JP2012174097A - Query extraction device and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device and a method capable of extracting a query whose retrieval number is increasing gradually with the lapse of time within a prescribed period excluding a query to which attention is paid cyclically.SOLUTION: A query extraction device 10 stores a query used for retrieval and the date and time when the query is used as a retrieval log, equally divides a prescribed evaluation period for a query to be a determination object, imparts a first order of orders from the latest date and time of the retrieval logs summed up by a prescribed time unit for each divided calculation object period and a second order of orders from the largest retrieval number, and calculates an order correlation coefficient. Then, the query extraction device 10 extracts the queries whose order correlation coefficient is equal to or larger than a first value as candidate queries, performs similar processing tracing back the start period of the prescribed evaluation period for the candidate queries, and excludes the query for which there is the calculation object period during which the order correlation coefficient becomes equal to or small than a second value from the candidate queries.

Description

  The present invention relates to a query extraction apparatus and method for extracting a query in which the number of searches gradually increases with time within a predetermined period.

  In a search system, an advertisement associated with a search query is distributed and earned. Frequently used search queries are often high-profile topics and news of the user, and the content displayed in the search results is expected to be highly effective in advertising because the user has a high chance of seeing it. . Therefore, what is a high-attention search query is very important information. Further, topics and news that are noticed by users often change frequently, and it is required to quickly grasp signs that the number of searches of a certain search query is increasing (see, for example, Patent Document 1).

JP 2004-013417 A

  Patent Document 1 describes that queries input to a search system are limited in date and time, and queries input within a predetermined period are aggregated, and queries whose number of searches is increasing within the period are extracted. ing. However, the method described in Patent Document 1 cannot extract a query in which the number of searches gradually increases with the passage of time within a predetermined period.

  On the other hand, there are queries in which the number of searches rises periodically, such as seasonal events such as Christmas and Tanabata. These queries can easily be expected to be used more frequently as the event gets closer, so it can be a noise when investigating what queries have a gradual increase in search regardless of seasonality. is there. In the method described in the cited document 1, it is not possible to distinguish these queries in which the number of searches increases periodically.

  An object of the present invention is to provide an apparatus and a method capable of extracting a query in which the number of searches gradually increases with the passage of time within a predetermined period, excluding queries that are periodically noticed. .

  (1) A query extraction device that extracts a query in which the number of searches gradually increases with the passage of time in a predetermined evaluation period, wherein the query and the date and time when the query was used for the search are used as a search log Search log storage means for storing in association; reception means for receiving an input of a query to be determined; dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period; Referring to the search log storage means, the determination target query is given a higher rank for each of the plurality of calculation target periods divided by the dividing means as the date and time of the search log aggregated in a predetermined time unit is newer. Each of the plurality of calculation target periods divided by the dividing unit with respect to the determination target query, with reference to the first rank assigning unit and the search log storage unit The second rank assigning means for assigning a higher rank as the number of searches counted in a predetermined time unit increases, and the rank assigned by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means. Correlation coefficient calculation means for calculating the rank correlation coefficient for the rank given by the second rank grant means, and the rank correlation coefficient calculated by the correlation coefficient calculation means are all the plurality of calculation target periods. And the extraction means for extracting the determination target query as a candidate query whose number of searches is gradually increasing, and the start of the predetermined evaluation period as the calculation target period unit. The evaluation period retroactive setting means for setting the evaluation evaluation period retroactively in the past, and the confirmation evaluation period set by the evaluation period retroactive setting means with the candidate query extracted by the extracting means as the confirmation target query The processing by the dividing means, the first rank assigning means, the second rank assigning means and the correlation coefficient calculating means is executed, and the rank correlation coefficient calculated by the correlation coefficient calculating means is the second When the calculation target period that is less than or equal to the value exists, the filtering means performs processing by the filtering means for excluding the confirmation target query from the candidate query, the evaluation period retroactive setting means, and the filtering means. The processing by the evaluation period retroactive setting means is sequentially performed in the past until the query is excluded from the candidate queries, or until the number of searches of the candidate queries aggregated in the calculation target period included in the confirmation evaluation period becomes zero A query extracting device comprising: repetitive means for performing retroactively and repeatedly.

The query extraction device described in (1) stores a query used for a search and a date and time when the query was used as a search log, and equally divides a predetermined evaluation period for a query to be determined Then, for each divided calculation target period, the first rank in the order of the date and time of the search log aggregated in a predetermined time unit, and the second rank in the descending order of the number of searches aggregated in the predetermined time unit To calculate the rank correlation coefficient for these ranks. Then, the query extraction device extracts a query whose rank correlation coefficient is equal to or higher than the first value in all calculation target periods as a candidate query, and further checks the start of a predetermined evaluation period using the candidate query as a check target query. The same processing is performed retroactively until the number of searches for the target query becomes zero, and a confirmation target query having a calculation target period in which the rank correlation coefficient is equal to or less than the second value is excluded from the candidate queries.
As a result, among the queries to be determined, the queries whose number of searches gradually increases with the passage of time in a predetermined evaluation period are extracted as candidate queries, which are used as confirmation target queries, and the predetermined evaluation The query is automatically retroactive until the number of searches reaches zero, and a query with a period that does not necessarily increase the number of searches gradually with the passage of time is a candidate query as a periodically fluctuating query Can be filtered (excluded).

  (2) In the confirmation evaluation period set by the evaluation period retrospective setting means, the filtering means includes the number of searches for the confirmation target query aggregated in the calculation target period included in the confirmation evaluation period, and the predetermined evaluation period. If the difference from the number of searches of the confirmation target query calculated at the end of the calculation target period included in is equal to or less than a third value, the confirmation target query is excluded from the candidate queries (1) Query extraction device.

  In addition to the query extraction device described in (1), the query extraction device described in (2) is a calculation target period within the confirmation evaluation period in a verification evaluation period in which the start of a predetermined evaluation period is retroactively performed in the past. If the difference between the aggregated number of searches for the query to be confirmed and the number of retrievals for the query to be confirmed aggregated at the end of the calculation target period within the predetermined evaluation period is equal to or less than the third value, the query is extracted from the candidate query. Can be filtered. Thereby, a query having a calculation target period that has the same number of searches in the past as an absolute value can be excluded from the candidate queries as a query that periodically changes.

  (3) A query extraction device that extracts a query in which the number of searches gradually decreases with time in a predetermined evaluation period, and uses the query and the date and time when the query was used for the search as a search log. Search log storage means for storing in association; reception means for receiving an input of a query to be determined; dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period; Referring to the search log storage means, the determination target query is given a higher rank for each of the plurality of calculation target periods divided by the dividing means as the date and time of the search log aggregated in a predetermined time unit is newer. Each of the plurality of calculation target periods divided by the dividing unit with respect to the determination target query, with reference to the first rank assigning unit and the search log storage unit The second rank assigning means for assigning a higher rank as the number of searches counted in a predetermined time unit increases, and the rank assigned by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means. Correlation coefficient calculation means for calculating the rank correlation coefficient for the rank given by the second rank grant means, and the rank correlation coefficient calculated by the correlation coefficient calculation means are all the plurality of calculation target periods. In the case where the determination target query is less than or equal to the first value, an extraction unit that extracts the determination target query as a candidate query in which the number of searches gradually decreases, and a start period of the predetermined evaluation period is set as the calculation target period unit. The evaluation period retroactive setting means for setting the evaluation evaluation period retroactively in the past, and the confirmation evaluation period set by the evaluation period retroactive setting means with the candidate query extracted by the extracting means as the confirmation target query The processing by the dividing means, the first rank assigning means, the second rank assigning means and the correlation coefficient calculating means is executed, and the rank correlation coefficient calculated by the correlation coefficient calculating means is the second When the calculation target period that is greater than or equal to the value exists, filtering means for excluding the confirmation target query from the candidate query, and the processing by the evaluation period retroactive setting means and the filtering means is performed by the filtering means. The processing by the evaluation period retroactive setting means is sequentially performed in the past until the query is excluded from the candidate queries, or until the number of searches of the candidate queries aggregated in the calculation target period included in the confirmation evaluation period becomes zero A query extracting device comprising: repetitive means for performing retroactively and repeatedly.

