readable recording medium having a dictionary data structure recorded according to a sixth aspect of the present invention has the following structure of the recording medium having the dictionary data structure recorded therein.

That is, the dictionary data structure is the array BA.
SE and array CHECK are used to hold words, nodes, and their information data.

On the other hand, the start node n ₁ , a string of characters c ₁ , c ₂ , ..., C _s representing the held word and the nodes n _2, ..., N _s , n associated with it The information of _{s + 1} and is represented by integers.

Furthermore, for an integer i (1 ≦ i ≦ s), the n _i- th element BASE [n _i ] of the array BASE and the n-th element of the array CHECK.
_{For the i +} 1th element CHECK [n _{i + 1} ] and two conditions n _{i + 1} = BASE [n _i ] + c _i CHECK [n _{i + 1} ] = n _i are both satisfied. To hold.

According to the present invention, it is possible to realize a dictionary having a trie structure encrypted by using an array.

According to a seventh aspect of the present invention, there is provided a dictionary look-up method in which the length of a dictionary is stored in a recording medium storing the dictionary data structure.
The information data held by being associated with the phrase expressed by the character string of s e ₁ , e ₂ , ... e _s ,
It is configured to include a sequential acquisition step and a data output step.

Here, in the sequential acquisition step, the integer i (1 ≦ i
≤ s), the sum of the n _i- th element BASE [n _i ] of the array BASE and the character e _i is t _i = BASE [n _i ] + e _i , and the condition CHECK [t _i ] =
When n _i is satisfied, it is sequentially acquired as an integer n _{i + 1} = t _i .

On the other hand, in the data output step, information data accumulated in a one-to-one correspondence with the obtained integer n _{s + 1}
Get d and output it.

The dictionary lookup method of the present invention can be configured so as to further include a reporting step.

Here, in the reporting step, if the condition is not satisfied in the sequential acquisition step, it is reported that there is no information data held in association with the word.

A word / phrase acquisition method according to an eighth aspect of the present invention is to obtain a word / phrase held in association with the information data d held in a recording medium storing the dictionary data structure. Then, a node acquisition process, a sequential acquisition process, and a character string output process are provided.

Here, in the node acquisition step, information data
An integer that represents the information of the node that is stored in a one-to-one correspondence with d or with a one-to-one reference to the information data d
Get m ₀ .

On the other hand, in the sequential acquisition process, an integer j (0 ≦ j)
For the integer m _j obtained earlier, the integer m _{j + 1} = CHE
CK [m _j ] and the character integer e _{j + 1} = m _j -BASE [m _j ] are sequentially acquired.

Further, in the character string output step, when the integer n ₁ expressing the information of the start node is equal to any of the sequentially obtained integers m _k , the sequence of the sequentially obtained character integers The e _k , e _k-1 , ... E ₂ , e ₁ are output as a string of characters expressing the word held in association with the information data d.

According to a ninth aspect of the present invention, there is provided a dictionary look-up device, in which a length of a dictionary medium for storing the dictionary data structure is
The information data held by being associated with the phrase expressed by the character string of s e ₁ , e ₂ , ... e _s ,
It is configured to include a sequential acquisition unit and a data output unit.

Here, in the sequential acquisition unit, the integer i (1 ≦ i ≦
s), the sum of the n _i- th element BASE [n _i ] of the array BASE and the character e _i is t _i = BASE [n _i ] + e _i , and the condition CHECK [t _i ] = n _i
When it satisfies, it is sequentially acquired as an integer n _{i + 1} = t _i .

On the other hand, in the data output section, the information data d stored in a one-to-one correspondence with the obtained integer n _{s + 1.}
To get and output.

Further, the dictionary lookup apparatus of the present invention can be configured to further include a reporting unit.

Here, if the condition is not satisfied in the sequential acquisition unit, the report unit reports that there is no information data held in association with the word / phrase.

A word / phrase acquisition device according to a tenth aspect of the present invention obtains a word / phrase held in association with information data d held in a recording medium storing the dictionary data structure. And a node acquisition unit, a sequential acquisition unit, and a character string output unit.

Here, the node acquisition unit makes one-to-one correspondence with the information data d, or one-to-one with the information data d.
An integer m ₀ that represents the information of the node that is accumulated by being referred to
To get.

