JP3260428B2 - Information retrieval processor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval processing apparatus, and more particularly to an information retrieval processing apparatus suitable for retrieving a character string or the like existing in a database such as a document or correcting a character string.

[0002]

2. Description of the Related Art Conventionally, in a data processing system, when searching for data (keywords) of a target character or the like in a database, it is configured to sequentially search all data or text stored in advance. A search method is generally employed. FIG. 12 is a conventional configuration diagram to which a sequential search method of a keyword is applied.
In FIG. 12, a keyword as character data to be searched is input to the matching unit 51.
The matching unit 51 is connected to a text memory 52 which stores text, that is, data such as characters stored in advance. Then, the matching unit 51 reads out the characters of the document one by one and determines whether or not the characters match the keyword.

However, in the conventional sequential search method, all characters in the text and the keyword are sequentially compared to determine whether or not the keyword exists. are doing. In particular, as the number of characters in the text increases, the search time also increases. Therefore, when searching for a large-capacity text, it takes a lot of time to search. In order to solve such a problem, Japanese Patent Application Laid-Open No. 64-35627 discloses a technique for performing a search in a short time.

FIG. 13 is a functional block diagram of a device employing a new search method. In FIG. 13, a keyword composed of a plurality of characters is input to a character chain extraction unit 53. The character chain extraction unit 53 includes a specific number of character strings composed of the plurality of characters and a keyword. Can be extracted. The character chain index 54 stores where in the text a particular number of consecutive characters present in the text to be searched are located. The index search means 55 can search the character chain index 54 based on the character string extracted by the character chain extraction means 53, and can search the position of the keyword in the text from the positional relationship of the plurality of character strings. ing.

[0005]

However, in the prior art, the search can be performed at a higher speed than in the sequential search method. However, when the text is corrected, all the pointers indicating the character positions after the correction must be updated. The problem arises that it must be done. That is, when the correction occurs, the character position of the character chain after the correction portion changes, and therefore, of the pointers corresponding to the character chains stored in the character chain index 54,
In the modified portion, all pointers corresponding to the character chain after the modified portion must be updated.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an information retrieval processing device capable of speeding up retrieval and shortening the time required for modifying a document.

[0007]

To SUMMARY OF THE INVENTION To achieve the above object, the present invention, a text file and the data of the text file be stored with the location data of the data string is composed of a plurality of characters such as the divided for each data specific character or a specific character string and the text generation means for generating a plurality of sub-text, in the sub-text of this character chain to consist character chain of a plurality of specific character from the sub text position and Rubun shaped linkage extracting means to extract character chain position 置De over data indicating a correspondence the character chain and character chain position data by extracting the character chain extracting means <br/> position location in the text a character chain storage means for storing, of the input
When performing a search process using the specified keyword,
Character chains and character chains extracted from keywords by the extraction means
Searching the character chain storage means from the position data ,
In the information retrieval apparatus which includes a index search means, for extracting the character chain of the same character chain by extraction of the character chain extracting unit together with the position data, before
Modify the substring to be modified to the modified string
Character chain correction means,
, The character chain extracting means, from the correction target character string, text
Character chain position data that returns the character chain and the position in the character string to be corrected
And the index search means extracts the sentence
The character of the correction target character string extracted by the character chain extraction unit
Based on the chain and character chain position data, the character chain memory
Search column and position data in the text of the character string to be corrected
And the text generating means extracts the text file.
Extracted from the file by the index search means.
Extracting sub text corresponding to the position data, the character
The chain correction means is extracted by the index search means.
Using the position data obtained and the corrected character string,
Modify the subtext extracted by the list generation means
Creating a modified sub-text, wherein the character chain extracting means comprises:
Fixed Sabute kiss modified by the character chain modification means
Characters that represent the position in the character chain and the modified subtext from the
Chain position data and the character chain correcting means,
Corrected subtext extracted by the character chain extraction means
The character chain and character chain position data of the
Modify subtext by storing it in storage
This constitutes an information retrieval processing device .

[0008]

[0009]

First, as a registration process, when character string data and its position data are stored in a text file, the data of the text file is input to the sub-text generating means. When the data of the text file is input to the subtext generating means, a plurality of subtexts are generated by dividing the data of the text file for each data of a specific character or a specific character string. The data of each subtext is input to the character chain extracting means. The character chain extracting means receives data relating to the character string of each subtext, and relates to a character chain composed of a specific number of characters from the data of each subtext and a character chain position indicating the position of the character chain in the subtext. Extract data. For example, a plurality of character strings in a three-character chain and the position of the character string in the subtext are obtained. Then, the character chain position data of the extracted character chain is converted into position data in the text, and the converted character chain data of the position data is stored in the character chain storage means in association with the position in the text.