The query extraction device described in (3) stores the query used for the search and the date and time when the query was used as a search log, and equally divides the predetermined evaluation period for the query to be determined Then, for each divided calculation target period, the first rank in the order of the date and time of the search log aggregated in a predetermined time unit, and the second rank in the descending order of the number of searches aggregated in the predetermined time unit To calculate the rank correlation coefficient for these ranks. Then, the query extraction device extracts a query whose rank correlation coefficient is equal to or lower than the first value in all calculation target periods as a candidate query, and further checks the start of a predetermined evaluation period using the candidate query as a check target query. The same processing is performed retroactively until the number of searches of the target query becomes zero, and a confirmation target query having a calculation target period in which the rank correlation coefficient is equal to or greater than the second value is excluded from the candidate queries.
As a result, a query that is subject to determination is extracted as a candidate query for a query whose number of searches gradually decreases with the passage of time in a predetermined evaluation period. The query is automatically retroactive until the number of searches reaches zero, and queries that have periods that do not necessarily have a gradual decrease in the number of searches over time can be used as periodically changing queries. Can be filtered (excluded).

  (4) A query extraction method for extracting a query in which the number of searches gradually increases with the passage of time in a predetermined evaluation period, wherein the computer calculates the query and the date and time when the query was used for the search. A search log storage unit that stores the log as a search log, and accepts an input of a query to be determined in the computer; and the predetermined evaluation period is a plurality of calculation target periods shorter than the predetermined evaluation period. With reference to the division step for equally dividing, and the search log storage means, the search log of the determination target is aggregated in a predetermined time unit for each of the plurality of calculation target periods divided in the division step. A first rank assigning step of assigning a higher rank as the date and time is new, and referring to the search log storage means, In addition, a second rank assigning step for assigning a higher rank for each of the plurality of calculation target periods divided in the dividing step as the number of searches totaled in the predetermined time unit increases, and the plural divided in the dividing step A correlation coefficient calculation step for calculating a rank correlation coefficient with respect to the rank given in the second rank assignment step for the rank given in the first rank assignment step, and a correlation coefficient calculation step Extraction that extracts the determination target query as a candidate query in which the number of searches gradually increases when the calculated rank correlation coefficient is equal to or greater than the first value in all the plurality of calculation target periods And an evaluation period retroactive setting step of retroactively setting the start of the predetermined evaluation period in the past in the calculation target period unit to set a confirmation evaluation period, and the extraction In the confirmation evaluation period set in the evaluation period retroactive setting step, the division step, the first rank assigning step, the second rank assigning step, and the correlation coefficient calculation are performed using the candidate query extracted in step as a confirmation target query. When the calculation target period in which the rank correlation coefficient calculated in the correlation coefficient calculation step is less than or equal to a second value exists, the confirmation target query is excluded from the candidate queries. The filtering step and the processing by the evaluation period retroactive setting step and the filtering step are either excluded from the candidate query in the filtering step or aggregated in the calculation target period included in the confirmation evaluation period. Until the number of searches for the candidate query is zero. A query extraction method for executing a recurring step of repeatedly performing the processing by the price period retroactive setting step sequentially retroactively in the past.

  The query determination method described in (4) can achieve the same effect as the query extraction device described in (1).

  (5) A query extraction device that extracts a query in which the number of searches gradually increases over time in a predetermined evaluation period, and associates the query and the date and time when the query was used for the search as a search log. A search log storage means for storing; a receiving means for receiving an input of a query to be determined; a dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period; and the search log Referring to the storage means, for the query to be determined, a higher rank is given to a date and time of the search log aggregated in a predetermined time unit for each of the plurality of calculation target periods divided by the dividing means. With reference to the one rank assigning means and the search log storage means, for each of the plurality of calculation target periods divided by the dividing means for the query to be determined, The second rank assigning means for assigning a higher rank as the number of searches counted in a fixed time unit increases, and the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the divider means. Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means, and the rank correlation coefficient calculated by the correlation coefficient calculating means for all the plurality of calculation target periods. Extraction means for extracting the query to be determined as a candidate query in which the number of searches is gradually increasing when it is equal to or greater than the first value, and a predetermined confirmation retroactively to the beginning of the predetermined evaluation period Confirmation evaluation period request accepting means for accepting an input of an evaluation period, and the candidate query extracted by the extracting means and confirmation for the predetermined confirmation evaluation period accepted by the confirmation evaluation period request accepting means As an elephant query, the processing by the dividing means, the first rank assigning means, the second rank assigning means, and the correlation coefficient calculating means is executed, and the absolute value of the rank correlation coefficient calculated by the correlation coefficient calculating means is calculated. A query extracting device comprising: filtering means for excluding the confirmation target query from the candidate query when the value is equal to or less than a fourth value.

The query extraction device described in (5) stores the query used for the search and the date and time when the query was used as a search log, and equally divides the predetermined evaluation period for the query to be determined Then, for each divided calculation target period, the first rank in the order of the date and time of the search log aggregated in a predetermined time unit, and the second rank in the descending order of the number of searches aggregated in the predetermined time unit To calculate the rank correlation coefficient for these ranks. Then, the query extraction device extracts a query whose rank correlation coefficient is equal to or higher than the first value in all calculation target periods as a candidate query, and further uses the candidate query as a confirmation target query, and for an evaluation period in which an input is received, The same processing is performed only once, and a confirmation target query having a calculation target period in which the absolute value of the rank correlation coefficient is equal to or smaller than the fourth value is excluded from candidate queries.
As a result, a query that is subject to determination is extracted as a candidate query with a search number that gradually increases with the passage of time in a predetermined evaluation period. A query that has a period in which the number of searches does not necessarily increase gradually with the passage of time is accepted from the candidate query as a periodically fluctuating query. Can be filtered (excluded). In some cases, a user who uses the query extraction apparatus can predict a query that changes to some extent periodically. In this case, the process can be simplified to determine the query.

  (6) A query extraction device that extracts a query in which the number of searches gradually decreases with time in a predetermined evaluation period, and associates the query and the date and time when the query was used for the search as a search log. A search log storage means for storing; a receiving means for receiving an input of a query to be determined; a dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period; and the search log Referring to the storage means, for the query to be determined, a higher rank is given to a date and time of the search log aggregated in a predetermined time unit for each of the plurality of calculation target periods divided by the dividing means. With reference to the one rank assigning means and the search log storage means, for each of the plurality of calculation target periods divided by the dividing means for the query to be determined, The second rank assigning means for assigning a higher rank as the number of searches counted in a fixed time unit increases, and the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the divider means. Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means, and the rank correlation coefficient calculated by the correlation coefficient calculating means for all the plurality of calculation target periods. Extraction means for extracting the query to be determined as a candidate query whose number of searches is gradually decreasing when it is equal to or less than a first value, and a predetermined confirmation retroactively to the beginning of the predetermined evaluation period Confirmation evaluation period request accepting means for accepting an input of an evaluation period, and the candidate query extracted by the extracting means and confirmation for the predetermined confirmation evaluation period accepted by the confirmation evaluation period request accepting means As an elephant query, the processing by the dividing means, the first rank assigning means, the second rank assigning means, and the correlation coefficient calculating means is executed, and the absolute value of the rank correlation coefficient calculated by the correlation coefficient calculating means is calculated. A query extracting device comprising: filtering means for excluding the confirmation target query from the candidate query when the value is equal to or less than a fourth value.