[0064] On the other hand, sequential acquisition unit, for integer j (0 ≦ j), the integer m _j obtained above, the integer m _{j + 1} = CHECK [m
_j ] and the character integer e _{j + 1} = m _j -BASE [m _j ] are sequentially acquired.

Further, when the integer n ₁ expressing the information of the start node is equal to any of the sequentially obtained integers m _k , the character string output unit outputs the sequence of the sequentially obtained character integers. The e _k , e _k-1 , ... E ₂ , e ₁ are output as a string of characters expressing the word held in association with the information data d.

A computer-readable recording medium recording a program according to an eleventh aspect of the present invention is configured to record a program that causes a computer to function as the dictionary lookup device or the phrase acquisition device.

[0067]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the accompanying drawings.

(Schematic Configuration of Information Processing Apparatus) FIG. 4 is a typical information processing apparatus (general-purpose computer, various terminals, mobile terminal, mobile phone, etc.) that functions as a dictionary lookup apparatus or word acquisition apparatus of the present invention. FIG. 3 is a schematic diagram showing a schematic configuration of a dedicated device such as a game device). Hereinafter, description will be given with reference to FIG.

The information processing device 401 has a CPU (Central
Processing Unit (Central Processing Unit) 402. When the information processing device 401 is powered on,
The CPU 402 is an IPL (Initial Program) stored in a ROM (Read Only Memory) 403.
Loader; initial program loader) is executed.

The IPL is a hard disk 404, FD
(Floppy Disk; FD, CD-ROM (Compac
t Disk ROM) CD-ROM installed in the drive 411
OS (Operating System) stored in a recording medium such as
Operating system) A program that reads and executes a program.

After booting the OS, the CPU 402 is stored in the hard disk or the like according to the user's instruction input through the keyboard 405 or the mouse 406, or according to the contents of the setting file previously described in the hard disk or the like. Run the application program.

In a small-scale information processing device such as a mobile terminal, an embodiment in which the IPL itself functions as an OS or an application program can be adopted.

The CPU 402 uses a RAM (Random Access Memory) 407 as a temporary work storage area when executing a program. In addition, a register or a cache (not shown) provided in the CPU 402 is used as a temporary work storage area.

In order to report the result to the user or show the progress along with the execution of the program, the CPU 402
Can display the information on a display device 408 such as a liquid crystal display or a CRT (Cathode Ray Tube). In the instruction operation using the mouse 406, the cursor displayed on the screen is moved by moving the mouse 406, and the menu item pointed to by the cursor can be selected by clicking the mouse 406.

The information processing device 401 is a NIC (Network
Interface Card) or modem interface 40
9 can communicate with a computer communication network such as the Internet. The document data received via the interface 409 can be processed, the processed result can be transmitted via the interface 409,
The program received via the interface 409 can be executed and so on.

(Outline of Trie Structure) FIG. 5 is a schematic diagram showing a state in which a dictionary having the same contents as the fused trie structure dictionary shown in FIG. 3 is constructed by the method of the present invention. Hereinafter, description will be given with reference to this figure.

To check whether or not a character string is registered in this dictionary 501, first, the start node 10
The characters in the character string are sequentially examined from 1, and the node 104 may be moved along the arrow 103 to which the character 102 is assigned. When the arrow 103 to which the end character 105 is assigned extends from the relevant 104, it means that the word or phrase corresponding to the character string up to that point is registered. The node 104 pointed to by the extended arrow 103 is the last node and is given a number that does not overlap with each other.
Although not shown, the nodes 104 other than the last node 104 are also given numbers (node numbers) that do not overlap with each other.

An arrow pointing to the spelling data 106 and the reading data 107 further extends from the last node 104. A number indicating the spelling or reading is assigned to the spelling data 106 or the reading data 107.

The following information is recorded in the dictionary 501 having a trie structure constructed by the method of the present invention.・ The spelling "I" (number 21) for the reading "Watakushi" (final node number 9). The reading "Watakushi" (number 9) for the spelling "Watakushi" (final node number 10) Spelling "I" (number 21), "I" (number 13) for reading "I" (final node number 12). The reading "I" (number 12) for the spelling "I" (final node number 13). The reading "Watakushi" (number 9) "I" (number 12) for the spelling "I" (final node number 21).

