JP2007249369A - Key search device, key search method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To speedily search a leading search key among search keys having a search object key value regardless of the number of search keys having the same key value included in a search object key column. <P>SOLUTION: When a search instruction including a search object key value is inputted by a user, a search key having the search object key value is searched by a binary search method from an array type index A by a key search part SK. Afterwards, one of a linear search method or the binary search method is selected based on a threshold to be determined according to the rate of a duplicate level calculated by a duplicate level determination part D to the mean costs of the linear search method and the binary search method by a boundary search part 11. At that time, when the duplicate level is less than a threshold, the linear search method is selected, and when the duplicate level is the threshold or more, the binary search method is selected. Afterwards, the leading search key of the search keys having search object key values is searched by the selected search method. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  According to the present invention, a key string including a search key in which a plurality of search keys are arranged in ascending or descending order and the same key value is used as a search target key string, such as an array type index having search keys that allow duplication. The present invention relates to a key search technique, and more particularly to a key search technique for searching for the first or last search key among search keys having a designated search target key value.

  When searching an array type index or file with a search key that allows duplication, searching for the first search key among the search keys with the key value (search target key value) specified by the user Traditionally done. The first search key having the search target key value is searched so that the subsequent processing can be performed efficiently. For example, in the array type index 101 as shown in FIG. 16, when the user designates “d” as the search target key value, the first search key among the search keys k3 to k10 having the key value “d”. Is retrieved, the user can refer to all the records pointed to by the keys k3 to k10 having the key value “d” only by operating the button requesting the next record. Improves.

  By the way, the following techniques are known as this kind of conventional techniques (for example, refer to Patent Document 1). The conventional information search device described in Patent Document 1 uses a log file whose time is a search key as a search target. When a user inputs a time to be searched, first, a bisection is performed. A record including the above time as a search key is searched by a search method. Thereafter, the record searched by the binary search method is used as a search start record, the log file is searched in the descending order by the linear search method, and the first record among the records including the time designated by the user is searched.

JP 2001-167094 A

  In the conventional technique described above, first, one of the records including the time instructed by the user is searched by the binary search method, and then the search record is set as the search start record in the descending direction by the linear search method. Therefore, the search process can be speeded up as compared with the case where all searches are performed by the linear search method. However, in the above-described conventional technique, after searching for a record including the time designated by the user by the binary search method, the first record is always searched by the linear search method, so the same key value is included. When there are a large number of records, there is a problem that the search process cannot be sufficiently accelerated.

  Although it is conceivable to search for the first record by the binary search method, in such a case, there is a problem that if there are few records including the same key value, it becomes inefficient.

(Object of invention)
Accordingly, an object of the present invention is to quickly search the first or last search key of search keys having a search target key value regardless of the number of search keys having the same key value included in the search target key column. To be able to search.

The first key retrieval apparatus according to the present invention is:
A plurality of search keys arranged in ascending or descending order, and a key having a designated search target key value and different from the search target key value out of a search target key string including a search key having the same key value A key search device for searching a search key that is a boundary with a search key having a value,
A key search unit for searching a search key having the search target key value from the search target key string by a binary search method;
Based on the search key duplication degree of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value And a boundary search unit for searching.

A second key search device according to the present invention is the first key search device,
The boundary search unit
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search unit for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search unit for searching by a search method;
The first duplicate key boundary search unit or the second duplicate key boundary search unit is based on the degree of duplication of the search key of the search target key column and the ratio of the average cost of the linear search method and the binary search method. A search method determination unit to be operated is provided.

A third key search device according to the present invention is the first or second key search device,
The degree of duplication increases as the number of search keys having the same key value increases.

A fourth key search device according to the present invention is the third key search device,
For each key value in the search target key column, a duplication degree management table in which the number of search keys having the key value is registered;
When a search key is added to the search target key column and when a search key is deleted from the search target key column, the duplication degree management table is updated, based on the contents of the updated duplication degree management table And a duplication degree determining unit that obtains an average value of the number of keys for each key value and uses the average value as a duplication degree.

The first key search method according to the present invention is as follows.
A plurality of search keys arranged in ascending or descending order, and a key having a designated search target key value and different from the search target key value out of a search target key string including a search key having the same key value A key search method for searching a search key that is a boundary with a search key having a value,
A key search step of searching a search key having the search target key value from the search target key string by a binary search method;
Based on the search key duplication degree of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value And a boundary search step for searching.