Next, as a search process, when a keyword including data relating to a designated character chain and the position of the character chain in the text is output from the keyword output means, a character composed of a plurality of specific characters is obtained from the keyword. Data relating to the chain and its character chain position is extracted by the character chain extracting means. For example, a character string of a three-character chain and the position of the character string in the keyword are obtained. The character chain storage means stores a character string existing in the text and its position data, and reads the positions in the text of a plurality of character strings input from the character chain extraction means from the character chain storage means. If the character string extracted from the above-described text is, for example, in units of three characters, three consecutive characters from the beginning, three consecutive characters one character later, and three consecutive characters two characters later. Since the character string is shifted in units of one character and is in units of three characters, when the keyword is located in the text, the position data read out from the character chain storage means has the continuity with the position data of the character string in the keyword. Will match. Therefore, when the index search means obtains the continuity from the positional relationship, it is possible to search for position data having the same character string as the keyword in the text.

Next, the correction processing will be described. When a plurality of correction target sub-texts obtained by dividing data of a text file for each specific character or specific character string data are input, the correction target character string is converted into a plurality of consecutive character strings of a specific number of characters, for example, 3 characters. The character string of the character chain and the position of the character string in the correction target character string are obtained. These data are stored in the character chain storage means in association with the position in the subtext. Then, similarly to the search processing, when the index search means searches the position of the character string in the correction target character string and the position data of the same character string as the correction target character string in the text, the correction target character string exists. The data of the subtext number is input to the character chain correcting means. When the correction target subtext generating means extracts a subtext corresponding to the subtext number from the text file, data relating to the subtext is transferred to the character chain extracting means. And by the character chain extraction means,
Multiple consecutive strings of a certain number of characters in the subtext,
For example, a character string of a three-character chain and the position of the character string in the subtext are obtained. Then, all the position data having the subtext number stored in the character chain storage means are deleted. Next, when a corrected character string is input, data relating to the corrected character string is input to the character chain extracting means. The corrected character string is obtained from a plurality of consecutive character strings having a specific number of characters, for example, a three-character chain character string and the position of the character string in the subtext. These data are output to the character chain correcting means. In this character chain correcting means, a corrected subtext is generated from the character string position where the obtained correction target character string exists, the subtext, and the corrected character string. Then, a plurality of character strings having a specific number of characters and position data in the text are transferred to the index search means from the subtext. The index search means stores the character string and the position data in the character chain storage means. Modifications of this text remain in the subtext, ie,
Text files divided for each character or specific character string data
Only the data of the file can be corrected, and the sub text after the corrected portion is not affected, so that the text can be corrected in a short time. In addition, the insertion and deletion of a character string can be performed in a short time by deleting or adding the position data in the corresponding subtext, similarly to the correction.

[0012]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is an overall configuration diagram of an information search processing device according to an embodiment of the present invention, FIG. 2 is a configuration diagram showing a configuration of a registration processing example of the present invention, and FIG. 3 is an operation explanation for explaining the operation of FIG. FIG. In FIG. 1 and FIG. 2, reference numeral 12 denotes a text file that stores character string data including a plurality of characters and the like together with its position data. When the data of the text file 12 is input to the subtext generator 13, the text data is divided into a plurality of subtexts for each specific character or specific character string data. The data of each subtext is transferred to the character chain extraction unit 14, and data on a character chain composed of a plurality of specific characters and a character chain position indicating the position of the character chain in the subtext is extracted from each subtext. You. When the extracted data is transferred to the character chain correcting unit 17, the character chain position data of the character chain is converted into position data in the text. That is, the character chain correction unit 17 is configured as a position data conversion unit. The character data whose position data has been converted by the character chain correction unit 17 is stored in the character chain index 15 in association with the position in the text. That is, the character chain index 15 is configured as character chain storage means. When a keyword including a designated character chain and data on the position of the character chain in the text is input to the character chain extracting unit 14, the character chain extracting unit 14 converts the keyword into a character chain composed of a plurality of specific characters. And data on the character chain position are extracted. That is, the character chain extraction unit 14 constitutes a first character chain extraction unit and a second character chain extraction unit. Then, the data extracted by the character chain extraction unit 14 is stored in the index search unit 1.
6, the same character chain as the character chain extracted by the character chain extraction unit 14 is extracted from the character chain index 15 together with the position data. That is, the index search unit 16 is configured as an index search unit.