The spelling data 106 and reading data 107
Stores the spelling and reading of a word or phrase consisting of a sequence of characters 102 scanned from the start node 101 to the spelling data 106 and reading data 107.
The node number of the trie structure dictionary 501 is stored as the reading information.

For example, in order from the start node 101,
By tracing "wa", "ta", "ku", "shi", "# 1", the spelling data 106 of the spelling "I" for the reading "watakushi" is obtained through the last node 104 (number 9).

From the start node 101, "I" and "#"
2 ”, the reading data 107 of the reading“ I ”and“ Watakushi ”for the spelling“ I ”is obtained through the last node 104 (number 21).

Here, the reading data for the spelling "Watakushi"
The number 9 is stored as the data 107 . this is,
“Wa” “ta” “ku” “shi” in order from the start node 101
This is the number of the last node 104 when “# 1” is traced.

The number 21 is stored as the reading data 106 for the reading "Watakushi". this is,
This is the number of the last node 104 when “I” and “# 2” are sequentially traced from the start node 101.

From the last node 104 to the start node 101
If the arrow is followed in the opposite direction and the characters encountered in the middle are rearranged in reverse order, the reading and spelling can be restored from the node numbers stored in the spelling data 106 and reading data 107.

For this reason, it becomes more difficult to plagiarize the dictionary 501 as compared with the case where the reading and the spelling are stored as they are in the spelling data 106 and the reading data 107. Also, since the reading and spelling are embedded in the tree of the trie structure, the dictionary 501
The overall size is expected to be smaller.

(Realization by Double Array) The following describes how the tri-structure dictionary 501 configured as described above is stored in the double array.

Whether or not there is an arrow 103 that moves from the current node 104 (node number r) by the character 102 of the character code a is determined as follows using the two arrays BASE and CHECK. That is, first, t ← BASE [r] + a is calculated. Next, check whether CHECK [t] = r holds. If it holds, next node 1
The node number of 04 is t. If not, the next node does not exist.

FIG. 6 shows that the double array is actually "I".
It is explanatory drawing which shows the numerical example in case "Watakushi" is registered. Note that the node numbers in this figure differ from the above example. Further, although the figure shows a dictionary for Japanese Kana-Kanji conversion for obtaining spelling information and the like from reading, spelling data 106 is not shown.

In this example, character codes are assigned as follows. End character → 1 “A” “I”… → 2 3… “Ki” “Ku” “Ke”… → 7 8 9… “Shi”… → 12… “Ta”… → 16… “Wa” → 44

In the diagram shown in FIG. 6, each element of the array is expressed by a box that is vertically overlapped (the upper box is BASE, the lower box is CHECK), and the subscript of the element is added on the box. Is described. The node number of the start node is 1. In this case, the procedure for scanning "I" will be briefly described below.

First, for the start node 1, BASE [1] = 1 as shown in the figure, and BASE [1] + wa (44) = 1 + 44 =
45. Since CHECK [45] = 1, it can be seen that the words beginning with "wa" are registered.

Next, as shown in FIG.
BASE [45] = 2 and BASE [45] + C (16) = 2 + 16 = 18
Is. Since CHECK [18] = 45, it can be seen that the words and phrases beginning with "wata" are registered.

Next, as shown in FIG.
BASE [18] = 1 and BASE [18] + then (12) = 13. C
Since HECK [13] = 18, it can be seen that the words beginning with "I" are registered.

Further, for node 13, BASE [13] = 1
And BASE [13] + # (1) = 2. Since CHECK [2] = 13, it can be seen that the word "I" is registered.

The BAS for the last node (number e)
Since E [e] does not have to be used to express the relationship between the nodes of the trie structure and the arrows, the identification numbers of the spelling data 106 and reading data 107, the addresses where these data are stored, etc. are stored here. You can keep it.

On the other hand, a procedure for tracing this in reverse will be described. In the dictionary, when it is possible to transfer from the node r to the node t by the character a, in order to obtain the character a and the node r from the node t, the following calculation may be performed from the above relationship. That is, r ← CHECK [t] a ← t-BASE [r]

In the following, in the example shown in this figure, the character string is restored from the node number 2 of the last node.