The query extraction device described in (6) stores the query used for the search and the date and time when the query was used as a search log, and equally divides a predetermined evaluation period for the query to be determined Then, for each divided calculation target period, the first rank in the order of the date and time of the search log aggregated in a predetermined time unit, and the second rank in the descending order of the number of searches aggregated in the predetermined time unit To calculate the rank correlation coefficient for these ranks. Then, the query extraction device extracts a query whose rank correlation coefficient is equal to or lower than the first value in all the calculation target periods as a candidate query, and further uses the candidate query as a confirmation target query, for the evaluation period in which the input is accepted. The same processing is performed only once, and a confirmation target query having a calculation target period in which the absolute value of the rank correlation coefficient is equal to or smaller than the fourth value is excluded from candidate queries.
As a result, a query that is subject to determination is extracted as a candidate query, and the query whose number of searches gradually decreases with the passage of time in a predetermined evaluation period. A query that has a period in which the number of searches is not necessarily declining gradually as time passes is accepted as a query that periodically fluctuates. Can be filtered (excluded). In some cases, a user who uses the query extraction apparatus can predict a query that changes to some extent periodically. In this case, the process can be simplified to determine the query.

  According to the present invention, it is possible to provide an apparatus and a method capable of extracting a query in which the number of searches gradually increases with the passage of time within a predetermined period, excluding queries that are periodically noticed. .

It is a figure which shows the functional block of the query extraction apparatus which concerns on 1st Embodiment of this invention, and the function outline | summary of its peripheral device. It is a figure which shows search log DB which concerns on 1st Embodiment of this invention. It is a figure which shows the graph which shows transition of the search number which concerns on 1st Embodiment of this invention. It is a flowchart of the process which the query extraction apparatus which concerns on 1st Embodiment of this invention performs. It is a flowchart of the process which the query extraction apparatus which concerns on 1st Embodiment of this invention performs. It is a figure which shows the example of a display of the query extraction apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the example of a display of the query extraction apparatus which concerns on 1st Embodiment of this invention. It is a figure which shows the functional outline of the query extraction apparatus which concerns on 3rd Embodiment of this invention, and its peripheral device function. It is a figure which shows the graph which shows transition of the search number which concerns on 3rd Embodiment of this invention. It is a flowchart of the process which the query extraction apparatus which concerns on 3rd Embodiment of this invention performs. It is a figure which shows the functional block of the query extraction apparatus which concerns on 4th Embodiment of this invention, and the functional outline | summary of the peripheral device. It is a flowchart of the process which the query extraction apparatus which concerns on 4th Embodiment of this invention performs. It is a flowchart of the process which the query extraction apparatus which concerns on 4th Embodiment of this invention performs. It is a figure which shows the example of a display of the query extraction apparatus which concerns on 4th Embodiment of this invention.

  Embodiments of the present invention will be described below with reference to the drawings. The embodiment described below is merely an example, and the technical scope of the present invention is not limited to this.

  With reference to FIGS. 1 to 3, a query extraction apparatus 10 according to a first embodiment of the present invention will be described. The embodiment described below is merely an example, and the technical scope of the present invention is not limited to this.

  A query extraction apparatus 10 according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram illustrating a functional block of the query extraction device 10 and a functional outline of peripheral devices.

  This embodiment is applied to a computer (query extraction device 10) and its peripheral devices. Each unit in the present embodiment is configured by hardware included in a computer and its peripheral devices, and software that controls the hardware.

  The hardware includes a CPU as a control unit, a storage unit, a communication unit, a display unit, and an input unit. Examples of the storage unit include a memory (RAM, ROM, etc.), a hard disk drive (HDD), and an optical disk (CD, DVD, etc.) drive. Examples of the communication unit include various wired and wireless interface devices. Examples of the display unit include various displays such as a liquid crystal display and a plasma display. Examples of the input unit include a keyboard and a pointing device (mouse, tracking ball, etc.).

  The software includes a computer program and data for controlling the hardware. The computer program and data are stored in the storage unit, and are appropriately executed and referenced by the control unit. The computer program and data can be distributed via a communication network, and can also be stored and distributed on a computer-readable medium such as a CD-ROM.

Hereinafter, the functional configuration of the query extraction device 10 will be described.
The query extraction device 10 includes an operation unit 11 that receives a direct input from an operator of the query extraction device 10, a display unit 12 that displays information on the function of the query extraction device 10, and a function of the query extraction device 10. A storage unit 13 that stores various programs (not shown) for executing the program, a program (not shown) for executing the functions of the present invention, and the like, and a control unit 14 that comprehensively controls each function related to the query extraction device 10. .

The operation unit 11 is implemented by, for example, a keyboard and a pointing device (mouse, tracking ball, etc.).
The display unit 12 is mounted by various displays such as a liquid crystal display and a plasma display, for example.

  In addition to the various programs described above, the storage unit 13 includes a search log DB 131 serving as a search log storage unit that stores a query used for Web page search as a search log.

  FIG. 2 is a diagram showing the search log DB 131 according to the first embodiment. The search log DB 131 uses, as a search log, data that associates a query input from a user terminal (not shown) to a search engine (not shown) that searches a Web page and the date and time when the query was used for the search. Remember. The search log DB 131 includes a query string in which a query used for the search is stored, and a search date / time string in which information indicating the date / time when the query is input to the search engine is stored.

  Returning to FIG. 1, the control unit 14 is configured by a CPU or the like, and performs control of each processing operation of the query extraction device 10 and information communication control. The control unit 14 includes a receiving unit 141 as a receiving unit, a dividing unit 142 as a dividing unit, a first ranking providing unit 143 as a first ranking providing unit, and a second ranking providing as a second ranking providing unit. Unit 144, correlation coefficient calculation unit 145 as a correlation coefficient calculation unit, extraction unit 146 as an extraction unit, evaluation period retroactive setting unit 147 as an evaluation period retroactive setting unit, and filtering unit 148 as a filtering unit And a repeater 149 as a repeater, and an output unit 150.

The accepting unit 141 accepts an input of a predetermined evaluation period for evaluating the query together with the query to be determined from the operator of the query extraction device 10 via the operation unit 11. One or more queries may be input. In the present embodiment, a description will be given assuming that a plurality of queries are input to the query extraction device 10.
The predetermined evaluation period is a target period for the query extraction apparatus 10 to determine whether or not the determination target query is a query whose search number is gradually increasing. The evaluation period may be determined in advance, and an arbitrary period may be set together with the query to be evaluated.

  The dividing unit 142 equally divides the predetermined evaluation period into a plurality of calculation target periods shorter than the evaluation period.

