JP5296011B2 - Related burst query extraction apparatus and method - Google Patents

Description

  The present invention relates to a related burst query extraction apparatus and method.

  2. Description of the Related Art Conventionally, a method has been proposed for extracting a query that has been noticed by a plurality of users using the search server, as a search query used in the search server.

  For example, in Patent Document 1, a query used in a search server in a predetermined period is totaled as a first total value for each query, and a query used in the search server in a period before the predetermined period is listed for each query. A method of presenting as a query with a burst of usage frequency, that is, a burst query, calculating the ratio of the first aggregated value to the second aggregated value, calculating the ratio of the first aggregated value relative to the second aggregated value. Yes.

  As described above, in the method proposed in Patent Document 1, since burst queries are presented in descending order of the calculated ratio, the user confirms the presented query and knows a query that is attracting public attention. Can do.

JP 2004-206517 A

  By the way, the search result output by the search server includes a URL (Uniform Resource Locator) of the Web page, a title of the Web page, and a snippet that is an explanatory text of the Web page. In particular, as for snippets, in recent years, rich snippets to which rich information is added in addition to information on conventional snippets have been presented to the user.

  To generate a rich snippet, the search server operator needs to collect information relevant to the query used for the search. Here, a plurality of burst queries described above may occur due to the same cause (for example, a query for a TV drama title or a query for the name of a leading actress of this TV drama). In this case, the search server operator can use this same cause as information relevant to these multiple burst queries and generate a rich snippet common to these multiple burst queries based on this information. There is.

  However, in the method described in Patent Document 1, only burst queries are presented in descending order of the rapid increase rate, so even if a plurality of burst queries are extracted, the possibility that these burst queries are generated due to the same cause is known. I can't. Therefore, the method described in Patent Literature 1 cannot determine whether a rich snippet common to these multiple burst queries can be generated.

  An object of the present invention is to provide a related burst query extraction apparatus and method capable of extracting a combination of a plurality of burst queries that may have occurred due to the same cause.

  (1) Based on the search log storage unit that stores the query used for the search and the time when the query was used in association as a search log, and the search log stored in the search log storage unit, For each query, a totaling unit that counts the number of uses for a certain period, and a rate of change and a number of fluctuations of the number of times used by the totaling unit with respect to the number of uses for a certain period immediately before the certain period Then, for each query, a fluctuation score calculation means for calculating a fluctuation score for each fixed period, and a query for which the fluctuation score for each fixed period for each query calculated by the fluctuation score calculation means exceeds a predetermined value Are extracted as burst queries, which are queries whose usage frequency is rapidly increasing, and the burst query extraction means An adjustment means for adjusting the scale of the fluctuation score for the extracted burst query, and a transition of the fluctuation score after the scale adjustment by the adjustment means by analyzing a time-series correlation of the fluctuation score for each predetermined period Related burst query extraction comprising: query set extraction means for extracting a combination of a plurality of burst queries similar to each other as a burst query set; and output means for outputting the burst query set extracted by the query set extraction means apparatus.

  The related burst query extraction device of (1) uses a query set extraction means to analyze a combination of a plurality of burst queries that are determined to have similar transitions of fluctuation scores by analyzing a time series of correlations of fluctuation scores. Extract as Subsequently, the related burst query extraction device outputs the burst query set extracted by the query set extraction means by the output means.

  Therefore, the related burst query extraction device of (1) can extract a plurality of burst queries having similar transition scores as a combination of a plurality of burst queries that may have burst due to the same cause, These burst queries can be presented to the user of this related burst query extraction device as candidate information that can share a rich snippet.

  (2) The burst query extraction means has a frequency of use of a query in which the fluctuation score for each of the certain periods calculated by the fluctuation score calculation means exceeds a predetermined value in the latest certain period. The adjustment means is retroactive from the latest fixed period and the fluctuation score exceeds the predetermined value for the burst query extracted by the burst query extraction means. The related burst query extraction device according to (1), wherein scale adjustment of the variation score of the continuous period is performed.

  The related burst query extraction device of (2) extracts a query whose fluctuation score exceeds a predetermined value in the latest fixed period as a burst query that is a rapidly increasing use frequency, and changes retroactively from the latest fixed period. The scale adjustment is performed for the fluctuation score for a continuous period in which the score exceeds a predetermined value. Therefore, since the related burst query extraction device is a target for extracting a burst query based on the latest fluctuation score and adjusting the scale of the fluctuation score, the user can share the rich snippet based on the latest information. Can be recognized.

  (3) The variation score calculation means uses, as a normalized number, a value obtained by dividing the number of times of use of all queries in a certain period immediately before the certain period by the number of uses of all queries in the certain period. Calculate and multiply the fluctuation count and fluctuation count for the number of uses for the fixed period immediately before the fixed period, and the number of uses for the fixed period normalized by multiplying the normalized number for each fixed period. Then, the related burst query extraction device according to (1) or (2), wherein the fluctuation score for each certain period is calculated for each query.