In the above configuration, the registration processing of this embodiment
This will be described with reference to FIGS. First, the registration text
Input from the text file 12 to the sub text generator 13
Then, the sub-text generation unit 13
Searching for a preset character or character string representing the end
Done. In FIG. 3, the period "."
And Then, the sub-text generation unit 13
When a dot "." Is detected, every time a period is detected,
A character string composed of all the characters that referred to the character string
Text and the order of appearance of the subtext in the text
Is the subtext number. For example, as shown in FIG.
Kist (...
Kikukeko ... …), The sub-text number
The number is (... i ₁, I_TwoAnd the subtext number i₁When
i_TwoThe sub texts are (...
・ ・. ) And (... Kikukukeko ...).

After the registration process is completed, a process for generating a plurality of subtexts is performed as shown in FIG. First, when text data is input from the text file 12 to the subtext generator 13, an initial set (step) OP1 is performed. That is, the subtext number cnt is “1”, the character position i from the beginning of the text is “2”, and the subtext end character string is p ₁ to p
to _k, it is set to "0" to match the number of times j of up to detect the p ₁ ~p _k. And strings and sub-text-terminated string p ₁ of k characters taken from the text (k are the same as k of p _k)
Determines whether ~p _k coincide in step OP2.
When it is determined not to match the process, the character string matching text and sub text terminated string p ₁ ~p _k is determined whether to end at step OP3. If it is determined in this step to continue, the character position i and the number of times of collation j are set to +1 in step 4. Step OP2
Repeating the subsequent processing, it determines whether there is a character string and the sub-text terminated string p ₁ ~p _k of k characters match to retrieve from the text to the end of the text. Next, in step OP2, the character string and the sub-text terminated string p ₁ ~p _k of k characters taken from text matches, in step OP5, consists of a character obtained by the matching string to date The sub-text “C _ij -C _{i + k−1} ” and the sub-text number cnt are the text segmentation results so far. Next, in step OP6, to search for the next subtext, the subtext number cnt is incremented by 1, and the number of collations j "-
In step OP3, when the character string collation up to the end of the text is confirmed, in step OP7, the subtext and subtext number created so far in step OP5 are set as the result of subtext division.

Further, as shown in FIG. 3, when the subtext and the subtext number are transferred from the subtext generator 13 to the character chain extractor 14, the character chain extractor 1
In step 4, a character chain composed of a plurality of specific characters and character chain position data indicating a position in the subtext are obtained from the subtext. For example, when a three-character chain is obtained in FIG. 3, the character chain and its position are:
・, <Ai, i _{1 to} pt ₁ >, <Iue, i _{1 to} pt
_2>, <Ueo _{i 1 ~pt 3>, ···,} < Kakiku i ₂ ~p
t ₁ '>, <Kikuke i ₂ ~pt _2'>, <Kukeko i ₂ ~pt
₃ '>, ... Here, pt ₁ and pt ₁ ′ indicate the position of the character chain.

Next, the operation of the character chain extraction processing will be described with reference to FIG. When the input S, that is, the sub text is input to the character chain extraction unit 14, an initial set OP8 is performed. That is, the number of characters a in the character chain is set to “3”,
The pointer i indicating the head of the character chain of the input S is set to 1.
Then, it is determined in step OP9 whether or not it is the last of the character chain extraction. Since the pointer i is changed by an operation described later and indicates the head of the character chain, it is determined in step OP9 whether or not the keyword is continuous by a character behind including the specified position. . When the sub text has 10 characters (...,...), N is 10, so that i + a−
Since 1> n is not satisfied (NO), next, a process of storing data for three characters from the position indicated by the pointer i in the register X is performed in step OP10. The register X is a matrix register, and changes from j = 1 to a, that is, 1, 2, 3, and stores three characters of inputs S (1) to S (n) first. When i = 1, X (1,1) ← S (1), X
(1,2) ← S (2) and X (1,3) ← S (3). In this operation, the three characters from the beginning are stored in the register X as one character chain. Then, the operation of incrementing the pointer i is performed in step OP11. That is, an operation for pointing to the next character is performed. Then, the discriminating operation is performed again in step OP9. By this repetitive operation, next, X (2,1) ← S (2), X (2,2)
2) ← S (3), X (2,3) ← S (4) are performed.

