JPH0475551B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a content search device that outputs data that satisfies one or more keywords and search conditions by providing them.

The content retrieval device described above is an important component of a database system. A database provides a service for extracting data related to a given keyword, and has a storage capacity of several gigabytes or more, so it is recorded on a large-capacity storage device such as a magnetic disk or magnetic tape. It is very difficult to search such a large capacity storage device in a short time. Therefore, input a keyword and search condition, find the storage address on the mass storage device of the data that meets the keyword, and then use this address to match the search keyword from the mass storage device and retrieve the data that meets the search condition. What you ask for is being done. At this time, a relationship table between keywords and storage addresses called an index table is used. In the index table, content search processing is performed to obtain a storage address from a search keyword.

In order to realize a content retrieval device that performs such content retrieval processing, it is desirable to use associative memory elements.
There is nothing practical in terms of storage capacity and price. Therefore, it is necessary to perform content retrieval processing using a normal storage element. The content retrieval device in this case reads the contents of all addresses in a storage device using ordinary storage elements, compares keywords at each address, and finds data at matching addresses.

In the search operation of such a content search device, search conditions are generally given that represent a plurality of keywords and their relationships in logical formulas. The execution time of a logical formula by a microprocessor or the like is several tens of times longer than the memory access time, so if search conditions are determined for each address, the search time will be enormous. For this reason, a method is used in which the search is repeated for each search keyword. For example, in a search using the product of three search keywords A, B, and C, first search with search keyword A, set an identification marker at the matching address, then search with search keyword B,
Reset identification markers for mismatched addresses. Similarly, by resetting the identification marker that does not match the search keyword C, and finally outputting the data of the address where the identification marker is set,
A search is performed using search conditions A, B, and C.

Such conventional content search devices have the disadvantage that the content of all addresses must be read multiple times in order to search for a plurality of keywords, and the search time is too long.

SUMMARY OF THE INVENTION An object of the present invention is to provide a content search device that easily solves the above-mentioned conventional drawbacks and can process searches using a plurality of keywords in a short time.

According to the present invention, a record storage means for storing a plurality of records each including N (N is a positive integer) keyword sections and a data section; and an address scanning means for sequentially outputting the contents of the record storage means;
A search keyword register that stores a plurality of externally given search keywords, and the contents of this search keyword register are compared with the contents of the plurality of keyword sections in the record output by the record storage means. Converting the search conditions given by the logical expressions of the N search keywords into a ^2N -bit search condition code indicating one of the ^2N unit search conditions that can be logically generated. A bit position in the ^2N- bit search condition code is specified by the output of the search condition encoding means and the keyword comparison means, and the value of the specified bit position is output as the search result based on the search condition. A content retrieval device is obtained, which is characterized in that it includes a switching means for switching.

The present invention will be described in detail below using the drawings.

FIGS. 1a and 1b are explanatory diagrams of the index table. The index table 120 is a large-capacity magnetic disk storage device 110 in which a data base is stored.
It shows the correspondence between the address of each record in , and the keyword 130 assigned to that record. This index table 120 is designed to quickly find the storage address of a record containing a search keyword.
Must be stored in fast memory. for example,
The search keyword is age 30, occupation is technology,
When search conditions such as age, job type, and name (the "." indicates a logical product) are given, such as the name Sato, 110 is output as the address.

FIG. 2 is a block diagram showing an embodiment of the content search device according to the present invention.

As shown in Figure 1, age, occupation, name, etc.
Records consisting of keywords and data such as addresses are stored in the first keyword section 211, second keyword section 212, third keyword section 213, and data section 214 of the record memory 210 via the data bus 201 at the time of registration. Stored. A large number of registered records are stored in the record memory 210. Address scanning circuit 2
20 is specified by the address of the record memory 210.
Select stored records.