  The related burst query extraction device of (3) calculates a normalized number for each fixed period, and multiplies the normalized number for each fixed period to normalize the number of times of use for the fixed period immediately before the fixed period. Multiplying the number of uses for a certain period by the rate of change and the number of fluctuations to calculate the fluctuation score for each fixed period for each query, so calculate the fluctuation score when the total number of uses for each fixed period is unchanged. it can. As a result, the related burst query extraction device can calculate a variation score based on the number of times of use of the normalized query in consideration of the influence on the variation score due to the variation in the total number of uses in each of a plurality of fixed periods. .

  (4) The query set extraction unit includes the bursts whose difference in the variation scores is equal to or larger than a predetermined value among a plurality of combinations of the burst queries having similar transitions of the variation scores after the scale adjustment by the adjustment unit. The related burst query extraction device according to any one of (1) to (3), wherein a combination of burst queries is extracted as the burst query set, excluding a query combination.

  The related burst query extraction device of (4) extracts burst query combinations with similar transitions of the variation score after the scale adjustment, except for the combination of burst queries in which the difference of the variation scores is a predetermined value or more. As a result, the related burst query extraction device further selects only combinations of burst queries whose difference in variation score is less than the predetermined value in addition to the variation of the variation score being similar, as a query set. Can be extracted.

  (5) A method executed by a burst query extraction device including a search log storage unit that associates a query used for a search with a time when the query is used and stores it as a search log, the search log storage unit A totaling step for totaling the number of uses for a certain period for each query based on the search log stored in the query, and a certain period immediately before the certain period of the number of uses counted in the aggregating step A variation score calculating step for calculating a variation score for each fixed period for each query based on a variation rate and a variation number with respect to the number of times of use, and the constant for each query calculated in the variation score calculation step A burst query that is a query whose usage frequency is rapidly increasing and a query in which the fluctuation score for each period exceeds a predetermined value; Extracting a burst query, adjusting the scale of the variation score for the burst query extracted in the burst query extraction step, and analyzing the time-series correlation of the variation score for each predetermined period A query set extraction step for extracting a combination of a plurality of burst queries having similar transitions of the variation score after the scale adjustment in the adjustment step as a burst query set, and the burst extracted in the query set extraction step An output step of outputting the query set.

  According to the method as described in (5), there exists an effect similar to the apparatus of (1).

  According to the present invention, it is possible to extract a combination of a plurality of burst queries that may have occurred due to the same cause.

It is a figure which shows the function structure of the related burst query extraction apparatus which concerns on this embodiment. It is a figure which shows an example of the search log table stored in search log DB which concerns on this embodiment. It is a figure which shows an example of the result totaled by the total part based on the search log memorize | stored in search log DB which concerns on this embodiment. It is a figure which shows an example of the fluctuation score table stored in fluctuation score DB which concerns on this embodiment. It is a figure which shows an example of the adjusted fluctuation score table stored in the adjusted fluctuation score DB which concerns on this embodiment. It is a figure which shows an example of the fluctuation score for every fixed period in the predetermined period of the some burst query which concerns on this embodiment. It is a figure which shows an example of the difference of the adjusted fluctuation score for every fixed period in the predetermined period of the some burst query which concerns on this embodiment. It is a flowchart which shows the process in the control part of the related burst query extraction apparatus which concerns on this embodiment. It is a figure which shows another example of the difference of the adjusted fluctuation score for every fixed period in the predetermined period of the some burst query which concerns on this embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[Function configuration]
FIG. 1 is a diagram showing a functional configuration of a related burst query extraction apparatus 1 according to the present embodiment. The related burst query extraction device 1 is a server that extracts a combination of a plurality of burst queries that may have a sudden increase in frequency (burst) due to the same cause.

  This embodiment is applied to a computer (related burst query extraction device 1) 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.

  In addition to the CPU as the control unit 10, the hardware includes a storage unit 20, a communication unit, a display unit, and an input unit. Examples of the storage unit 20 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 20 and appropriately executed and referenced by the control unit 10. Further, the computer program and data can be distributed via a communication line, and can also be recorded and distributed on a computer-readable medium such as a CD-ROM.

  The control unit 10 of the related burst query extraction apparatus 1 includes a totaling unit 11 (totaling unit), a fluctuation score calculation unit 12 (fluctuation score calculation unit), a burst query extraction unit 13 (burst query extraction unit), and an adjustment unit 14. (Adjustment means), a query set extraction section 15 (query set extraction means), and an output section 16 (output means). The storage unit 20 of the related burst query extraction device 1 includes a search log DB (database) 21 (search log storage means), a fluctuation score DB 22, and an adjusted fluctuation score DB 23.

