JPH05266067A - Data retrieving device - Google Patents

Abstract

PURPOSE:To efficiently retrieve desired data from a dictionary within a minimum retrieval range by a corresponding address concerning the relevant flag when there are the corresponding data by firstly checking whether there are data corresponding to a retrieval code or not by the setting of the corresponding flag. CONSTITUTION:A retrieval part 14 detects the block corresponding to the retrieval code from a corresponding part 16 with the front part of the retrieval code inputted from an input part 11 as a retrieval key. Next, the corresponding flag of the detected corresponding block is checked by one part next to the utilized front part of the retrieval code, when the relevant flag shows zero, it is expressed that there are no corresponding data, and a message showing that the relevant data are not discovered is displayed by an output part 19. When the relevant flag is '1', it means that there are the corresponding data, and the data in this range are read from a dictionary 18 by the corresponding address of the relevant flag and the next address. A detection part 17 detects the corresponding data coincident to the retrieval code.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data retrieval device capable of detecting information corresponding to an input / translation system such as a language translation device, a word processor and a desktop publishing system at high speed.

[0002]

2. Description of the Related Art The practicality of a data retrieval device is concerned with the storage space and retrieval speed of data. Conventionally, as a storage method and a search method for large-capacity dictionary data, there is a method described in Japanese Patent Laid-Open No. 62-197822. A system block diagram of the embodiment is shown in FIG. 4 (a). The master index, the sub index, and the dictionary body are stored in a file as shown in FIG. 4B, and the dictionary body 312 is divided into blocks of fixed length, and only the head entry word of each block is stored in the sub index 311. Remember. The sub index is also divided into blocks having a fixed length, and only the head entry word of each block is stored in the master index 310. A11, a21, a in FIG. 4 (b)
Since 31, ... Are all headword strings in the dictionary body, they are stored in the sub-index. The head entry words a11, b11, ..., X11 of the sub index are similarly stored in the master index.

The master index file 310 is read into the master index area 307, and it is sequentially determined that the character string to be searched for is the same as the nth index word in the master index file or between the nth and (n + 1) th. Search (primary). Next, the nth sector of the sub index file is read into the buffer area 308. The character string to be searched matches with the m-th index word string of the read sub index, or the m-th and (m +
1) It is sequentially searched for the second position (secondary).
Next, the (k + m) th sector of the dictionary data file is read using the offset value k for the dictionary data file set for each sector of the sub index. The headword of the read dictionary data file is compared with the character string to be searched, and the matching headword is searched (third order).

[0004]

In the above-mentioned conventional example, if the input utterance symbol string is used for the retrieval without being processed in advance, the collation stored in the master index, the sub-index, and the dictionary body will be used. Since the length of the voicing symbol string is not determined, the voicing symbol strings having different lengths are searched three times in sequence, so that the search space cannot be reduced sequentially. Of course, the search efficiency cannot be improved.

Further, when the original utterance symbol string is used as it is for searching, the length of the utterance symbol string in the dictionary body is indefinite and stored in a block of a fixed length.
A space of indefinite length remains at the end of each block, wasting dictionary space.

[0006]

In order to solve the above problems, the present invention comprises a dictionary storing index codes and corresponding data, and a plurality of blocks, each block comprising a plurality of bits. Corresponding flag group that indicates whether or not there is data corresponding to the search code according to the flag value of 0 or 1 for each bit, and if the flag value is 1, the storage position of the index code and the corresponding data in the dictionary The corresponding portion formed of corresponding address groups for storing the corresponding search portion, the search portion for detecting the corresponding block of the corresponding portion by the front portion of the input search code, and the detected portion obtained by comparing a part of the input search code. A data retrieving apparatus, comprising: a detection unit that extracts a corresponding flag and a corresponding address of a copy and detects corresponding data of a search code from a dictionary.