If the subtext is 10 characters and the pointer i indicates 9, i + a-1 = 11, n = 1
Since it is 0, i + a-1> n is satisfied (YES), the process of outputting X to the index search unit 16 is performed in step 12, and all the operations are completed.

With the above operation, the index search unit 1
5, a character chain and its position data are stored. Note that
The value of i in the trix register X (i, j) is
Indicates the position of the character chain. And character chain index
As shown in FIG. 3, a character chain (..
・ 、 Ah, it's like ... ・ ・ ・ Kikuke, Kukeko ...
・) And the position in the text
Pointer (..., i ₁-Pt₁, I₁-Pt_Two, ...
・ I_Two-Pt_Two’, I_Two-Pt_Three’, ...)
You. For example, like that, Iue, Kikuke, Kukeko
Correspondingly i₁-Pt₁, I ₁-Pt_Two, I_Two-Pt₁’, I_Two−
pt_Three'Is stored. Note that the character chain index 1
5, in addition to three-letter chains, one-letter and two-letter chains
Also remember their character chains and pointers so they can be handled
Have been.

Next, the retrieval process will be described with reference to FIGS. First, the keyword is extracted from the keyword output means in the character chain extraction unit 14, from the keyword, a character chain composed of a plurality of specific characters and character chain position data indicating a position in the keyword. For example, FIG.
When the three-character chain is obtained in step (1), the character chain and its character chain position data are <A, 1> and <I, 2>. Here, "1" and "2" represent character chain position data. In the character chain extraction operation, the same processing as in FIG. 5 is performed. The character chain and character chain position data are converted from the character chain extraction unit 14 to the index search unit 16 by using the above-described character chain (Ai, Iue). Assuming that P and Q are pointer strings, P and Q are obtained. The pointer strings P and Q generally consist of a plurality of pointers, and the pointer string P is p (1)
~ P (l). The pointer sequence Q is q
It is a set of (1) to q (m). This pointer string indicates the position in the text where the character chain of “Aoi” and “Iue” exists. When the keyword “Aiue” exists in the text, p (l) −
There is a pointer satisfying q (m) =-1. The extraction of the pointer is also performed by the index search unit 16.

FIG. 8 is a flowchart showing the operation of pointer extraction. When the index search unit 16 obtains the pointer strings P and Q corresponding to the character chain by searching the character chain index 15, the operation of setting i, j and h is then performed in step OP13. i and j are pointers for specifying the pointer strings p (1) to p (l) and q (1) to q (m), respectively, and h is the target character string, that is, the same character chain as the keyword exists. sometimes,
Processing result register Y (h) storing the character chain position
Is a pointer for instructing. After the initial set operation is performed in step OP13, the pointer i,
It is determined in step OP14 that j is i> 1, j> m, respectively.
Done in l and m represent the final pointers of the pointer strings P and Q, and when both are not satisfied, i, j
A determination is made in step OP15 as to whether or not the pointers p (i) and q (j) indicated by satisfies the conditions for a series of character chains in the keyword. That is, using the character chain position data ax and bx extracted from the keyword, the difference between the character positions of the pointer strings P and Q is calculated as ao ← p (i) -ax,
Correct as bo ← q (j) −bx. If the values of the registers aa and bo are the same in Steps OP16 and OP18, it is a continuous series of character chains. Therefore, in Step OP20, ao at this time is stored in the processing result Y (h), and Y (h) is stored. ) Is set to +1. If ao is larger than bo in step 16, the pointer p
Since (i) indicates the point in the sentence from the pointer q (j), the pointer j of the pointer sequence Q is set to +1 in step OP17. On the other hand, in step 18, bo is a
When it is determined to be larger than o, the pointer q (j)
Indicates the point in the sentence from the pointer p (i), the pointer i of the pointer sequence P is set to +1 in step OP19. When ao and bo match, the pointers p (i) and q (j) indicate the same series of character chains, and they should be the next pointers. j is set to +1. By repeating such an operation, the position of the same character chain as the keyword is obtained, and the respective position data is stored in the register Y.

On the other hand, at step OP14, i> 1,
When one of j> m is satisfied, there is no more pointer in the satisfied pointer row, and the processing for outputting the processing result Y is performed in step 22. With this operation, the processing of the index search unit 16 ends.