  The first rank assigning unit 143 refers to the search log DB 131 and aggregates the date and time when the query to be determined is used for each of a plurality of calculation target periods in a predetermined time unit, and the date and time from the newest to the oldest Is given a ranking. This assigned rank is set as the first rank. That is, the first rank is a rank in which the time ranges divided into predetermined time units are arranged in the new order in the calculation target period. Here, the predetermined time unit may be any time such as 1 hour unit, 6 hour unit, 12 hour unit, 24 hour unit (one day), or the like. Further, the predetermined time unit is preferably a time range in which the calculation target period can be equally divided.

  The second rank assigning unit 144 refers to the search log DB 131 and totals the number of searches in which the query to be determined is used for each of a plurality of calculation target periods in a predetermined time unit used by the first rank assigning unit. The ranking is given in descending order of the number of searches. This assigned rank is set as the second rank. That is, the second rank is a rank in which the time ranges divided by the predetermined time range in the calculation target period are arranged in descending order of the number of searches of the query to be determined that are aggregated in a predetermined time unit in the calculation target period.

The correlation coefficient calculation unit 145 calculates a rank correlation coefficient (search score) for the second rank of the first rank for each calculation target period. In the following description, an example in which the predetermined time unit is one day (24 hours) will be described.
Here, the following expression (1) is used to calculate the search score.

  In the equation (1), ρ is a rank correlation coefficient. The rank correlation coefficient is a correlation coefficient between the first rank and the second rank in each calculation target period. Specifically, the correlation coefficient calculation unit 145 uses the following (2) to rank A correlation coefficient (Spearman's rank correlation coefficient) is calculated.

上記の（２）式において、Ｎは計算対象期間であり、各計算対象期間における、所定の時間単位で集計した検索ログにおける、第１の順位と第２の順位との差をＤとする。ここで、順位相関係数が取り得る値は、−１から１である。

In the above equation (2), N is the calculation target period, and D is the difference between the first rank and the second rank in the search log totaled in a predetermined time unit in each calculation target period. Here, the value that the rank correlation coefficient can take is −1 to 1.

  The extraction unit 146 determines that the rank correlation coefficient (search score) calculated by the correlation coefficient calculation unit 145 is greater than or equal to the first value in each time range divided into predetermined time units of all calculation target periods. It is determined whether or not there is. The extraction unit 146 extracts, as candidate queries, queries whose search scores for all calculation target periods are greater than or equal to the first value.

  Here, the description will be continued by taking the case where the first value is 0.7 as an example. The extraction unit 146 determines the number of searches in the evaluation period when the search score calculated by the correlation coefficient calculation unit 145 is 0.7 or more in each time range divided into predetermined time units of all calculation target periods. Is extracted as a candidate query to be extracted by the query extraction device 10.

  The evaluation period retroactive setting unit 147 sets a confirmation evaluation period in which the start of the evaluation period is retroactively set for the candidate query. The unit of the retroactive period is preferably the calculation target period unit of the evaluation period. As a result, for the candidate query, a confirmation evaluation period in which the start date is retroactive from the evaluation period is newly set.

  The filtering unit 148 filters (excludes) queries whose number of searches periodically increases, such as seasonal events. Specifically, the filtering unit 148 uses the candidate query as a confirmation target query, and for the confirmation evaluation period newly set by the evaluation period retroactive setting unit 147, the dividing unit 142, the first rank assigning unit 143, and the second rank assigning unit 144. Similarly, the processes of the correlation coefficient calculation unit 145 and the extraction unit 146 are sequentially executed, and it is determined whether or not there is a calculation target period in which the search score is equal to or less than the second value in the confirmation evaluation period. In the confirmation evaluation period, the filtering unit 148 excludes the confirmation target query from the candidate queries when there is a calculation target period in which the search score is equal to or smaller than the second value.

In the present embodiment, for example, the second value is -0.7.
If there is a calculation target period that is less than or equal to the second value in the confirmation evaluation period, it means that there was a period in which the number of searches gradually decreased in the period (confirmation evaluation period) that includes the past period from the start of the evaluation period To do. Queries having a period in which the number of searches decreases in the past, such as Tanabata and Christmas, tend to decrease the number of searches if a seasonal event elapses despite the periodic increase in the number of searches. It is a strong query. The filtering unit 148 excludes such queries that periodically increase the number of searches from candidate queries.

  For the confirmation target query, the repetition unit 149 determines whether the filtering unit 148 has excluded the confirmation target query from the candidate queries, or whether the number of candidate query searches aggregated in the calculation target period included in the confirmation evaluation period has become zero. Determine whether. When the filtering unit 148 excludes the confirmation target query from the candidate queries, the repetition unit 149 ends the process for the confirmation target query. Moreover, the repetition part 149 also complete | finishes the process with respect to the said confirmation object query, when there exists a calculation object period whose search number is zero in a confirmation object period. In the latter case, the repetition unit 149 does not exclude the confirmation target query from the candidate queries, and outputs the query as a processing result of the query extraction device 10.

  In addition, when the filtering unit 148 does not exclude the confirmation target query from the candidate queries, or when there is no calculation target period in which the number of searches is zero in the confirmation evaluation period, the repetition unit 149 continues processing for the confirmation target query. . That is, the repetition unit 149 causes the evaluation period retroactive setting unit 147 to exclude the filtering target 148 from the candidate queries or until it determines that there is a calculation target period in which the number of searches is zero in the confirmation evaluation period. A confirmation evaluation period in which the start period is retroactively set in the past is set, and the same processing as described above is performed in the dividing unit 142, the first rank assigning unit 143, the second rank assigning unit 144, the correlation coefficient calculating unit 145, and the extracting unit 146. Repeatedly.

  The output unit 150 outputs the processing result of the query extraction device 10. In this embodiment, it is displayed on the display unit 12. This processing result is a result of extracting a query in which the number of searches gradually increases in a predetermined evaluation period by excluding a query in which the number of searches periodically increases from among a plurality of queries input to the reception unit 141. It is.

As described above, the query extraction device 10 determines whether or not the number of searches for the determination target query is gradually increasing in the predetermined evaluation period received by the reception unit 141 for the determination target query received by the reception unit 141. Determine.
Specifically, as shown in FIG. 3, for example, a case where the evaluation period is one month will be described. In FIG. 3, this is the period indicated by A.

First, for one month, which is a predetermined evaluation period received by the receiving unit 141, the dividing unit 142 divides the predetermined evaluation period at equal intervals. Here, it is assumed that the dividing unit 142 divides the calculation target period in units of one week (period B in FIG. 3).
Next, the first ranking assigning unit 143 further aggregates the calculation target periods in a predetermined time unit such as a day unit, and ranks the time ranges divided into the predetermined time units in order of newest (first order). 1), the second ranking assigning unit 144 performs ranking in descending order of the number of searches for each time range used by the first ranking granting unit 143 (second ranking). Then, the correlation coefficient calculation unit 145 calculates the rank correlation coefficient (search score) between the first rank and the second rank, and the extraction section 146 calculates the rank correlation coefficient in all calculation target periods. Queries having a value of 1 (for example, 0.7) or more are extracted as candidate queries.

The query extraction device 10 further determines whether or not the extracted candidate query is a query whose number of searches periodically changes.
First, the evaluation period retroactive setting unit 147 sets, for the candidate query, a confirmation evaluation period in which the start of the evaluation period is retroactively (period C in FIG. 3). Next, the filtering unit 148 uses the candidate query as a confirmation target query, and determines whether or not the rank correlation coefficient (search score) in the confirmation evaluation period is equal to or less than the second value. The filtering unit 148 excludes the confirmation target query whose rank correlation coefficient in the calculation target period in the confirmation evaluation period is equal to or less than the second value from the candidate query. If there is no calculation target period that is equal to or less than the second value in the confirmation evaluation period, the repetition unit 149 further retroactively extends the start of the confirmation evaluation period to the past, and adds a new confirmation evaluation period (see FIG. 3), and repeatedly determines whether or not the rank correlation coefficient for the confirmation target query is equal to or less than the second value for the newly set confirmation evaluation period.