[0007]

According to the above-mentioned structure of the present invention, the front part of the input search code is used as the search key to detect the corresponding block of the corresponding part by the search part, and then the search code used by the detection means is searched. The next part of the part is used to extract the corresponding address of the corresponding part and determine the minimum search range for the dictionary. Finally, the corresponding data is detected from the dictionary by comparison with the search code of the next part, and the search result is output by the output unit. A large capacity dictionary data file can be searched at high speed.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The operation of the present invention will be described by taking a search of Chinese dictionary data as an example. In the present embodiment, the Chinese reading symbols are coded and the search comparison code, the word, and the frequency of use of the word are used as the corresponding data.

There are over 1,300 valid readings in Chinese, and every reading must be represented by two bytes.
This method can save the storage space of the dictionary and is convenient for data retrieval. Then, the Chinese vowels and Chinese vowels are placed together in one byte, the vowels and tones are placed in another byte, and are placed in the ASCII character table as shown in FIG. Each reading will be able to be displayed with a unique 2-byte representative code. Here, this 2-byte representative code is referred to as a search code. The first, second, and third bytes of the converted search code are referred to as first, second, and third search codes.

One sequence value is given to each of the Chinese voicing symbols shown in FIG. 6, and the input voicing symbol string is converted into the corresponding retrieval code by the simple judgment and calculation as shown in FIG. can do. reading

[0011]

[Outer 1]

An example will be described. Referring to the sequence value in FIG.

[0013]

[Outside 2]

Is the tenth item in the order of the voice mother,

[0015]

[Outside 3]

Is the second of the ensemble order, the first search code is transformed as follows. 21H + 10 ^* 4 + 2 = 4bH, 4bH corresponds to "K" in the ASCII code.

Rhyme

[0018]

[Outside 4]

Is the eighth in the rhyme order and the tone "..." is the zero in the tone order, so the second search code is transformed as follows:

26H + 8 ^* 5 + 0 = 4eH, 4eH corresponds to "N" of the ASCII code.

By the above, the retrieval code for reading (outer 1) is set to "KN". FIG. 1 is a system block diagram of an embodiment of a data search device of the present invention. In FIG. 1, reference numeral 11 is an input unit capable of inputting a vocal symbol string of an arbitrary length. A storage unit 12 has a register and a buffer for storage. Of which, R register 12
Reference numeral 1 is a register that stores a search code to be compared when searching. The H register 122 stores the upper limit value for the binary search. The P register 123 stores a lower limit value when performing a binary search. The B register 124 stores the size of the limited range when the binary search and the corresponding word are detected. The Q register 125 is a register that stores a code value to be compared when performing a binary search. Reference numeral 13 denotes a search code processing unit that converts the input utterance symbol string into a search code for search according to the constituent features of Chinese utterance. An index storage unit 15 stores a main index table and a sub index table for retrieval. As shown in FIG. 8, the structure of the index storage unit 15 stores a code that can be the first search code in the main index table. Each item in the primary index table points to a specific block in the secondary index table. All valid second search codes relating to the first search code are stored in the corresponding sub-index table block. The size of each block in the secondary index table is determined by the difference between the pointers of the adjacent two primary indexes. By searching the main index table and the sub index table using the second search code, the corresponding block of the corresponding unit 16 can be obtained.

The correspondence unit 16 is composed of blocks corresponding to the hierarchical index table of the index storage unit 15. As shown in FIG. 9, the structure of the correspondence unit 16 is such that each block is divided into a correspondence flag group 161 and a correspondence address group 162. Correspondence flag group 161
The display stores whether or not there is a word corresponding to the next code (the third search code in this embodiment) of the search code used in the index storage unit 15. As shown in FIG. 5, there are 88 third search codes, and if each search code is represented by 1 bit, 11 bytes are required. Therefore, in the present embodiment, the corresponding flag group 161 has a size of 11 bytes. Each search code corresponds to the corresponding bit of the corresponding flag group 161 according to the ASCII order as shown in FIG. If the corresponding bit is set to "1", it means that the corresponding word of the corresponding search code exists. If the corresponding bit is "0", it means that the corresponding search code has no corresponding word. As shown in FIG. 9, the corresponding address group 162 stores the index value of the bit set to "1" of the corresponding flag group 161, that is, the storage start address of the word corresponding to the relevant search code in the dictionary 18. ing. In this embodiment, this corresponding address is stored in 2 bytes.