A second key search method according to the present invention is the first key search method,
The boundary search step includes:
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search step for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search step for searching by a search method;
The first duplicate key boundary search step or the second duplicate key boundary search step is performed based on the degree of duplication of the search key of the search target key string and the ratio of the average cost of the linear search method and the binary search method. And a search method determination step to be activated.

A third key retrieval method according to the present invention is the first or second key retrieval method,
The degree of duplication increases as the number of search keys having the same key value increases.

A fourth key search method according to the present invention is the third key search method,
When a search key is added to the search target key column and when a search key is deleted from the search target key column, for each key value in the search target key column, the key of the search key having that key value The duplication degree management table in which the number is registered is updated, the average value of the number of keys for each key value is obtained based on the contents of the updated duplication degree management table, and the duplication degree is determined by using the average value as the duplication degree Including steps.

The first program according to the present invention is:
A plurality of search keys arranged in ascending or descending order, and a search target key string including a search key having the same key value, the specified search target key value and the search target key value; Is a program for functioning as a key search device that searches for a search key that is a boundary with a search key having a different key value,
The computer,
A key search unit for searching a search key having the search target key value from the search target key string by a binary search method;
Based on the search key duplication degree of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value It functions as a boundary search unit for searching.

The second program according to the present invention is the first program,
The boundary search unit
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search unit for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search unit for searching by a search method;
The first duplicate key boundary search unit or the second duplicate key boundary search unit is based on the degree of duplication of the search key of the search target key column and the ratio of the average cost of the linear search method and the binary search method. A search method determination unit to be operated is provided.

The third program according to the present invention is the first or second program,
The degree of duplication increases as the number of search keys having the same key value increases.

A fourth program according to the present invention is the third program,
The computer,
When a search key is added to the search target key column and when a search key is deleted from the search target key column, for each key value in the search target key column, the key of the search key having that key value The duplication degree management table in which the number is registered is updated, the average value of the number of keys for each key value is obtained based on the contents of the updated duplication degree management table, and the duplication degree is determined by using the average value as the duplication degree Function as a part.

[Action]
When a search key is added to the search target key string and when a search key is deleted from the search target key string, the duplication degree determination unit recalculates the duplication degree of the search target key string. The degree of overlap increases as the number of search keys having the same key value increases. For example, an average value of the number of keys for each key value can be adopted.

  When the search target key value is designated by the user, first, the key search unit searches the search key having the search target key value from the search target key string by the binary search method. After that, the boundary search unit selects one of the linear search method and the binary search method based on the redundancy calculated by the redundancy determination unit and the ratio of the average cost of the linear search method and the binary search method. A search key searched by the key search unit by the selected search method as a search start key, and a boundary between the search key having the search target key value and a key value different from the search target key value Search for a search key.

  According to the present invention, regardless of the number of search keys having the same key value included in the search target key column, the search key at the head or the end among the search keys having the search target key value is searched at high speed. The effect that it is possible can be obtained. The reason is that one of the linear search method and the binary search method is selected based on the redundancy of the search key in the search target key column and the ratio of the average cost of the linear search method and the binary search method. By using the search method, the search key searched by the key search unit is used as a search start key, and a search key that is a boundary between a search key having a search target key value and a key value different from the search target key value is searched. This is because a boundary search unit is provided.

  Next, the best mode for carrying out the invention will be described in detail with reference to the drawings.

[Principle of the present invention]
First, before describing embodiments of the present invention, the principle of the present invention will be described.

When searching for one key from N pieces of data in an array type index, if the binary search method is used for an index that does not allow duplication of keys, the search ends with a maximum of (log ₂ N + 1) comparisons. In the case of an array type index that allows duplication of keys, in addition to this process, a process of “detecting the leading position of the same search key group having the same key value” is required.

  There are generally two methods for this processing.

  The first method is a method of detecting the head position of the same search key group by comparing the array type indexes in reverse order by the linear search method. The second method is a method of detecting the head position of the same search key group while narrowing down the search range by the binary search method.

The maximum number of comparisons when searching for one key from N data is N times for the linear search method and (log ₂ N + 1) times for the binary search method. FIG. 13 shows the maximum number of comparisons of the linear search method and the binary search method when the number of data N is 1 to 10.