First, CHECK [2] = 13, 2-BASE [13] = 2-1 = 1 As a result, the character “# (1)” and the previous node 13 are obtained.

Next, from CHECK [13] = 18, 13-BASE [18] = 13-1 = 12, the character "shi (12)" and the previous node 18 are obtained.

Further, from CHECK [18] = 45, 18-BASE [45] = 18-2 = 16, the character "ta (16)" and the previous node 45 are obtained.

Further, from CHECK [45] = 1, 45-1 = 44, the character "wa (44)" and the previous node 1 are obtained. Since this is the start node, the above iteration ends here. The characters “#” “shi” “ta” obtained in order
If you arrange "wa" in reverse, you will get "me".

As described above, by using the double array, it is possible to obtain a character string for tracing the trie structure from the node number of the last node to the node. This is because the pair (r, a) and the number t are stored in a one-to-one correspondence with each other when a certain node r and a certain character a move to another node t.

The method of sequentially registering words and phrases in the double array and constructing the dictionary is disclosed in the above-referenced paper, and the description thereof is omitted here.

(Dictionary Lookup Procedure) FIG. 7 shows the following when the character strings c ₁ , c ₂ , ..., C _s (c _s is an end character) are given in the embodiment using the double array. It is a flowchart which shows the procedure of the dictionary look-up process which investigates whether the term is registered and acquires the information (reading data or spelling data) of the term. Hereinafter, description will be given with reference to this figure.

Note that these processes are performed by the information processing device 40.
CPU 402 of No. 1 is RAM 407 or hard disk 40
4. The CD-ROM drive 411 and the like are controlled and executed. In addition, it is assumed that areas for variables t, r, and i are secured in advance in the register, the RAM 407, and the like.

First, the integer 1 is assigned to the variable i (step S701). This is a counter for checking how many characters have been scanned.

Then, the node number of the start node is assigned to the variable r (step S702).

Next, calculate the BASE [r] + c _i, assign the result to the variable t (step S703).

Further, it is checked whether CHECK [t] = r (step S704), and if not (step S7).
04; No), and reports that the character string is not registered (step S705), and ends this processing.

On the other hand, if CHECK [t] = r (step S704; Yes), it is checked whether i = s (step S706), and if not (step S706;
No), the value of i is increased by 1 (step S707), and the variable t
Is assigned to the variable r (step S708), and the process returns to step S703.

On the other hand, if i = s (step S70)
6; Yes), that is, if c _i is the ending character, BA
The identification number or address of the information recorded in SE [t] is acquired, the information is output (step S709), and this processing ends.

(Procedure for obtaining a phrase) FIG. 8 is a phrase for obtaining a character string representing a phrase reaching the information when the node number of the last node is stored in information such as spelling data and reading data. It is a flowchart which shows the procedure of an acquisition process. Hereinafter, description will be given with reference to this figure. It should be noted that this procedure is executed in the information processing apparatus 401 that is configured and set similarly to the procedure of the dictionary lookup processing. Furthermore, an array e for storing a character string expressing a phrase is prepared in advance.

First, the value of the variable i is set to 1 (step S801).

The node number of the last node is assigned to the variable t (step S802).

Next, the value of CHECK [t] is obtained, and this is set to the variable r.
(Step S803).

Further, it is checked whether r is equal to the node number of the start node (step S804). If they are not equal (step S804; No), t-BASE [r] is calculated and is substituted for the variable a (step S805).

Then, the value of the variable a is substituted into the i-th element of the array e (step S806).

Further, the value of the variable i is incremented by 1 (step S807), and the value of the variable r is substituted into the variable t (step S807).
808) and the process returns to step S803.

On the other hand, if r is equal to the node number of the start node (step S804; Yes), the element e [i-
1], e [i-2], ..., e [1] are output in this order (step S
809), and this processing ends.