  Based on the search log stored in the search log DB 21, the totaling unit 11 totals the number of uses for each fixed period (for example, every 15 minutes).

  Here, the search log DB 21 stores a query used for search in a search service provided on the Web and a time when this query is used in association with each other as a search log. The search service may be implemented, for example, in a search server (not shown) that is communicably connected to the related burst query extraction device 1 or may be implemented in the related burst query extraction device 1.

FIG. 2 is a diagram illustrating an example of a search log table stored in the search log DB 21 according to the present embodiment.
In the search log table, the query used for the search in the search service and the time when the query was used for the search are associated and stored as a search log.

  FIG. 3 shows only the search logs stored in the search log DB 21 according to the present embodiment (“◎ Surin”, “× Emperor ▽ child”, “○ River △ flower”, “□ Yama ☆ Akira”). It is a figure which shows an example of the result totaled by the total part 11 based on Suppose). Here, the number of uses for a certain period (15 minutes) in a predetermined period (2010/5/6 17:00 to 18:44) of a plurality of queries is shown. It can be confirmed that it is difficult to grasp a similar query from the transition of fluctuations from the number of times of use for a certain period (15 minutes) shown in FIG.

  Returning to FIG. 1, the fluctuation score calculation unit 12 changes, for each query, for each query based on the rate of change and the number of fluctuations of the number of uses counted by the counting unit 11 with respect to the number of uses for the immediately preceding fixed period. Calculate the score.

  Here, the variation score calculation unit 12 may calculate the variation score after normalizing the aggregated number of times of use in order to eliminate the influence on the variation score due to variations in the number of uses of all queries. More specifically, the variation score calculation unit 12 obtains a value obtained by dividing the number of uses of all queries in a certain period immediately before the certain period by the number of uses of all queries in the certain period for each certain period. Is calculated as a normalized number. Subsequently, the variation score calculation unit 12 calculates the variation rate and variation of the number of times of use in a certain period normalized by multiplying the number of normalizations for each certain period with respect to the number of times of use in a certain period immediately before the certain period. Multiply the number to calculate a variation score for each fixed period for each query.

That is, the fluctuation score calculation unit 12 normalizes the number of uses in each of a plurality of fixed periods, and for each query based on the fluctuation rate and the number of fluctuations of the normalized number of uses with respect to the number of uses immediately before the fixed period. , Calculate the fluctuation score for each fixed period.
Subsequently, the variation score calculation unit 12 stores the calculated variation score in the variation score DB 22.
Note that when the variation rate is less than 1, the variation score calculation unit 12 may calculate the variation score using the reciprocal of the variation rate.

FIG. 4 is a diagram illustrating an example of a variation score table stored in the variation score DB 22 according to the present embodiment.
In the variation score table, a query, a period during which the query is used (a certain period), and a variation score calculated in this period are stored in association with each other.

  Returning to FIG. 1, the burst query extraction unit 13 uses a query whose frequency of use exceeds a predetermined value for each query calculated by the variation score calculation unit 12 and whose variation score for each fixed period is stored in the variation score DB 22. It is extracted as a burst query that is a rapidly increasing query. Here, the burst query extraction unit 13 may extract a query that exceeds a predetermined value in the latest fixed period as a burst query that is a query whose use frequency is rapidly increasing.

The adjustment unit 14 refers to the variation score DB 22 for the burst query extracted by the burst query extraction unit 13 and adjusts the scale of the variation score. Here, the adjustment unit 14 may adjust the fluctuation score of a continuous fixed period in which the fluctuation score exceeds a predetermined value retroactively from the latest fixed period. The adjustment unit 14 calculates the logarithm of the variation score to be adjusted. Subsequently, the adjustment unit 14 adjusts the scale of the variation score by using the logarithm of the variation score as the adjusted variation score. Specifically, when the variation score is 1 or more, the adjustment unit 14 calculates the logarithm of the variation score and sets the calculated value as the adjusted variation score. The adjustment unit 14 sets the adjusted variation score to 0 when the variation score is greater than −1 and less than 1. Moreover, the adjustment part 14 calculates the logarithm of the absolute value of a fluctuation score, when the fluctuation score is -1 or less, and the adjusted fluctuation score as a fluctuation score after scale adjustment is obtained by multiplying the calculated value by -1. And In the present embodiment, the adjustment unit 14 calculates the natural logarithm, but is not limited thereto, and the logarithm value may be calculated using a logarithm different from the natural logarithm such as a common logarithm.
Subsequently, the adjustment unit 14 stores the adjusted variation score in the adjusted variation score DB 23.