  The processing contents and the average cost (average number of processing steps at the assembler level) of the linear search method and the binary search method when the head position of the same search key group can be detected by n comparison processes are shown in FIG. 14 is obtained. Note that SL4 and the like in the processing contents of the linear search method indicate step numbers in the flowchart of FIG. 11, and SB2 and the like in the processing contents of the binary search method indicate step numbers in the flowchart of FIG. As shown in FIG. 14, the average cost of the binary search method is 1.5 times the average cost of the linear search method.

  In consideration of the maximum number of searches of the linear search method and the binary search method shown in FIG. 13 and the average cost of the linear search method and the binary search method shown in FIG. FIG. 15 shows the calculated costs of the linear search method and the binary search method.

  As can be seen with reference to FIG. 15, the linear search method is more efficient when the number of data N is small, and the binary search method is more efficient when the number of data is large. The boundary value is “6” in this example.

  Therefore, it is possible to increase the speed by using a linear search method when the number of records with duplicate search keys is smaller than the expected value “6” and using a binary search method when the search key is larger than “6”. Actually, both methods have additional processing, so when the expected value is “6”, the total cost is “36 + 6α” in the linear search method and “36+” in the binary search method. 4α ”(α is an additional processing cost for one comparison), so the binary search method is used.

[Description of Configuration of Embodiment]
Next, the configuration of the embodiment of the key search device according to the present invention will be described.

  As shown in FIG. 1, a storage device 2 such as a hard disk and an input device 3 such as a keyboard are connected to the key search device 1 of the present embodiment.

  In the storage device 2, an array type index A to be searched by the key search device 1 is registered. The array type index A allows search keys to be duplicated, and a plurality of search keys are arranged in ascending order. In the example of FIG. 1, the key values of the search keys k3 to k10 are all “d”, and the key values of the search keys k11 and k12 are “e”. Although not shown in FIG. 1, the storage device 2 also stores records (data) pointed to by the search keys k1, k2,. In the present embodiment, an example in which search keys are arranged in ascending order as array type index A will be described as an example. However, the present invention is applied even if search keys are arranged in descending order. be able to.

  The key search device 1 includes a key search unit SK, a search method determination unit SM, a boundary search unit 11 including a first duplicate key boundary search unit SL, and a second duplicate key boundary search unit SB, and a redundancy degree determination. It consists of a part D and a redundancy management table DT.

  The key search unit SK has a function of searching the array type index A by a binary search method when a search instruction is input from the user via the input device 3. FIG. 2 shows an outline of processing performed by the key search unit SK. The search instruction includes a key value (search target key value) to be searched.

  In the duplication degree management table DT, for each key value b, c, d,... In the array type index A, the number of search keys having the key value (number of duplicate keys) and the average value of the number of duplicate keys are stored. It is registered. In this embodiment, as shown in FIG. 3, a record having items “key value” and “number of duplicate keys” is registered, and an average value of the number of duplicate keys is registered as the degree of duplication. The example in the figure shows that the number of duplicate keys of search keys having key values b, c, d,... Is 2, 1, 8,. 5.8.

  The duplication degree determination unit D has a function of referring to and updating the duplication degree management table DT every time the array type index A is updated, and recalculating the duplication degree (average value of the number of duplicate keys). FIG. 3 shows an overview of the processing performed by the redundancy determining unit D. If the array type index A is not updated, the duplication degree determination unit D and the duplication degree management table DT are not necessary. Instead, a memory in which the duplication degree of the array type index A is registered is used. Prepare it.

  The boundary detection unit 11 selects one of the linear search method and the binary search method based on the duplication degree determined by the duplication degree determination unit D and a predetermined threshold Th, and the key search is performed by the selected search method. The search key searched in the part SK is used as a search start key, and the first search key among the keys having the search target key value is searched. Note that the threshold value Th is a value determined in advance based on a ratio of average costs of the linear search method and the binary search method.

  After the processing of the key search unit SK is completed, the search method determination unit SM in the boundary search unit 11 performs the first operation based on the degree of duplication calculated by the duplication degree determination unit D and a predetermined threshold Th. , Has a function of determining which one of the second duplicate key boundary search units SL and SB is to be operated.