  Here, with respect to a certain confirmation target query, the rank correlation coefficient of the calculation target period in the confirmation evaluation period is equal to or smaller than the second value is a case where a transition like E shown in FIG. That is, when the number of searches shows a tendency to gradually decrease, the rank correlation coefficient is equal to or less than the second value. If the number of searches temporarily decreases and then the number of searches gradually increases, the confirmation target query is likely to be a periodically changing query. The query can be removed.

  Processing performed by the query extraction device 10 will be described with reference to FIG. FIG. 4 is a flowchart of processing performed by the query extraction device 10.

  In step S <b> 11, the reception unit 141 of the query extraction device 10 receives an input of a query to be determined by the operation unit 11. The query received by the receiving unit 141 is not limited to a single query, and may be a plurality of queries. Hereinafter, the description will be given assuming that the reception unit 141 receives a plurality of queries. The reception unit 141 also receives an input of an evaluation period.

  In step S12, the dividing unit 142 of the query extraction device 10 divides the evaluation period received by the receiving unit 141 into a plurality of calculation target periods. The dividing unit 142 equally divides the evaluation period into a plurality of calculation target periods.

  In step S13, the first rank assigning unit 143 of the query extraction device 10 assigns the first rank. Specifically, the first rank assigning unit 143 aggregates each calculation target period for each predetermined time unit, and assigns a higher rank to the time range divided by the time unit from the newest to the oldest. To do.

  In step S14, the second rank assigning unit 144 of the query extraction device 10 assigns the second rank. Specifically, the second rank assigning unit 144 counts the number of searches for each predetermined time unit used by the first rank assigning unit 143, and for the time range divided by the predetermined time unit, A higher ranking is assigned in descending order of the number of searches.

  In step S15, the correlation coefficient calculation unit 145 of the query extraction device 10 calculates a rank correlation coefficient for each calculation target period.

  In step S16, the extraction unit 146 of the query extraction device 10 determines whether or not the rank correlation coefficient (search score) in all the calculation target periods is equal to or greater than the first value. The extraction unit 146 moves the process to step S17 for the query if YES in this process, and moves the process for the query to step S21 if NO. Here, the processing from step S <b> 12 to step S <b> 16 is performed for all the determination target queries received by the receiving unit 141.

  In step S17, the extraction unit 146 of the query extraction device 10 extracts candidate queries. The extraction unit 146 extracts, as candidate queries, queries whose rank correlation coefficients (search scores) in all target periods are greater than or equal to the first value among the determination target queries received by the reception unit 141.

  In step S18, the evaluation period retrospective setting unit 147 of the query extraction device 10 sets one query among the candidate queries extracted in step S17 as a check target query, and sets a check evaluation period for the check target query. Specifically, the evaluation period retroactive setting unit 147 newly sets a period in which the start of the evaluation period accepted by the accepting unit 141 is retroactively set as a confirmation evaluation period.

  In step S19, the filtering unit 148 of the query extraction device 10 performs a filtering process. Details will be described later.

  In step S20, the repetition unit 149 of the query extraction device 10 determines whether all candidate queries have been processed. If this process is YES, the repetition unit 149 moves the process to step S21, and if NO, returns the process to step S18 for a different candidate query. Here, in step S18, the evaluation period retroactive setting unit 147 newly sets a period in which the start of the evaluation period received by the receiving unit 141 is further retroactively set as a new evaluation query for a new candidate query.

  In step S <b> 21, the output unit 150 of the query extraction device 10 outputs the processing result to the display unit 12.

  With reference to FIG. 5, the filtering process which the query extraction apparatus 10 performs is demonstrated. FIG. 5 is a flowchart of the filtering process.

  In step S31, the dividing unit 142 of the query extraction device 10 divides the confirmation evaluation period set in step S18 into a plurality of calculation target periods. The calculation target period is preferably a period obtained by equally dividing the confirmation target period.

  In step S32, the first rank assigning unit 143 of the query extraction device 10 assigns the first rank. Specifically, the first rank assigning unit 143 aggregates the calculation target periods for each predetermined time unit, and ranks the time ranges divided into the predetermined time units from the newest to the oldest. Give.

  In step S33, the second rank assigning unit 144 of the query extraction device 10 assigns the second rank. Specifically, the second ranking assigning unit 144 has a large number of searches for the query to be evaluated in the time range divided by the predetermined time unit used by the first ranking granting unit 143 for each of a plurality of calculation target periods. Give higher ranks in order.

  In step S34, the correlation coefficient calculation unit 145 of the query extraction device 10 calculates a rank correlation coefficient for each calculation target period.

  In step S35, the extraction unit 146 of the query extraction device 10 determines whether there is a calculation target period that is equal to or less than the second value among the rank correlation coefficients (search scores) for each of the plurality of calculation target periods. . The extraction unit 146 moves the process to step S36 if YES in this process, and moves the process to step S37 if NO.

  In step S36, the filtering unit 148 of the query extraction apparatus 10 excludes the confirmation target query having the calculation target period that is equal to or less than the second value from the candidate query, and moves the process to step S20 in FIG.

  In step S <b> 37, the repetition unit 149 of the query extraction device 10 determines whether there is a period in which the number of searches is zero among the calculation target periods in the confirmation evaluation period. The repetition unit 149 moves the process to step S20 in FIG. 4 if YES in this process, and moves the process to step S18 in FIG. 4 if NO. Here, in step S18, the evaluation period retroactive setting unit 147 newly sets a period in which the start of the confirmation evaluation period determined in step S19 is further retroactively set as the confirmation evaluation period.

A display example of the processing result will be described with reference to FIGS. 6 and 7 are diagrams showing display examples displayed on the display unit 12 by the output unit 150. FIG.
According to FIG. 6, it is determined on the display screen 60 that the evaluation period 61 input to the reception unit 141 by the operation unit 11 is displayed, and that the number of searches gradually increases during the evaluation period. The extracted query is displayed in the search increase query part 62. Further, the query filtered by the filtering unit 148 is displayed on the exclusion query unit 63.

  Further, according to FIG. 6, all the queries input to the receiving unit 141 are displayed in the ranking display unit 65 in the ranking format in descending order of the number of searches in the evaluation period. Here, the queries corresponding to the first and third places are displayed with different background colors. This indicates that the query is filtered by the filtering unit 148. Thereby, it is possible to make the user of the query extraction device 10 recognize which query is a filtering target, that is, whether the query is a query whose number of searches periodically changes.

[Second Embodiment]
The second embodiment will be described below. The second embodiment differs in the calculation method of the rank correlation coefficient calculated by the correlation coefficient calculation unit 145. The following description will focus on differences from the first embodiment, and parts not specifically described are the same as in the first embodiment.

ここで、上記の（３）式で、ρは第１実施形態と同様の順位相関係数であるので説明を省略する。また、上記の（３）式で、Ｗ１及びＷ２は重み付けの係数である。 The correlation coefficient calculation unit 145 calculates the rank correlation coefficient using the following formula.

Here, in the above equation (3), ρ is a rank correlation coefficient similar to that in the first embodiment, and thus the description thereof is omitted. In the above equation (3), W1 and W2 are weighting coefficients.