[0121]

As described above, according to the present invention,
Suitable for realizing an encrypted dictionary and a high-speed dictionary lookup using the same, a recording medium recording a dictionary data structure, a dictionary lookup method, a phrase acquisition method, a dictionary lookup device, a phrase acquisition device, and A recording medium recording the program can be provided.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing an outline of a trie structure that realizes a part of a kana-kanji conversion dictionary.

FIG. 2 is a schematic diagram showing an outline of a trie structure that realizes a part of a morphological analysis dictionary.

FIG. 3 is a schematic diagram showing an outline of a trie structure in which a part of the dictionary is fused.

FIG. 4 is a schematic diagram showing a schematic configuration of a typical information processing device that functions as a dictionary lookup device or a phrase acquisition device of the present invention.

FIG. 5 is a schematic diagram showing a schematic configuration of a dictionary having a trie structure according to the present invention.

FIG. 6 is an explanatory diagram showing a state in which the trie-structured dictionary of the present invention is realized by a double array.

FIG. 7 is a flowchart showing a dictionary lookup procedure of the present invention.

FIG. 8 is a flow chart showing the procedure of word acquisition according to the present invention.

[Explanation of symbols]

101 Start node 102 characters 103 arrow 104 nodes 105 end character 106 Spelling data 107 reading data 401 Information processing device 402 CPU 403 ROM 404 hard disk 405 keyboard 406 mouse 407 RAM 408 display device 409 interface 410 FD drive 411 CD-ROM drive

─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-11-7451 (JP, A) Jun'ichi Aoe, search of natural language dictionary High-speed digital search using double array algorithm, bit, Japan, Kyoritsu Publishing, 1989 May 1st, Issue 270, pp. 776-784 Makoto Nagao, Rifumi Sato, Sadao Kurohashi, Tatsuhiko Tsunoda, Iwanami Course Software Science 15 Natural Language Processing, Japan, Iwanami Shoten, October 5, 1999 , No. 4, pp. 250-253 (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06F 17/30-17/30 419 G06F 17/22 520 JISST file (JOIS)

Claims (15)