  The first duplicate key boundary search unit SL has a function of detecting the head position of the same search key group by the linear search method from the key position detected by the key search unit SK. FIG. 4 shows an outline of processing performed by the first duplicate key boundary search unit SL.

  The second duplicate key boundary search unit SB has a function of detecting the head position of the same search key group by the binary search method from the key position detected by the key search unit SK. FIG. 5 shows an outline of processing performed by the second duplicate key boundary search unit SB.

  In the present embodiment, the first and second duplicate key boundary search units SL and SB detect the start position of the same search key group, but may detect the end position. The key search device 1 can be realized by a computer. When the key search device 1 is realized by a computer, for example, the following is performed. A disk, a semiconductor memory, and other recording media on which a program for causing the computer to function as the key search device 1 is recorded are prepared, and the program is read by the computer. The computer controls its own operation according to the read program, so that a key search unit SK, a search method determination unit SM, a first duplicate key boundary search unit SL, a second duplicate key boundary search unit are provided on the computer. The SB and the redundancy determining unit D are realized.

[Description of Operation of Embodiment]
Next, the operation of the present embodiment will be described in detail with reference to the drawings.

  First, the redundancy determination process performed by the redundancy determination unit D when the array type index A is updated will be described with reference to the flowchart of FIG.

  When the process of updating (including addition / deletion) of the element of the array type index A is performed (step D0 in FIG. 6), the duplication degree determination unit D performs an event of adding or deleting a search key for the array type index A. Is determined (step D1).

  If it is determined in step D1 that an additional event has occurred, the redundancy management table addition process DA is executed, and the process proceeds to step D4, which is the end of the event loop (step D2 → D4). The redundancy management table addition process DA will be described in detail later with reference to the flowchart of FIG.

  If it is determined in step D1 that a deletion event has occurred, the redundancy management table subtraction process DS is executed, and the process proceeds to step D4, which is the end of the event loop (step D3 → D4). The redundancy management table subtraction process DS will be described in detail later with reference to the flowchart of FIG.

  If it is determined in step D1 that neither an additional event nor a deletion event has occurred, the process proceeds to step D4 which is the end of the event loop (step D1 → D4).

  In step D4, when it is determined that there are remaining events (when the determination result is yes), the process returns to step D1. On the other hand, when it is determined in step D4 that there are no remaining events (when the determination result is no), the process proceeds to step D5.

  In step D5, the duplication degree management table DT is referred to, and an average value of the number of duplicate keys registered in association with each key value is calculated as the duplication degree. Thereby, the multiplicity determination unit D ends the process.

  Next, the redundancy management table addition process DA performed in step D2 of FIG. 6 will be described in detail with reference to the flowchart of FIG.

  First, the duplication degree management table DT is searched by the binary search method to find a record having the same key value as the key value k of the target search key (added search key) (step DA1).

  If the corresponding record does not exist (step DA2 is no), a record having the key value “k” and the number of duplicate keys “0” is added to the duplicate management table DT, and further, the duplicate management table for the key is added. After the records in DT are sorted in ascending order (step DA3), the process proceeds to step DA4. On the other hand, if the corresponding record exists (step DA2 is yes), the process immediately proceeds to step DA4.

  In step DA4, the number of duplicate keys of the record having the key value k in the duplicate management table DT is incremented by one. The above is the details of the redundancy management table addition process DA performed in step D2 of FIG.

  Next, the redundancy management table subtraction process DS performed in step D3 of FIG. 6 will be described in detail with reference to the flowchart of FIG.

  First, the duplication degree management table DT is searched by the binary search method to find a record having the same key value as the key value k of the target search key (deleted search key) (step DS1). Next, the number of duplicate keys in the found record is decremented by 1 (step DS2).

  If the number of duplicate keys in the record becomes “0” (Yes in step DS3), the record is deleted from the duplicate management table DT, and then the records in the duplicate management table DT are arranged in ascending order by the key. Sorting is performed (step DS4), and the duplication degree management table subtraction process DS is terminated. On the other hand, if the number of duplicate keys in the record does not become “0” (step DS3 is no), the duplicate management table subtraction process DS is immediately terminated. The above is the details of the redundancy management table subtraction process performed in step D3 of FIG.

  Next, the operation when searching for the array type index A will be described.

  When searching for the array type index A, the user inputs a search instruction from the input device 3. This search instruction includes the key value k of the search key to be searched.