（４）式でＮは計算対象期間であり、ｚｄは検索数が０である日数である。 Specifically, W1 is calculated using the following equation (4). This is for adding the number of days in which the number of searches is 0 to the calculation result, and for correcting the value of ρ that decreases the reliability when the value is large in 0.

In Expression (4), N is a calculation target period, and zd is the number of days in which the number of searches is zero.

  In the above equation (3), W2 is a second weighting coefficient, and is calculated in detail using the following equation (5). This is for taking into account the degree to which the aggregated values of a plurality of calculation target periods included in the evaluation period are included in the respective aggregated value ranges at the start and end of the evaluation period.

上記（５）式でＮは計算対象期間であり、ｘｄは評価期間の始期と終期におけるそれぞれの集計値の範囲から集計値が逸脱する計算対象期間の数である。

In the above equation (5), N is the calculation target period, and xd is the number of calculation target periods where the total value deviates from the range of the total value at the start and end of the evaluation period.

  As described above, the value of the rank correlation coefficient ρ can be corrected by adding a weighting coefficient.

[Third Embodiment]
Hereinafter, the third embodiment will be described with reference to FIGS. 8 to 10.
The third embodiment differs from the first and second embodiments in the processing performed by the filtering unit 158. The following description will focus on differences from the first embodiment, and parts not specifically described are the same as in the first embodiment.

  With reference to FIG.8 and FIG.9, the query extraction apparatus 10 in 3rd Embodiment is demonstrated. FIG. 8 is a diagram showing a functional block of the query extraction device 10 and its peripheral devices in the third embodiment, and FIG. 9 is a diagram showing a graph showing the transition of the number of searches in the third embodiment. .

  The filtering unit 158 of the query extraction device 10 according to the third embodiment calculates the difference between the number of searches in the calculation target period included in the confirmation evaluation period and the number of searches in the calculation target period included in the evaluation period for the confirmation target query. It is calculated and it is determined whether or not the difference value is equal to or smaller than the third value.

  Specifically, the filtering unit 158 refers to the search log DB 131 and totals the number of searches for the confirmation target query and the calculation target period of the confirmation evaluation period. Further, the filtering unit 158 refers to the search log DB 131 and totals the number of searches for the calculation target period of the evaluation period for the confirmation target query (the period in which the reception unit 141 has received input). Then, the filtering unit 158 compares the number of searches for each calculation target period included in the evaluation period with the number of searches for each calculation target period included in the confirmation target period. At this time, the object to be compared with the filtering unit 158 compares the number of searches indicating the maximum value among the number of searches in the calculation target period included in the evaluation period and the number of searches in the calculation target period included in the confirmation target period. In addition, although the evaluation period is included in the confirmation target period, when the filtering unit 158 compares, the calculation in the period that does not include the evaluation period (that is, the period past the start of the evaluation period) in the confirmation target period. The number of searches in the target period is preferable. Then, when the difference value is equal to or smaller than the third value, the filtering unit 158 excludes the confirmation target query from the candidate queries as a filtering target. This third value is a value indicating the range of F shown in FIG.

Here, the number of searches in the evaluation period compared by the filtering unit 158 may be the number of searches in the calculation target period including the end of the evaluation period. This is because the number of searches in the calculation target period including the end of the evaluation period is often close to the period that the user wants to evaluate most, and often indicates the maximum value.
Since the confirmation evaluation period is a period including a period earlier than the evaluation period, when the difference (that is, the third value) between the number of searches in the confirmation evaluation period and the maximum number of searches in the evaluation period is sufficiently small In the past, it can be inferred that there has been a similar change in the number of searches, and it can be estimated that the query is a query in which the number of searches periodically changes.

  Further, the third value may be 10% of the maximum value of the number of searches in the evaluation period.

  With reference to FIG. 10, the filtering process in 2nd Embodiment is demonstrated. FIG. 10 is a flowchart of the filtering process in the third embodiment.

  In step S41, the dividing unit 142 of the query extraction device 10 divides the confirmation evaluation period into a plurality of calculation target periods for the confirmation target query.

  In step S42, the filtering unit 158 of the query extraction device 10 totals the number of searches in each calculation target period in the confirmation evaluation period.

  In step S43, the filtering unit 158 of the query extraction device 10 aggregates the number of searches in the calculation target period including the end of the evaluation period or the number of searches in each calculation target period in the evaluation period for the confirmation target query.

  In step S44, the filtering unit 158 of the query extraction device 10 calculates the difference between the number of searches in each calculation target period of the confirmation evaluation period and the number of searches in each calculation target period of the evaluation period.

  In step S45, the filtering unit 158 of the query extraction device 10 determines whether or not the difference values calculated in step S46 are each equal to or smaller than the third value. If YES in this process, the filtering unit 158 moves the process to step S46, and if NO, moves the process to step S20 in FIG.

  In step S46, the filtering unit 158 of the query extraction device 10 excludes the confirmation target query that is equal to or lower than the third value from the candidate queries, and moves the process to step S20 in FIG.

  According to the above-described embodiment, the query extraction device 10 can extract a query in which the number of searches gradually increases with the passage of time within a predetermined period, excluding queries that are periodically noticed.

[Fourth Embodiment]
Hereinafter, the fourth embodiment will be described with reference to FIGS. 11 to 14.
The fourth embodiment is different from the first embodiment in that an input about a confirmation evaluation period is received from a user. The following description will focus on differences from the first embodiment, and parts not specifically described are the same as in the first embodiment.

  With reference to FIG. 11, the query extraction apparatus 10 in 4th Embodiment is demonstrated. FIG. 11 is a diagram illustrating functional outlines of functional blocks of the query extraction device 10 and peripheral devices thereof according to the fourth embodiment.

  The query extraction device 10 according to the fourth embodiment includes a confirmation evaluation period reception unit 167 as confirmation evaluation period request reception means. The confirmation evaluation period receiving unit 167 receives an input about the start period from the user for a predetermined evaluation period.

The filtering unit 168 includes a dividing unit 142, a first rank assigning unit 143, a first ranking unit from the start date of the confirmation evaluation period received by the confirmation evaluation period reception unit 167 to the end of the predetermined evaluation period. Similarly, the processes of the second rank assigning unit 144, the correlation coefficient calculating unit 145, and the extracting unit 146 are sequentially executed to determine whether or not the absolute value of the search score is equal to or smaller than the fourth value in the confirmation evaluation period. The filtering unit 168 excludes the confirmation target query from the candidate queries when the absolute value of the search score becomes the fourth value or less during the confirmation evaluation period.
For example, the fourth value may be 0.1.

  With reference to FIG.12 and FIG.13, the process which the query extraction apparatus 10 in 4th Embodiment performs is demonstrated. FIG. 12 is a flowchart of processing performed by the query extraction device 10.

  The processing from step S51 to step S57 is the same as the processing from step S11 to step S17 in the first embodiment of FIG.

  In step S58, the confirmation evaluation period reception unit 167 of the query extraction device 10 displays on the display unit 12 the candidate query extracted by the extraction unit 146 in step S57 on the display unit 12. In step S <b> 59, the confirmation evaluation period reception unit 167 receives an input about the start of the confirmation evaluation period in accordance with the operation of the operation unit 11.

  In step S60, the filtering unit 168 of the query extraction device 10 performs a filtering process. Details will be described later.