During a search, three search keywords A, B, and C are sent via the data bus 201 to A, B, and C, respectively.
Stored in C registers 231, 232, 233,
Also, a search condition 241 is supplied to a search condition encoder 240 . Search condition 241 is search keyword A,
It is expressed as the logical product or logical sum of B and C. For example, the search condition that search keyword A is equal to search keyword C or equal to search keyword B is expressed as A.C+B. When such a search condition A・C+B is input, the search condition encoder 240 converts the search condition into the prime additive standard form, that is, A・C→A・B・C+A・・C B→A・B・C+A・B・+・B・C+
Transforms into A.B. Here, eight unit search conditions A・B・C, A・B・, A・・C, A・,
C,・B・C,・B・,,・・C,
A... as a decimal number 7, 6, 5, 4, 3, 2,
1, 0 or binary code 111, 110, 101, 100,
If it corresponds to 011, 010, 001, 000, the unit corresponding to the binary code indicating the position depends on which bit (decimal number) of the 8-bit parallel search condition code is "1". You can display whether the search condition is included or not. For example, the search condition for A.B. is 2.
Since it includes a unit search condition corresponding to base number 110 or decimal number 6, it is converted to a search condition code of the bit pattern 〓01000000〓 where the right end is 0 and the 6th is 1. In addition, the search conditions for A/C+B above are unit search conditions corresponding to 111, 110, 101, 011, 010 in binary, that is, 7, 6, 5, 3, 2 in decimal.
Since it includes ABC, AB, A, C, BC, and B, it is converted to the search condition code 〓11101100〓.
Generally, a search condition regarding n keywords is converted into a 2n ^- bit search condition code. After generating the search condition code described above, the record memory 21
Start searching for the contents of 0.

The search keywords A, B, and C stored in the A, B, and C registers 231, 232, and 233 are compared with the first, second, and third keyword sections 211, 212, and 213 in the record memory 240, respectively. This is performed by first, second, and third comparators 251, 252, and 253. The 3 bits of each comparison result are supplied to a switching circuit 260, which selects 1 bit within the 8-bit search condition code. This selected one
The bits will indicate the search results. In the search condition shown above, the search condition code 〓11101100〓 stored in the search condition encoder 240 is input to the switching circuit 260 in 8 bits in parallel, so the comparison result of the search keywords A, B, and C in the order is〓010〓、〓
In any case of 011〓, 〓101〓, 〓110〓, 〓111〓 (1 indicates a match), it is determined that the search conditions A, C+B including the corresponding unit search conditions are satisfied. Search result 1 is switching circuit 26
Output from 0. When the search result 〓1〓 occurs, the bus buffer 270 becomes active, and the data of the matched record is outputted from the data section 214 in the record memory 210 onto the data bus 201.

Since the search for one record has been completed above, the address of the record memory 210 indicated by the address scanning circuit 220 is increased, and the search for the next record is continued. In this way, when the address scanning circuit 220 finishes scanning all addresses in the record memory 210, the search for all records is completed.

Even when searching using a plurality of keywords in this manner, the search is completed by scanning all addresses in the record memory 210 once. Furthermore, the operation of determining the search condition for each record can be performed at high speed because it is sufficient to simply select one bit in the 8-bit search condition code using the 3-bit search result using the switching circuit 260.

FIG. 3 shows an example of the configuration of a search condition encoder 240 used in the content search device according to the present invention. This search condition encoder 240 converts a search condition expressed by a logical formula into a search condition code. This conversion is performed for each product term of the search conditions.

First, the first product term of the search condition is encoded and input into the register 310.

In the previous search condition A・C+B, the first product term A・
〓101〓 indicating C is stored in the register 310,
The inverted output 〓010〓 is the OR gate 341, 34
2,343. Thereafter, ²ⁿ clock signals 302 (n is the number of keywords), ie, 8 clock signals 302 are supplied. A 3-bit counter 320 is incremented from 0 by a clock signal 302, and the contents of an 8-bit shift register 330 are cycled through. The negative pulse signal clock signal 302 is inverted by an inverter 360 and provided to the clock input of the shift register 330. Therefore, when one clock signal 302 is applied, the contents of shift register 330 are first shifted by one bit, and then counter 320 is incremented. When the output of the register 310 is 〓010〓, the output of the AND gate 350 means that the contents of the counter 320 are 〓101〓 or 〓111〓.
When 〓, it becomes 〓1〓. That is, the 6th and 8th
The shift register 33 receives the clock signal 302.
〓1〓 is written to 0. Therefore, the shift register 3 after being supplied with eight clock signals 302
The content of 30 becomes from 〓00000000〓 to 〓10100000〓.