  When a search instruction is input, first, key search processing shown in the flowchart of FIG. 9 is performed by the key search unit SK.

  First, the key search unit SK sets the upper limit position (array suffix) of the array type index A to the upper limit pointer u and the lower limit position to the lower limit pointer l (step SK1).

  Next, an intermediate position between the upper limit position set in the upper limit pointer u and the lower limit position set in the lower limit pointer l is obtained and set in the search pointer m (step SK2). Here, the intermediate position is {(l + u) +1} / 2. However, when m is a decimal, the decimal part is rounded down.

  If the key value A [m] at the search position indicated by the search pointer m (the mth key value of the array type index A) is equal to the search target key value k (step SK3 is yes), m is detected. The pointer k0 is set to 1 as the search lower limit pointer kl, and after the search method determination unit SM is activated (step SK8), the key search process is terminated. On the other hand, if A [m] ≠ k (step SK3 is no), it is checked whether A [m]> k (step SK4).

  If A [m]> k (step SK4 is yes), the upper limit pointer u is updated by m−1 (step SK5), and the process proceeds to step SK7. On the other hand, if A [m] ≦ K (step SK4 is no), the lower limit pointer l is updated with m + 1 (step SK6), and the process proceeds to step SK7.

  In step SK7, if l> u (determination result is yes), the key search process is terminated as no corresponding data. At this time, a message indicating that there is no corresponding data may be displayed on a display device (not shown). On the other hand, if l ≦ u (step SK7 is no), the process returns to step SK2.

  Next, search method determination processing performed by the search method determination unit SM will be described with reference to the flowchart of FIG.

  When activated by the key search unit SK, the search method determination unit SM first compares the duplication degree d determined by the duplication degree determination unit D with a predetermined threshold value Th (step SM1). Here, as described above, the threshold value Th is determined in advance according to the ratio of the average cost of the linear search method and the average cost of the binary search method, and the costs of the linear search method and the binary search method are the same. In the case shown in FIG. 15, the threshold value Th is “6” as described above.

  If the duplication degree d is less than the threshold Th (step SM1 is yes), the first duplicate key boundary search unit SL that performs the search by the linear search method is activated (step SM2), and the search method determination process is terminated. On the other hand, when the duplication degree d is equal to or greater than the threshold value Th (step SM1 is no), the second duplicate key boundary search unit SB that performs the search by the binary search method is activated (step SM3), and the search method is determined. The process ends.

  Next, the operation of the first duplicate key boundary search unit SL will be described. When the duplicate key boundary search unit SL is activated by the search method determination unit SM, the duplicate key boundary search unit SL performs a duplicate key boundary search process using the linear search method shown in the flowchart of FIG.

  First, the duplicate key boundary search unit SL sets the value set in the detection position pointer k0 by the key search unit SK as the search pointer m and the value set in the search lower limit pointer kl as the lower limit pointer l (step) SL1).

  Thereafter, it is checked whether or not the key value A [m−1] of the element immediately before the search position indicated by the search pointer m matches the search target key value k (step SL2).

  If the key value A [m-1] is not the search target key value k (step SL2 is no), m is the start position of the target key, so m is set as the detection position pointer k0 (step SL5). The duplicate key boundary search process (for linear search) is terminated. On the other hand, when the key value A [m−1] is the search target key value k (step SL2 is yes), the process proceeds to step SL3 to determine the search lower limit.

  In step SL3, it is checked whether or not the search pointer m and the lower limit pointer l are equal. If m = 1 (step SL3 is yes), all the search is performed up to the search lower limit, so m is set as the detection position pointer k0 (step SL5), and the duplicate key boundary search process is terminated. On the other hand, when m is not 1 (step SL3 is no), the search pointer m is decremented (step SL4) and the process returns to step SL2 in order to update the search position.

  Next, the operation of the second duplicate key boundary search unit SB will be described. When the duplicate key boundary search unit SB is activated from the search method determination unit SM, the duplicate key boundary search unit SB performs a duplicate key boundary search process using the binary search method shown in the flowchart of FIG.

  First, the duplicate key boundary search unit SB sets the values set by the key search unit SK to the detection position pointer k0 and the search lower limit pointer kl as the upper limit pointer u and the lower limit pointer l, respectively (step SB1).