As shown in FIG. 9, the dictionary 18 stores index codes, corresponding words and related information. Since the search code of the first character is already stored in the index storage unit 15, the dictionary 18
The search code starting from the second character is stored in. In this embodiment, the usage frequency of the corresponding word is stored in 1 byte as the related information.

The search unit 14 refers to the index table of the index storage unit 15 with the previous two search codes obtained by the search code processing unit 13 and applies the binary search to the corresponding two search codes from the corresponding unit 16. Take out the corresponding block. Reference numeral 17 is a detection means capable of detecting the corresponding word from the dictionary 18 in the minimum search space based on the search value of the corresponding flag of the corresponding block and the corresponding address obtained by the search unit 14. The output unit 18 outputs the word detected by the detection unit 17 and the correspondence information.

With respect to the embodiment of the present invention configured as described above, the search operation of the data search apparatus in the embodiment of the present invention will be described with reference to the processing flows of FIGS. First, the utterance symbol string input from the input unit 11 in S1 and S2 is stored in the B buffer 127, and then in B3 in S3.
The voicing symbol string stored in the buffer 127 is converted into a search code as in the process of FIG. 6 and then stored in the A buffer 126.

Next, the processing of the search unit 14 starts. The lower limit value is set to "0" and the upper limit value is set to the number of items in the main index table (55 in this embodiment) by S4, and the P register is set for each.
123 and the H register 122. Next, A buffer 12
The first search code stored in 6 is read into the R register 121. In the operation from S5 to S12, the P register 123
Within the range of the lower limit value of H and the upper limit value of the H register 122, the first search code stored in the R register 121 is used to perform a binary search on the main index table and the obtained sub index is searched. Is set in the P register 123, the position before the block next to the block of the sub index search start position is set in the H register 122, and the second search code is read from the A buffer 126 to read the R register. Store in 121. Then, the operations from S13 to S19 are performed, and the second search code stored in the R register 121 is used to perform a binary search for the sub index table within the limited value range of the P register 123 and the H register 122. The obtained position of the corresponding block pointed to by the corresponding unit 16 is stored in the Q register 125.

Then, the processing of the detecting means from S20 to S25 is performed. First, the third search code is read from the A buffer 126 and stored in the R register 121. Next, the corresponding block pointed to the corresponding portion 16 according to the position stored in the Q register 125 is read into the B buffer 127. In S22, R
By the third search code stored in the register 121,
It is checked whether there is a corresponding word by comparing with the setting of the corresponding flag group 161 of the corresponding block. When there is no corresponding word, the output unit 19 displays a message that the word cannot be found. If there is a corresponding word, in S23 S22
By comparing the corresponding address value of the corresponding address group 162 with P
Store it in register 123 and store the next address value in H register 1.
22 and the difference between the P register 123 and the H register 122 (that is, the minimum search space) in the B register 124. In S24, the data stored in the dictionary 18 from the search start address value stored in the P register 123 to the value stored in the B register 124 is read into the B buffer 127. Then, the previous two search codes in the A buffer 126 are deleted. In S25, the minimum search space stored in the B register 124 is limited by the remaining search code in the A buffer 126, and the contents stored in the B buffer 127 are sequentially compared to find a matching corresponding word and Output corresponding information 19
To output.

Referring to FIGS. 2 and 3 below, a Chinese utterance symbol string

[0029]

[Outside 5]

As an example, the operation of the present invention will be described in detail. For better explanation, R register 121, H register 122, P register 123, B register 124, Q register 1
The values of 25 are r, h, p, b and q, respectively.