  In step S61, the repetition unit 149 of the query extraction device 10 determines whether all candidate queries have been processed. If this process is YES, the repetition unit 149 moves the process to step S62, and if NO, returns the process to step S60 to process different candidate queries.

  With reference to FIG. 13, the filtering process which the query extraction apparatus 10 in 4th Embodiment performs is demonstrated. FIG. 13 is a flowchart of the filtering process performed by the query extraction device 10.

  Steps S71 to S74 are the same as steps S31 to S34 in the first embodiment shown in FIG.

  In step S75, the extraction unit 146 of the query extraction device 10 determines whether or not the absolute value of the rank correlation coefficient (search score) in the designated calculation target period is equal to or less than the fourth value. The extraction unit 146 moves the process to step S76 if YES in this process, and ends this process if NO.

  In step S76, the filtering unit 168 of the query extraction apparatus 10 excludes the confirmation target query having a calculation target period that is equal to or greater than zero and equal to or less than the fourth value from the candidate query, and moves the process to step S61 in FIG.

  With reference to FIG. 14, the example of a display displayed on the display part 12 when the start period of the confirmation evaluation period reception part 167 confirmation evaluation period is received from a user is demonstrated. FIG. 14 is a diagram showing a display example displayed on the display unit 12.

  On the display screen 60, a list 71 of candidate queries extracted by the extraction unit 146 is displayed. In addition, the current evaluation period 72 is displayed on the display screen 60, and an input field 73 for inputting the start of the confirmation evaluation period is displayed below the display period 60. The user of the query extraction device 10 inputs the start of the confirmation evaluation period into the input field 73 via the operation unit 11. Then, the query extraction apparatus 10 performs a filtering process in response to accepting a pressing operation on the OK button 74.

  According to the above-described fourth embodiment, it is possible to receive an input from the user at the start of the confirmation evaluation period, and to filter a query whose number of searches periodically changes based on the input confirmation evaluation period. For example, when a user confirms a candidate query, the user may have an idea as to what kind of query changes periodically. In such a case, the process performed by the query extraction device 10 can be simplified by receiving an input for the start of the confirmation evaluation period and performing the process only once for the confirmation evaluation period.

  As mentioned above, although embodiment of this invention was described, this invention is not restricted to embodiment mentioned above. The effects described in the embodiments of the present invention are only the most preferable effects resulting from the present invention, and the effects of the present invention are limited to those described in the embodiments of the present invention. is not.

  The output unit 150 displays the processing result of the query extraction device 10 on the display unit 12, but is not limited thereto. It may be displayed on another computer through a printout by a printer or a network.

  As shown in the first embodiment and the second embodiment, the filtering units 148 and 158 may perform both the process shown in FIG. 5 and the process shown in FIG. In this case, the filtering units 148 and 158 can perform steps S42 to S46 in FIG. 10 after steps S36 and S37 in FIG. 5, and the processing performed by the filtering unit 148 and the filtering unit 158 The processing to be performed may be reversed.

  Further, in the above-described embodiment, the query extraction device 10 extracts a query in which the number of query searches is gradually increasing and does not change periodically, but is not limited thereto. For example, a query in which the number of searches gradually decreases may be extracted. Specifically, the extraction unit 146 determines whether or not the rank correlation coefficient (search score) calculated by the correlation coefficient calculation unit 145 is equal to or smaller than a first value, and the rank correlator has the first value as the number. In the following cases, the query is extracted as a candidate query. Also, the filtering units 148 and 158 exclude, from the candidate queries, confirmation target queries whose rank correlation coefficient in the calculation target period in the confirmation evaluation period is equal to or greater than the second value. Then, the repetition unit 149 repeatedly determines whether or not the rank correlation coefficient of the confirmation target query is equal to or greater than the second value for the newly set confirmation evaluation period in which the start of the confirmation evaluation period is retroactive. . Similarly, in the fourth embodiment, a query in which the number of searches gradually decreases may be extracted.

  In the fourth embodiment, the query extraction apparatus 10 displays a list of candidate queries extracted by the extraction unit 146 on the display unit 12 before performing the process for the confirmation target query, and inputs the start of the confirmation evaluation period. It was accepted, but not limited to this. For example, after displaying a list of candidate queries on the display unit 12, the process may be continued without inputting the start of the confirmation evaluation period. For example, when a pressing operation of the OK button 74 is accepted as a process continuation instruction without inputting the start period, the same process as in the first embodiment may be performed. That is, the query extraction device 10 may include an evaluation period retroactive setting unit 147, and automatically perform a filtering process by setting a period in which the start of the confirmation evaluation period is retroactively performed in the past.

  In the fourth embodiment, the query extraction apparatus 10 performs the filtering process only once based on the start of the confirmation evaluation period received by the confirmation evaluation period reception unit 167, but the present invention is not limited to this. For example, after confirming the screen displaying the processing result, an input regarding the confirmation evaluation period may be received again and the processing may be performed. In this case, for example, on the display screen 60 that displays the processing result of FIG. 6 or FIG. 7, the input column 73 and the OK button 74 for accepting the input about the start of the confirmation evaluation period of FIG. ) Or the ranking display unit 65 (FIG. 7).

  In the fourth embodiment, in the query extraction device 10, the confirmation evaluation period reception unit 167 accepts only the input of the start of the confirmation evaluation period, but the present invention is not limited to this. You may receive the input about a period. In this case, the confirmation evaluation period receiving unit 167 receives an input of a period indicating an expected period such as one year, one month, one week, and sets the start period by performing reverse calculation from a predetermined end period.

DESCRIPTION OF SYMBOLS 10 Query extraction apparatus 11 Operation part 12 Display part 13 Storage part 14 Control part 141 Reception part 142 Division part 143 1st rank provision part 144 Second rank provision part 145 Correlation coefficient calculation part 146 Extraction part 147 Evaluation period retroactive setting part 148 158 Filtering unit 149 Repeating unit 150 Output unit

Claims (6)