(57) [Claims] 1. Correspondence between reading of a phrase and spelling of the phrase by holding a character string and information data in association with each other.
A dictionary data structure to be attached and held , wherein the dictionary data structure accumulates a start node n ₁ , and the dictionary data structure is a character string c ₁ , c ₂ , ..., C of length s.
and _s, the information data d, the (a) the start node n ₁ and pair with the character _{c 1 (n 1, c 1} )
And node n ₂ is stored in a one-to-one correspondence, and (b) for integer i (2 ≦ i ≦ s), node n _i and the character
c _i and the pair (n _i, c _i) and, to a one-to-one correspondence to the storage node n _{i + 1,} the information data d by a one-to-one correspondence to the (c) the node n _{s + 1}
By accumulating, it is associated and retained (hereinafter, node
Let n _{s + 1} be the " final node for the character string c ₁ , c ₂ , ..., c _s "
Call. ), The dictionary data structure uses the first and second terminating characters that are distinguishable from other characters to read the phrase and
Spelling and (p) " string of characters obtained by adding the first terminating character to the string of the reading" and "information data of the spelling" are stored in association with each other, and (q) "the spelling The “character string in which the second terminal character is added to the character string of” and the “information data of the reading” are stored in association with each other, and (r) “information data of the spelling” is written with ““ Sentence
"Character string in which the second terminal character is added to the character string"
Information that refers to the “final node” is accumulated, and (s) “information data of the relevant reading” indicates ““ text of the relevant reading.
"Character string in which the first terminating character is added to the character string"
By accumulating information that refers to the "final node,"
Computer readable recording medium storing a dictionary data structure, characterized in that retaining attach response. 2. A storage medium storing the dictionary data structure according to claim 1, which is held in association with a word or phrase expressed by a character string e ₁ , e ₂ , ... E _{s having} a length s. A dictionary lookup method for pulling information data, which acquires a node n ₂ accumulated in a one-to-one correspondence with the pair (n ₁ , e ₁ ) of the start node n ₁ and the character e ₁ The acquisition step, and for the integer i (2 ≦ i ≦ s), stored in a one-to-one correspondence with the pair (n _i , e _i ) of the previously acquired node n _i and the character e _i . comprising a sequential acquisition step of sequentially acquiring the node n _{i + 1,} a data output step of the obtained node n _{s + 1} to obtain the stored information data d to correspond to one-to-one output, the , Spelling information data stored in association with a certain reading
When referring to,
It used after adding the terminator ", reading of the information data stored in association with some spelling
When you draw a
A dictionary lookup method characterized by using "the one with the addition of a terminal character" . 3. In the case where a node accumulated in a one-to-one correspondence cannot be acquired in the head acquisition step or the sequential acquisition step, it is reported that there is no information data stored in association with the word or phrase. The dictionary lookup method according to claim 2, further comprising a reporting step of: 4. A word / phrase acquisition method for obtaining a word / phrase held in association with the information data d held in a recording medium storing the dictionary data structure according to claim 1, the information data d to correspond to one-to-one, or a node acquisition step of acquiring node m ₀ stored is referred to the information data d, stored in correspondence with one-to-one to the node m ₀ of the acquired For the leading acquisition step of acquiring the pair (m ₁ , e ₁ ), and the integer j (1 ≦ j), 1 is set to the node m _j acquired earlier.
A sequential acquisition step of sequentially acquiring pairs (m _{j + 1} , e _{j + 1} ) stored in association with pair ₁ , and a start element n _k of one of the pairs from which the start node n ₁ is sequentially acquired , The sequence of the last elements of the pair obtained sequentially, e _k , e _k-1 , ... E ₂ , e ₁ of the character representing the word held in association with the information data d includes a character string output step of outputting as a column, and wherein in the character string output step, information of the information data is read
If it is data, remove the terminator at the end of the resulting phrase.
If the information string is the spelling information data, the obtained string of characters is output as the spelling.
Read the string of characters with the trailing terminating characters removed.
The phrase acquisition method is characterized by outputting as . 5. A recording medium storing the dictionary data structure according to claim 1, which is held in association with a word or phrase expressed by a character string e ₁ , e ₂ , ... E _{s having} a length s. A dictionary look-up device for drawing information data, which acquires a node n ₂ accumulated in a one-to-one correspondence with the pair (n ₁ , e ₁ ) of the start node n ₁ and the character e ₁ Acquiring unit, for the integer i (2 ≦ i ≦ s), stored in a one-to-one correspondence with the pair (n _i , e _i ) of the previously acquired node n _i and the character e _i . A sequential acquisition unit that sequentially acquires the nodes n _{i + 1,} and a data output unit that acquires and outputs the information data d stored in one-to-one correspondence with the acquired nodes n _{s + 1} ,
And spelling information data that is stored in association with a certain reading
When referring to,
It used after adding the terminator ", reading of the information data stored in association with some spelling
When you draw a