When the corresponding vocalization symbol string is input, the input unit 11
The utterance symbol string input by is stored in the B buffer 127, and then converted by the search code processing unit 13 into a search code for search as shown in FIGS. The converted result is “1 (%. A & C + wV1G2 #). This search code is stored in the A buffer 126. Next, p is zero and h is h.
Is set to the total number of items in the main index table (in this embodiment, there are 55 sets of all combinations of valid Chinese and Chinese vowels collected in the main index table). And A
In order to write the first search code '1' stored in the buffer 126 into the R register 121 and perform a binary search, h,
The intermediate value of p (27 in this embodiment) is stored in the B register 124. Then, as shown in FIG. 8, the representative value'E 'of the intermediate values stored in the B register 124 is set to the Q register 125.
Fill in. Next, according to the ASCII character code order, q
With r. If r is greater than q, then replace h with b-1, and if less, h with b + 1
Change to. Until q and r become the same, the operation of performing the binary search from S5 to S10 in FIG. 2 and correcting the value of the register is repeated as described above. According to the above, the comparison result of q and r and the change of the value of each register are as shown in (Table 1) below.

[0032]

[Table 1]

Then, as shown in FIG. 10, the search start position 214 and the search position 242 of the next block pointed to by the sub index table can be obtained. Accordingly, p is 214 and h is 242.
Set -1 = 241.

Next, in order to write the second search code '(' stored in the A buffer 126 into the R register 121 and perform a binary search, the representative value'E 'of the intermediate values stored in the B register 124. To Q register 125. Then q
Until r and r become the same, the operation of performing the binary search from S13 to S17 of FIG. 2 and correcting the value of the register is repeated. According to the above processing, the comparison result of q and r and the change in the value of each register are as shown in (Table 2) below.

[0035]

[Table 2]

Then, as shown in FIG. 10, the corresponding block position 53128 pointed to by the corresponding portion 16 of "1 (", the next block position 5
You can get 4044.

Next, the processing of the detecting means 17 is started. First, A
The third search code '%' stored in the buffer 126 is written in the R register 121, and the content of the extracted corresponding block is written in the B buffer 127. Since the third search code '%' is located at the fifth position in this embodiment as shown in FIG. 5A, it is set to 1 when the fifth bit of the corresponding flag group 161 is searched. , It is displayed that the preceding three search codes have corresponding words. The corresponding address 72 of the fifth bit of the corresponding flag group 161 is stored in the P register 123, the next corresponding address 138 is stored in the H register 122, and the difference 66 between p and h is stored in the B register 124. That is, 53128 is the storage start position of the dictionary of corresponding words, and 66 is the detection range.

Then, the data within the range of b (66) counting from p (53128) of the dictionary 18 is read into the B buffer 127.
Next, the previous two search codes stored in the A buffer 126 are deleted, and the remaining “% .a
& C + wV1G2 # ”is compared with the index code of the corresponding word read in the B buffer 127, and the matching index code, the word“ Break-up Hotpot Question ”and corresponding information is detected,
It is output by the output unit 19.

In the present embodiment, the third search code is used to perform a search of three layers on the principle of branch and bound.
The search speed can be improved as the search range is successively reduced. Moreover, since the search range is shortened, not only the search time is shortened, but also an error such as an input that does not match the pronunciation rule or an incorrect input can be immediately reacted in the search of each layer. Also, the search file organization structure in which a set of valid first search codes is stored in the main index table, a set of valid second search codes is stored in the sub index table, and a third search code is stored in the corresponding flag is effective for searching. Can be done. If the extremely small two previous search codes are stored in the main memory, the storage section can be saved. It is very practical for data retrieval such as Chinese.