  Next, an intermediate position between the upper limit position set in the upper limit pointer u and the lower limit position set in the lower limit pointer l is obtained and set in the search pointer m (step SB2). Here, the intermediate position is {(l + u) +1} / 2. However, when m is a decimal, the decimal part is rounded down.

  Thereafter, it is checked whether or not the value of the search pointer m and the value of the upper limit pointer u are the same (step SB3).

  If m = u (Yes in step SB3), u-l ≦ 1 and the search is substantially completed, so the process proceeds to step SB8. On the other hand, when m = u is not satisfied (step SB3 is no), the process proceeds to step SB4.

  In step SB4, it is checked whether or not the key value A [m] indicated by the search pointer m is equal to the search target key value k.

  If A [m] = k (step SB4 is yes), the process proceeds to step SB6 to determine the start position of the target key. On the other hand, if A [m] ≠ k (step SB4 is no), m + 1 is set to the lower limit pointer l (step SB5), and the process returns to step SB2.

  In step SB6, it is checked whether A [m-1] = k.

  If A [m-1] = k (step SB6 is yes), m-1 is set as the upper limit pointer u (step SB7), and the process returns to step SB2. On the other hand, when A [m−1] ≠ k (step SB6 is no), m is the start position of the target key, so m is set as the detection position pointer k0 (step SB10), and the duplicate key is set. The head search process ends.

  In step SB8, it is checked whether A [l] = k.

  If A [l] = k (step SB8 is yes), l becomes the start position of the target key, so l is set as the detection position pointer k0 (step SB9), and the duplicate key head search process SB is executed. finish. On the other hand, if A [l] ≠ k (step SB8 is no), m is the start position of the target key, so m is set as the search position pointer k0 (step SB10), and duplicate key head search is performed. The process (for binary search) is terminated.

[Effect of the embodiment]
According to the present embodiment, regardless of the number of search keys having the same key value included in the array type index A, the first search key among the search keys having the search target key value is searched at high speed. The effect that it is possible can be obtained. The reason is that one of the linear search method and the binary search method is selected and selected based on the redundancy of the search key of the array type index A and the ratio of the average cost of the linear search method and the binary search method. Because the search method includes a boundary search unit 11 that searches the search key searched by the key search unit SK as a search start key and searches for the first search key among the search keys having the search target key value. is there.

  Further, in the present embodiment, even when a search key is added to the array type index A or when a search key is deleted from the array type index A, the search key among the search keys having the search target key value Thus, it is possible to obtain the effect that the head search key can be searched at high speed. The reason is that when the search key is added to the array type index A and when the search key is deleted from the array type index A, the duplication degree management table DT is updated, and the contents of the updated duplication degree management table DT are added. This is because there is provided a duplication degree determining unit D that calculates an average value of the number of keys for each key value as the duplication degree.

  INDUSTRIAL APPLICABILITY According to the present invention, the present invention can be applied to general search for data strings having overlapping keys including array type indexes.

It is a block diagram which shows the structural example of embodiment of the key search device 1 concerning this invention. It is a figure which shows the outline | summary of the process which the key search part SK performs. It is a figure which shows the example of the content of the duplication degree management table | surface DT, and shows the outline | summary of the process which the duplication degree determination part D performs. It is a figure which shows the outline | summary of the process which 1st duplication key boundary search part SL performs. It is a figure which shows the outline | summary of the process which 2nd duplication key boundary search part SB performs. 6 is a flowchart illustrating an example of processing of a redundancy determining unit D. It is a flowchart which shows the detail of step D2 of the flowchart of FIG. It is a flowchart which shows the detail of step D3 of the flowchart of FIG. It is a flowchart which shows the process example of the key search part SK. It is a flowchart which shows the process example of the search method determination part SM. It is a flowchart which shows the process example of 1st duplication key boundary search part SL. It is a flowchart which shows the process example of 2nd duplication key boundary search part SB. It is a figure which shows the maximum frequency | count of comparison of a linear search method and a binary search method when the number of data is 1-10. It is a figure which shows an example of the average cost of a linear search method and a binary search method. It is a figure which shows the cost of a linear search method and a binary search method when the number of data is 1-10. It is a figure which shows an example of an array type | mold index.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Key search device D ... Duplication degree determination part DT ... Duplication degree management table SK ... Key search part 11 ... Boundary search part SM ... Search method determination part SL ... 1st duplicate key boundary search part (linear search method)
SB: Second duplicate key boundary search unit (binary search method)
2 ... Storage device A ... Array type index 3 ... Input device