Next, the correction processing will be described. FIG. 9 is a block diagram showing an embodiment of the correction processing of the present invention, and FIG. 10 is an operation explanatory diagram showing the operation. The sub-text generating unit 13 constitutes a correction target sub-text generating unit, and the character chain correcting unit 17 constitutes first and second correction target character string data generating units and data correcting unit. The character string to be corrected represents a character string to be corrected, and the character string to be corrected is a character string for rewriting the character string to be corrected. Note that the insertion processing is performed when there is no correction target character string, and the deletion processing is performed when there is no correction character string. When the correction target character string is input to the character chain extraction unit 14, the character chain extraction unit 14
A character chain composed of a plurality of specific characters and character chain position data indicating a position in the correction target character string are obtained. For example, in FIG. 10, when three character chains are obtained, the character chain and the character chain position data are <A,
1><Ue,2>.

When these character chains and character chain position data are input to the index search unit 16, a pointer indicating the location where the character string to be corrected exists in the text is obtained from the character chain index 15 in the same manner as in the search processing. The process for obtaining is performed. In Figure _{10, i 1 -p t1, i} 1 -p
_t2 is a pointer. This pointer is used as the character chain correction unit 1
When 7 is input, the subtext number i ₁ is transferred to the subtext generator 13. The sub text generation unit 13
From the text file 12, this subtext number i ₁
Is extracted, and the character chain correcting unit 17
Output to At this time, the corresponding sub text is, for example, (...,...).

When the character chain correcting unit 17 inputs data relating to the correction target subtext, these data are transferred to the character chain extracting unit 14. The character chain extraction unit 14
From the data of the sub-text to be corrected, a character chain composed of a plurality of specific characters and character chain position data indicating a position in the character string to be corrected are obtained, and these data are transferred to the character chain correcting unit 17. For example, in FIG. 10, when the sought character chain 3 characters, character chain and character chain position data ... <abc, i ₁ -p _t1>, <Iue i ₁ -
p _t2 > _,. Then, the character chain correcting unit 17
Based on the character chain and character chain position data before correction, the character chain and character chain position data before correction stored in the character chain index 15 are deleted.

Next, the character chain modifying unit 17, by using the correction target string pointer i ₁ -p _t1 and i ₁ -p _t2 input from just the index search unit 16, the target modified from the correction target sub text Delete the string. Then, a separately input correction character string is inserted at the pointer position of the correction target character string to generate a correction subtext, and the generated correction subtext is transferred to the character chain extraction unit 14. For example, in FIG. 10, since the correction character string is abc, the correction sub-text is (... abc ...). Then, the character chain extraction unit 14 obtains a character chain composed of a specific number of characters and character chain position data indicating a position in the corrected subtext from the corrected subtext. For example, FIG.
When sought character chain three characters is 0, and its position data character chain _{···, <abc, i 1 -p} t1>, <bc
, I ₁ -p _t2>, become .... Next, when the character chain correcting unit 17 receives the character chain of the corrected subtext and the character chain position data from the character chain extracting unit 14, the pointer sequence corresponding to the corrected character chain of the character chain index 15 contains
The corrected character chain position data is stored, sorted in ascending order in the pointer column, and a corrected character chain index is created.

FIG. 11 is a flowchart showing the operation of extracting the subtext to be corrected in the correction processing of the present invention. In FIG. 11, when text data is input from the text file 12, initial setting is first performed in step OP23. That is, the sub text number cn
t to "1", the character position i from the beginning of the text to "1", the sub-text-terminated string to p ₁ ~p _k, p ₁
The processing number j until detection of the ~p _k to "0", and sets the extracted sub text number I input from character chain modification unit 17. Then, k characters (k
Is the same as K of p _k ), and a determination is made in OP24 as to whether or not the subtext end character strings p _{1 to} p _k match. If it is determined in this step that they do not match, in step OP25, the text and the subtext end character string p ₁
String matching with ~p _k it is determined whether finished.
If it is determined that the process is to be continued, the process proceeds to step OP2.
In step 6, the character position i and the number of collations j are set to +1. Further steps OP42 repeat the subsequent processes until the end of the text, whether the judging string and sub text terminated string p ₁ ~p _k of k characters match to retrieve from the text is performed.

[0027] When the character string and the sub-text terminated string p ₁ ~p _k of k characters extracted from the text matches in step OP 24, in step OP27, number of sub-text matches the extracted sub text number I Is determined. If it is determined in this step that they do not match, in step OP28, the subtext number cnt is incremented by 1 to search for the next subtext.
The number of times of collation j is set to “−1”. In step OP 25, a determination is string matching of text and sub text terminated string p ₁ ~p _k is whether to end. Further step OP
The processing after step 24 is repeated, and the character string is collated until the end of the text.