As described above, according to the data search apparatus of the present invention, the branch limit (branch ann
It is possible to improve the speed of the search as the search range is successively reduced by performing a search in three layers based on the principle of d bound). In addition, it is possible to immediately react to reading an erroneous input by searching each layer. Also, the search file organization structure in which a set of valid first search codes is stored in the main index table, a set of second search codes is stored in the sub index table, and a third search code is stored at the beginning of the block of the dictionary is effective for the search. Can be done. Moreover, by storing the previous two search codes in the main memory, the branch search can be efficiently performed by using only a very small and fixed storage section. Since the data storage space and the search speed are simultaneously taken into consideration when searching data such as Chinese, it is very practical.

The present invention is not limited to the above-mentioned embodiments, but can be carried out by appropriately modifying it without departing from the scope of the invention. For example,
The symbol to be input is not limited to the vocal symbol, and a simple Kuramo symbol may be used. If the main and sub index tables of the index storage unit and the dictionary are configured with the simple Kurasu code, it can be performed. Also,
The present invention can easily perform a search for Japanese dictionary data.

[0042]

As described above, according to the data search apparatus of the present invention, since the corresponding flag and the corresponding address of the corresponding portion are provided, the storage area can be saved, the search range can be minimized, and the search speed can be reduced. Can be improved. Very practical.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a data search device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing the processing steps of the embodiment.

FIG. 3 is a flowchart showing the processing steps of the embodiment.

FIG. 4 (a) is a block diagram showing a configuration of a conventional data search device, and FIG. 4 (b) is an explanatory diagram illustrating a conventional dictionary structure.

FIG. 5 is an explanatory diagram showing the coding of a kind of vocalization symbol in Chinese according to the present invention.

FIG. 6 is an explanatory diagram showing the code order of Chinese vocalization symbols.

FIG. 7 is a flowchart showing the operation of the search code processing unit of the present invention.

FIG. 8 is an explanatory diagram for explaining a configuration of an index storage unit of the present invention.

FIG. 9 is an explanatory diagram illustrating a configuration of a corresponding unit and a dictionary according to the embodiment.

FIG. 10 is an explanatory diagram illustrating a search method for an index storage unit for detecting a corresponding block of a corresponding unit of the embodiment.

[Explanation of symbols]

 11 Input section 12 Storage section 13 Search code processing section 14 Search section 15 Index storage section 16 Corresponding section 17 Detection means 18 Dictionary 19 Output section 121 R register 122H register 123P register 124B register 125Q register 126A buffer 127B buffer 161 Corresponding flag group 162 Corresponding address group

Front Page Continuation (72) Inventor Lin Chi Shen Taiwan Taipai Si Tha An Chi 10628 Ren Ile San Touan 136 Hao Rosun Sha Tien Chi Chi Shu Kai Far Koo Feng Yu Xieng Kong Su

Claims (2)

[Claims] 1. A dictionary for storing an index code and corresponding data and a plurality of blocks, each block comprising a plurality of bits, each bit corresponding to a search code by a flag value of 0 or 1. Corresponding flag group indicating whether or not there is data to be stored and the above flag value is 1, a corresponding portion including a corresponding address group storing the index code in the dictionary and the storage location of the corresponding data, and the input search code By comparing the search part for detecting the corresponding block of the corresponding part with a part of the input search code by the front part of the above, the corresponding flag and the corresponding address of the detected corresponding part are extracted and the corresponding data of the search code is dictionary. A data search device comprising: 2. A dictionary for storing index codes and corresponding words, and a plurality of blocks, each block consisting of a plurality of bits, each bit corresponding to a search code by a flag value of 0 or 1. A corresponding flag group indicating whether or not there is a corresponding word, and the flag value is 1, a corresponding portion including an index code in the dictionary and a corresponding address group storing the storage location of the corresponding word, and the input search code By comparing the search unit that detects the corresponding block of the corresponding portion with the code of the first character of the above and the code of the second character of the input search code, the corresponding flag and the corresponding address of the detected corresponding portion are extracted. A Chinese dictionary data search device comprising: a detection unit that detects a corresponding word of a search code from a dictionary.