A query extraction device that extracts a query in which the number of searches gradually increases with time in a predetermined evaluation period,
Search log storage means for storing the query and the date and time when the query was used for the search as a search log;
A receiving means for receiving an input of a query to be judged;
Dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period;
Referring to the search log storage means, for the determination target query, for each of the plurality of calculation target periods divided by the division means, the higher the ranking is, the newer the date and time of the search log aggregated in a predetermined time unit. A first order giving means for giving,
With reference to the search log storage unit, a higher rank is assigned to the determination target query as the number of searches aggregated in a predetermined time unit increases for each of the plurality of calculation target periods divided by the division unit. 2 rank granting means;
Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means of the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means; ,
When the rank correlation coefficient calculated by the correlation coefficient calculation means is equal to or greater than a first value in all the plurality of calculation target periods, the number of searches for the determination target query is gradually increased. Extraction means for extracting as a candidate query,
An evaluation period retroactive setting means for retroactively setting the start of the predetermined evaluation period in the past in the calculation target period unit and setting the confirmation evaluation period;
In the confirmation evaluation period set by the evaluation period retrospective setting means, the dividing means, the first rank assigning means, the second rank assigning means, and the correlation are set with the candidate query extracted by the extracting means as a check target query. When the calculation target period in which the rank correlation coefficient calculated by the correlation coefficient calculation means is less than or equal to a second value exists, the confirmation target query is determined from the candidate query. Filtering means to exclude,
The candidate query that the filtering means excludes the confirmation target query from the candidate query or that is aggregated in the calculation target period included in the confirmation evaluation period for the processing by the evaluation period retroactive setting means and the filtering means A query extracting device comprising: a repetitive unit that repeatedly executes the process by the evaluation period retrospective setting unit sequentially in the past until the number of searches becomes zero.   In the confirmation evaluation period set by the evaluation period retroactive setting means, the filtering means is included in the number of searches for the confirmation target query aggregated in the calculation target period included in the confirmation evaluation period, and included in the predetermined evaluation period. 2. The query extraction device according to claim 1, wherein the query to be confirmed is excluded from the candidate queries when the difference from the number of searches of the query to be confirmed calculated at the end of the calculation target period is equal to or smaller than a third value. . A query extraction device that extracts a query in which the number of searches gradually decreases with time in a predetermined evaluation period,
Search log storage means for storing the query and the date and time when the query was used for the search as a search log;
A receiving means for receiving an input of a query to be judged;
Dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period;
Referring to the search log storage means, for the determination target query, for each of the plurality of calculation target periods divided by the division means, the higher the ranking is, the newer the date and time of the search log aggregated in a predetermined time unit. A first order giving means for giving,
With reference to the search log storage unit, a higher rank is assigned to the determination target query as the number of searches aggregated in a predetermined time unit increases for each of the plurality of calculation target periods divided by the division unit. 2 rank granting means;
Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means of the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means; ,
When the rank correlation coefficient calculated by the correlation coefficient calculation means is less than or equal to the first value in all the plurality of calculation target periods, the number of searches for the determination target query is gradually decreased. Extraction means for extracting as a candidate query,
An evaluation period retroactive setting means for retroactively setting the start of the predetermined evaluation period in the past in the calculation target period unit and setting the confirmation evaluation period;
In the confirmation evaluation period set by the evaluation period retrospective setting means, the dividing means, the first rank assigning means, the second rank assigning means, and the correlation are set with the candidate query extracted by the extracting means as a check target query. When the calculation target period in which the rank correlation coefficient calculated by the correlation coefficient calculation means is greater than or equal to a second value exists, the confirmation target query is determined from the candidate query. Filtering means to exclude,
The candidate query that the filtering means excludes the confirmation target query from the candidate query or that is aggregated in the calculation target period included in the confirmation evaluation period for the processing by the evaluation period retroactive setting means and the filtering means A query extracting device comprising: a repetitive unit that repeatedly executes the process by the evaluation period retrospective setting unit sequentially in the past until the number of searches becomes zero. A query extraction method for extracting a query in which the number of searches gradually increases with time in a predetermined evaluation period,
Computer
Search log storage means for storing the query and the date and time when the query was used for the search in association with the search log,
In the computer,
A reception step for receiving an input of a query to be judged;
A division step of equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period;
With reference to the search log storage means, for the determination target query, for each of the plurality of calculation target periods divided in the dividing step, the higher the ranking is, the newer the date and time of the search log aggregated in a predetermined time unit. A first ranking granting step,
Referring to the search log storage means, the determination target query is given a higher rank for each of the plurality of calculation target periods divided in the dividing step as the number of searches aggregated in the predetermined time unit increases. A second ranking step;
A correlation coefficient calculating step for calculating a rank correlation coefficient for the rank given in the second rank assigning step for the rank given in the first rank assigning step for each of the plurality of calculation target periods divided in the dividing step; ,
When the rank correlation coefficient calculated in the correlation coefficient calculation step is equal to or greater than a first value in all the plurality of calculation target periods, the number of searches for the determination target query is gradually increased. An extraction step for extracting as a candidate query,
An evaluation period retroactive setting step of setting the confirmation evaluation period retroactively in the past in the calculation target period unit, the start of the predetermined evaluation period;
In the confirmation evaluation period set in the evaluation period retroactive setting step, the division step, the first rank assigning step, the second rank assigning step, and the correlation are performed using the candidate query extracted in the extraction step as a confirmation target query. When the calculation target period in which the rank correlation coefficient calculated in the correlation coefficient calculation step is less than or equal to a second value exists, the confirmation target query is determined from the candidate query. Filtering steps to exclude,
The candidate query obtained by excluding the confirmation target query from the candidate query in the filtering step in the evaluation period retroactive setting step and the filtering step, or aggregated in the calculation target period included in the confirmation evaluation period A query extraction method for executing a repetitive step of repeatedly performing the process of the evaluation period retroactive setting step sequentially and repeatedly in the past until the number of searches becomes zero. A query extraction device that extracts a query in which the number of searches gradually increases with time in a predetermined evaluation period,
Search log storage means for storing the query and the date and time when the query was used for the search as a search log;
A receiving means for receiving an input of a query to be judged;
Dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period;
Referring to the search log storage means, for the determination target query, for each of the plurality of calculation target periods divided by the division means, the higher the ranking is, the newer the date and time of the search log aggregated in a predetermined time unit. A first order giving means for giving,
With reference to the search log storage unit, a higher rank is assigned to the determination target query as the number of searches aggregated in a predetermined time unit increases for each of the plurality of calculation target periods divided by the division unit. 2 rank granting means;
Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means of the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means; ,
When the rank correlation coefficient calculated by the correlation coefficient calculation means is equal to or greater than a first value in all the plurality of calculation target periods, the number of searches for the determination target query is gradually increased. Extraction means for extracting as a candidate query,
A confirmation evaluation period request accepting means for accepting an input of a predetermined confirmation evaluation period retroactive to the beginning of the predetermined evaluation period;
For the predetermined confirmation evaluation period received by the confirmation evaluation period request receiving means, the dividing means, the first rank assigning means, and the second rank assigning means as the candidate query and confirmation target query extracted by the extracting means. And when the absolute value of the rank correlation coefficient calculated by the correlation coefficient calculation unit is equal to or less than a fourth value, the confirmation target query is determined from the candidate query. A query extracting device comprising filtering means for exclusion. A query extraction device that extracts a query in which the number of searches gradually decreases with time in a predetermined evaluation period,
Search log storage means for storing the query and the date and time when the query was used for the search as a search log;
A receiving means for receiving an input of a query to be judged;
Dividing means for equally dividing the predetermined evaluation period into a plurality of calculation target periods shorter than the predetermined evaluation period;
Referring to the search log storage means, for the determination target query, for each of the plurality of calculation target periods divided by the division means, the higher the ranking is, the newer the date and time of the search log aggregated in a predetermined time unit. A first order giving means for giving,
With reference to the search log storage unit, a higher rank is assigned to the determination target query as the number of searches aggregated in a predetermined time unit increases for each of the plurality of calculation target periods divided by the division unit. 2 rank granting means;
Correlation coefficient calculating means for calculating a rank correlation coefficient for the rank given by the second rank assigning means of the rank given by the first rank assigning means for each of the plurality of calculation target periods divided by the dividing means; ,
When the rank correlation coefficient calculated by the correlation coefficient calculation unit is equal to or less than the first value in all the plurality of calculation target periods, the number of searches for the determination target query is gradually decreased. Extraction means for extracting as a candidate query,
A confirmation evaluation period request accepting means for accepting an input of a predetermined confirmation evaluation period retroactive to the beginning of the predetermined evaluation period;
For the predetermined confirmation evaluation period received by the confirmation evaluation period request receiving means, the dividing means, the first rank assigning means, and the second rank assigning means as the candidate query and confirmation target query extracted by the extracting means. And when the absolute value of the rank correlation coefficient calculated by the correlation coefficient calculation unit is equal to or less than a fourth value, the confirmation target query is determined from the candidate query. A query extracting device comprising filtering means for exclusion.

Images