Claims (12)

A plurality of search keys arranged in ascending or descending order, and a key having a designated search target key value and different from the search target key value out of a search target key string including a search key having the same key value A key search device for searching a search key that is a boundary with a search key having a value,
A key search unit for searching a search key having the search target key value from the search target key string by a binary search method;
Based on the search key duplication degree of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value A key search device comprising a boundary search unit for searching. The key search device according to claim 1,
The boundary search unit
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search unit for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search unit for searching by a search method;
The first duplicate key boundary search unit or the second duplicate key boundary search unit is based on the degree of duplication of the search key of the search target key column and the ratio of the average cost of the linear search method and the binary search method. A key search device comprising a search method determination unit to be operated. The key search device according to claim 1 or 2,
The key search device according to claim 1, wherein the degree of duplication increases as the number of search keys having the same key value increases. The key search device according to claim 3, wherein
For each key value in the search target key column, a duplication degree management table in which the number of search keys having the key value is registered;
When a search key is added to the search target key column and when a search key is deleted from the search target key column, the duplication degree management table is updated, based on the contents of the updated duplication degree management table And a duplication degree determining unit that obtains an average value of the number of keys for each key value and uses the average value as a duplication degree. A plurality of search keys arranged in ascending or descending order, and a key having a designated search target key value and different from the search target key value out of a search target key string including a search key having the same key value A key search method for searching a search key that is a boundary with a search key having a value,
A key search step of searching a search key having the search target key value from the search target key string by a binary search method;
Based on the duplication degree of the search key of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value And a boundary search step for searching. The key search method according to claim 5, wherein
The boundary search step includes:
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search step for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search step for searching by a search method;
The first duplicate key boundary search step or the second duplicate key boundary search step is performed based on the degree of duplication of the search key of the search target key string and the ratio of the average cost of the linear search method and the binary search method. A key retrieval method comprising: a retrieval method determination step to be activated. The key search method according to claim 5 or 6,
The key search method according to claim 1, wherein the degree of duplication increases as the number of search keys having the same key value increases. The key search method according to claim 7,
When a search key is added to the search target key column and when a search key is deleted from the search target key column, for each key value in the search target key column, the key of the search key having that key value The duplication degree management table in which the number is registered is updated, the average value of the number of keys for each key value is obtained based on the contents of the updated duplication degree management table, and the duplication degree is determined by using the average value as the duplication degree A key search method comprising steps. A plurality of search keys arranged in ascending or descending order, and a search target key string including a search key having the same key value, the specified search target key value and the search target key value; Is a program for functioning as a key search device that searches for a search key that is a boundary with a search key having a different key value,
The computer,
A key search unit for searching a search key having the search target key value from the search target key string by a binary search method;
Based on the search key duplication degree of the search target key column and the ratio of the average cost of the linear search method and the binary search method, one of the linear search method and the binary search method is selected, and the selected search method By using the search key searched by the key search unit as a search start key, a search key serving as a boundary with a search key having the search target key value and having a key value different from the search target key value A program for functioning as a boundary search unit for searching. The program according to claim 9, wherein
The boundary search unit
Using the search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is linearly A first duplicate key boundary search unit for searching by a search method;
Using a search key searched by the key search unit as a search start key, a search key that is a boundary between a search key having the search target key value and a key value different from the search target key value is divided into two A second duplicate key boundary search unit for searching by a search method;
The first duplicate key boundary search unit or the second duplicate key boundary search unit is based on the degree of duplication of the search key of the search target key column and the ratio of the average cost of the linear search method and the binary search method. A program characterized by comprising a search method determination unit to be operated. The program according to claim 9 or 10,
The degree of duplication becomes a larger value as the number of search keys having the same key value increases. The program according to claim 11, wherein
The computer,
When a search key is added to the search target key column and when a search key is deleted from the search target key column, for each key value in the search target key column, the key of the search key having that key value The duplication degree management table in which the number is registered is updated, the average value of the number of keys for each key value is obtained based on the contents of the updated duplication degree management table, and the duplication degree determination is performed using the average value as the duplication degree Program to function as a part.

Images