A dictionary lookup device characterized by using "the one with the addition of a terminal character" . 6. If the head acquisition unit or the sequential acquisition unit cannot acquire information because there is no node accumulated in a one-to-one correspondence, there is no information data stored in association with the word or phrase. The dictionary lookup apparatus according to claim 5, further comprising a reporting unit that reports the effect. 7. A word / phrase acquisition device for obtaining a word / phrase held in association with the information data d held in a recording medium storing the dictionary data structure according to claim 1, the information data d In a one-to-one correspondence with each other, or with reference to the information data d, to acquire the accumulated node m _0, and a node acquisition unit, and in a one-to-one correspondence with the acquired node m ₀ For the head acquisition unit that acquires the pair (m ₁ , e ₁ ), and the integer j (1 ≦ j), 1 is set for the node m _j acquired earlier.
A sequential acquisition unit that sequentially acquires pairs (m _{j + 1} , e _{j + 1} ) stored in association with pair ₁ , and the start node n ₁ is the head element m of one of the sequentially acquired pairs. _When it matches with _k , the sequence e _k , e _k-1 , ... E ₂ , e ₁ of the last element of the pair obtained in sequence is expressed as the word held in association with the information data d. A character string output unit for outputting as a character string , wherein the character string output unit is an information data reading device for reading the information data.
Remove the trailing terminator of the resulting phrase,
Character of the string is output as a spelling, when the information data is information data for spelling, obtained
Read the string of characters with the trailing terminating characters removed.
The phrase acquisition device is characterized by outputting as . 8. The dictionary data structure includes an array BASE and an array CH.
ECK is used to hold a word, a node, and its information data, and a start node n ₁ and a character string c ₁ , c ₂ , ..., c _s representing the held word, and The information of the associated nodes n _2, ..., N _s , n _{s + 1} is represented by integers, and for the integer i (1 ≦ i ≦ s), the n _i- th element BASE [n of the array BASE _i ] and the n _{i +} 1th element CHECK [n of the array CHECK
_{i + 1} ] and two conditions n _{i + 1} = BASE [n _i ] + c _i CHECK [n _{i + 1} ] = n _i are held so that both are satisfied. The recording medium according to claim 1. 9. A recording medium storing the dictionary data structure according to claim 8, which is held in association with a word or phrase expressed by a character string e ₁ , e ₂ , ... E _s of length s. A dictionary lookup method for pulling information data, wherein, for an integer i (1 ≦ i ≦ s), the sum of the n _i- th element BASE [n _i ] of the array BASE and the character e _i t _i = BASE [n _i ] + e _i is the condition
When CHECK [t _i ] = n _i is satisfied, a sequential acquisition step of sequentially acquiring as integer n _{i + 1} = t _i , and information accumulated in a one-to-one correspondence with the acquired integer n _{s + 1} A data output step of acquiring and outputting the data d,
A dictionary lookup method comprising: 10. The method according to claim 9, further comprising a reporting step of reporting that there is no information data held in association with the phrase when the condition is not satisfied in the sequential acquisition step. How to look up the dictionary. 11. A phrase acquisition method for obtaining a phrase retained in association with an information data d retained in a recording medium storing the dictionary data structure according to claim 8, the information data d The node acquisition step of acquiring the integer m ₀ expressing the information of the node accumulated in one-to-one correspondence with or in reference to the information data d, and the integer j (0 ≦ j) are acquired first. From the integer m _j
The integer m _{j + 1} = CHECK [m _j ] and the character integer e _{j + 1} = m _j -BASE [m
_j ] and a sequential acquisition step of sequentially acquiring, and an integer n ₁ expressing the information of the start node, and any of the sequentially obtained integers m _k , if the sequentially obtained character integer A character string output step of outputting the strings e _k , e _k-1 , ... E ₂ , e ₁ as a string of characters expressing a word held in association with the information data d. How to get words and phrases. 12. A storage medium, which stores the dictionary data structure according to claim 8, is held in association with a word or phrase represented by a character string e ₁ , e ₂ , ... E _{s having} a length s. A dictionary look-up device for drawing information data, wherein, for an integer i (1 ≦ i ≦ s), the sum of the n _i- th element BASE [n _i ] of the array BASE and the character e _i t _i = BASE [n _i ] + e _i is the condition
When CHECK [t _i ] = n _i is satisfied, a sequential acquisition unit that sequentially acquires as an integer n _{i + 1} = t _i , and information accumulated in a one-to-one correspondence with the acquired integer n _{s + 1} A dictionary lookup device, comprising: a data output unit that acquires and outputs data d. 13. The sequential acquisition unit further comprises a reporting unit for reporting that there is no information data held in association with the phrase when the condition is not satisfied. The dictionary lookup device described. 14. A word / phrase acquisition device for obtaining a word / phrase held in association with the information data d held in a recording medium storing the dictionary data structure according to claim 8, the information data d The node acquisition unit that acquires the integer m ₀ that represents the information of the node that is stored in a one-to-one correspondence with or in reference to the information data d, and the integer j (0 ≦ j) From the integer m _j
The integer m _{j + 1} = CHECK [m _j ] and the character integer e _{j + 1} = m _j -BASE [m
_j ] and a sequential acquisition unit that sequentially acquires the integer n ₁ that represents the information of the start node, and any of the sequentially obtained integers m _k , if the sequentially obtained character integer A character string output unit that outputs the strings e _k , e _k-1 , ... E ₂ , e ₁ as a string of characters expressing a word held in association with the information data d. A phrase acquisition device. 15. A computer recording a program for causing a computer to function as the dictionary lookup device according to claim 5, 6, 12 or 13 or the word acquisition device according to claim 7 or 14. A readable recording medium.