When the sub-text number cnt matches the extracted text number I in step OP27, in step OP29 the matching sub-text "C
Outputs _{ij ~C i + k-1 "} , and ends the extracted correction target sub text. Note that when a string matching the text and sub text terminated string p ₁ ~p _k ended in step OP 25, step In OP30, an error message “not applicable” is output, and the processing in this routine ends.

In this embodiment, the sub-text terminating character string is a period (.), And a sub-text is constructed for each sentence. A sub text can also be configured for each. Further, for example, a sub-text can be configured for each use unit by inserting a use unit delimiter character string in consideration of the use unit in the registered text. further,
For example, a use unit delimiter character string may be inserted into the registered text in consideration of the use unit, and a subtext may be configured for each position use unit.

Further, in the above embodiment, the case where the number of characters of the character chain index is set to three or less has been described. However, a character chain of four or more characters can be searched by the same processing as in the above embodiment. At the time of retrieval, a character chain can be extracted from a keyword such that the degree of overlap of the character chain is minimized. For example, when the keyword is “Aieo”, the character chain can be “Aoi” and “Ueo”.

Further, the present invention can be applied not only to character chains but also to general data.

[0032]

According to the present invention, as is apparent from the above embodiment, a text is divided into a plurality of subtexts by a preset subtext end character string, and a character position is managed for each subtext. To correct the text
At the time of insertion / deletion, the change of the character position after the correction / insertion / deletion part is kept in the subtext, so that the text correction / insertion / deletion can be shortened. Also, in the case of the proximity operation search processing in which all keywords are required to be present in the same sentence or the same paragraph, it is only necessary to check whether or not all the subtext numbers of the pointers obtained from all keywords match. This can also contribute to speeding up the search processing.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram showing an embodiment of the present invention.

FIG. 2 is a configuration diagram showing a registration processing example of the present invention.

FIG. 3 is an operation explanatory diagram for explaining an operation of a registration processing example of the present invention;

FIG. 4 is a flowchart illustrating a sub-text division process.

FIG. 5 is a flowchart for explaining the operation of character chain extraction.

FIG. 6 is a configuration diagram showing a search process according to the present invention.

FIG. 7 is an operation explanatory diagram for explaining the operation of the search processing according to the present invention;

FIG. 8 is a flowchart for explaining the operation of pointer extraction;

FIG. 9 is a configuration diagram illustrating an example of a correction process according to the present invention.

FIG. 10 is an operation explanatory diagram for explaining an operation of a modification processing example of the present invention;

FIG. 11 is a flowchart illustrating an operation of extracting a correction target subtext.

FIG. 12 is a configuration diagram of a conventional example.

FIG. 13 is a configuration diagram of another conventional example.

[Explanation of symbols]

 12 Text file 13 Subtext generator 14 Character chain extraction unit 15 Character chain index 16 Index search unit 17 Character chain correction unit

Claims (1)

(57) [Claims] 1. A text file that stores data of a string that consists of a plurality of characters such as together with its position data, a plurality of divided data for each data of the specific character or a specific character string of the text file a text generation means for generating a sub-text, Rubun characters to extract character chain position 置De over data indicating a position in the sub-text of this character chain to consist character chain of a plurality of specific character from the sub text linkage extracting means, a character chain storage means for storing in association with the character chain and character chain position data by extracting the character chain extracting means position in the text, the time of the search processing by the input keyword, Said sentence
Character chain and sentence extracted from keyword by character chain extraction means
And a character chain position data, by searching the character chain storage unit, index search means and, in which the information retrieval process comprising the extracting character chain of the same character chain together with its position data by extracting the character chain extracting means In the device , the correction target character string of the sub text is corrected to a correction character string.
Character chain correction means for performing correction processing.
The character chain extracting means may be configured to convert a character string
Character chain position data that returns the position in the chain and the string to be modified
And the index search means sends the character chain extraction means
Character chain and character chain position of the string to be corrected extracted from
Search and correct the character chain storage means from the location data
Extracting position data in the text of the target character string, the text generating means, from the text file,
Position data extracted by the index search means
Is extracted, and the character chain correcting unit sends the sub-text to the index searching unit.
Using the extracted position data and the corrected character string,
Sub text extracted by the text generation means
Modify to create a modified sub-text, wherein the character chain extracting means is
Character chain and correction sub-text from the corrected correction sub-text
Extracts the character chain position data representing the position in the text, the character chain correcting means, by the character chain extracting means
Character chain and character chain position of extracted modified subtext
By storing data in the character chain storage means,
And an information retrieval processing device that corrects sub text.