Next, 〓010〓 indicating the second product term B of the search conditions A and C+B is input to the register 330, and 〓101〓 is supplied to OR gates 341, 342, and 343, respectively. When eight clock signals 302 are supplied in the same way as for the first product term, the output of the AND gate 350 indicates that the contents of the counter 320 are 〓010〓, 〓011
When 〓, 〓110〓, 〓111〓, it becomes 〓1〓.

Therefore, the contents of the shift register 330 are changed from 〓10100000〓 to 〓11101100〓 by adding the output of the AND gate 350 at the OR gate 344. Since search conditions A and C+B are composed of two product terms, the generation of the search condition code is now completed. If more product terms are included, similar operations are repeated for each product term. When the operations for all product terms are completed, a search condition code 331 is output from the shift register 330. The search condition code in the example of search conditions A and C+B is 〓11101100〓.

This search condition code 331 is supplied to the switching circuit 260 in FIG.

Note that the shift register 330 is a counter 320.
It is also possible to replace the contents of with a RAM that uses the address input. Further, this RAM can also be used as a switching circuit 260.

This search condition code generator is applied to search conditions expressed by AND and OR. To process a search condition including inversion, that is, A,, an inversion register indicating the inversion part in the product term and an exclusive OR gate are provided between the counter 320 and the OR gates 341, 342, 343, and the contents of the inversion register are processed. This is possible by inverting the contents of the counter 320 based on , and supplying it to the OR gates 341, 342, and 343.

As described above, even in a search using a plurality of keywords, the content search device of the present invention can complete the search by scanning all addresses in the record memory 210 once. With traditional identification markers,
In a content search device that determines search conditions, it is necessary to scan all addresses in the record memory 210 as many times as the number of search keywords in order to find a record that satisfies the search conditions. Therefore, when the number of search keywords is four, the search time of the content search device of the present invention is reduced to one-fourth of the conventional search time.

Note that the search condition encoder shown in FIG. 3 is an example, and the search condition code can also be generated by software using a microprocessor or the like. Further, it is possible to use a larger number of search keywords, and the present invention can also be applied when performing a content search using keywords that can be generated by hashing keywords instead of keywords. Therefore, the above description is not intended to limit the scope of the claimed invention.

[Brief explanation of drawings]

1A and 1B are block diagrams of a general database, FIG. 2 is a block diagram of a content retrieval device according to an embodiment of the present invention, and FIG. 3 is a block diagram of a search condition encoder. 110...Magnetic disk, 120...Index
Table, 130...keyword, 210...record memory, 211...first keyword section, 212...
Second keyword part, 213...Third keyword part,
214...data section, 220...address scanning circuit,
231...A register, 232...B register, 23
3...C register, 240...Search condition encoder, 25
1...first comparator, 252...second comparator, 253...
Third comparator, 260... switching circuit, 270... bus buffer, 310... register, 320... counter, 330... shift register, 341, 342,
343, 344...OR gate, 350...AND gate, 360...Inverter.

Claims (1)

[Claims] 1 record storage means for storing a plurality of records consisting of N (N is a positive integer) keyword parts and data parts; address scanning means for sequentially outputting the contents of this record storage means; a search keyword register for storing search keywords; keyword comparison means for comparing the contents of the search keyword register with the contents of a plurality of keyword parts in the record output by the record storage means; A search condition register of N bits or more that stores search conditions given by logical expressions of N search keywords, a counter of N bits or more that increases in synchronization with a clock signal, and an output of the N bits of this counter and the search. N first OR circuits each receiving the N-bit inverted output of the condition register, an AND circuit receiving the output of the first OR circuit, and a search condition of 2 ^N bits or more. a shift register that converts the code into a code; a second OR circuit that receives the output of the AND circuit and the output of the shift register and whose output is connected to the shift register ^; 1. A content retrieval device comprising a switching means for specifying a bit position in a retrieval condition code and outputting the value of the specified bit position as a search result based on the retrieval condition.