FIG. 5 is a diagram illustrating an example of the adjusted variation score table stored in the adjusted variation score DB 23 according to the present embodiment.
In the adjusted variation score table, a burst query, a period during which the burst query is used (a fixed period), and an adjusted variation score calculated in this period are stored in association with each other.

  Returning to FIG. 1, the query set extraction unit 15 has a plurality of similar changes in the variation score adjusted by the adjustment unit 14 stored in the adjusted variation score DB 23 by analyzing the time series of the variation score for each fixed period. Are extracted as a burst query set.

  Here, in the extraction of the burst query set, the query set extraction unit 15 determines that the difference in the variation score is the first among the combinations of the plurality of burst queries with similar transitions of the variation score after the scale adjustment by the adjustment unit 14. A combination of burst queries may be extracted as a burst query set, except for combinations of burst queries that are equal to or greater than a value (predetermined value).

The function of the query set extraction unit 15 will be described in detail with reference to FIGS.
FIG. 6 shows a predetermined period (2010/5/6) of a plurality of burst queries (“◎ Surin”, “× Emperor ▽ child”, “○ River △ flower”, “□ Yama ☆ Akira”) according to the present embodiment. 17:15 to 18:44) is a diagram illustrating an example of a variation score for each fixed period (15 minutes). 6A is a diagram illustrating an example of a variation score before adjustment by the adjustment unit 14, and FIG. 6B is a diagram illustrating an example of a variation score after adjustment. In the fluctuation score before scale adjustment shown in FIG. 6 (1), it is difficult to grasp the transition of the time series fluctuation of each burst query, and it is also difficult to grasp the burst query having the similar fluctuation transition. It is. On the other hand, in the adjusted fluctuation score shown in FIG. 6 (2), the scale adjustment is performed, so that it is easy to grasp the transition of the time series fluctuation of each burst query. It is also easy to grasp burst queries with similar transitions. That is, by performing the scale adjustment as shown in FIG. 6B, the query set extraction unit 15 can extract a combination of a plurality of burst queries that may have occurred due to the same cause.

  FIG. 7 illustrates a predetermined period (2010/5/6) of a plurality of burst queries according to the present embodiment (“◎ Surin”, “× Emperor ▽ child”, “○ River △ flower”, “□ Yama ☆ Akira”). 17:15 to 18:44) is a diagram illustrating an example of the difference in the adjusted variation score after every fixed period (15 minutes). That is, FIG. 7A is a diagram illustrating an example of a table of differences in adjusted variation scores, and FIG. 7B is a diagram illustrating an example of a graph of differences in adjusted variation scores.

  The query set extraction unit 15 calculates the difference between the adjusted fluctuation scores for the same fixed period for the adjusted fluctuation scores for each fixed period of the plurality of burst queries stored in the adjusted fluctuation score DB 23.

  Subsequently, the query set extraction unit 15 extracts a combination of burst queries in which the time-series transition of the difference calculated every certain period falls within the range of the first value (for example, “0.5”). In the example illustrated in FIG. 7, the query set extraction unit 15 extracts a combination of “◎ Surin” and “□ Yama ☆ Akira” and a combination of “○ River Δ Hana” and “× Emperor”.

  Here, the query set extraction unit 15 excludes a combination of burst queries in which the difference calculated for each fixed period of the extracted burst queries is a second value (for example, “3”) or more, for each fixed period. A combination of burst queries in which the time-series transition of the difference calculated in (1) falls within the range of the first value (for example, “0.5”) may be extracted. In the example shown in FIG. 7, the query set extraction unit 15 uses the difference itself for the combination of “◎ Surin” and “□ Yama ☆ Akira”, although the time-series transition of the difference is within the range of “0.5”. Is excluded because it is greater than or equal to “3”, and as a result, only a combination of “○ River Δ Hana” and “× Emperor” is extracted.

  Subsequently, the query set extraction unit 15 may use the extracted burst query combination as a burst query combination that is determined to have similar transitions of the variation score, and the combination may be burst for the same cause. A burst query set. In the example shown in FIG. 7, a combination of “○ river Δ flower” and “× emperor ▽ child” is a burst query set.

  The output unit 16 outputs the burst query set extracted by the query set extraction unit 15. Specifically, the output unit 16 outputs a burst query set to a display unit (not shown) of the related burst query extraction device 1.

[Processing flow]
FIG. 8 is a flowchart showing processing in the control unit 10 of the related burst query extraction device 1 according to the present embodiment.

  In step S <b> 1 (aggregation step), the control unit 10 (aggregation unit 11) aggregates the number of uses for a certain period for each query based on the search log stored in the search log DB 21.

  In step S2 (fluctuation score calculation step), the control unit 10 (variation score calculation unit 12) divides the number of uses of all queries in a certain period immediately before the certain period by the number of uses of all queries in the certain period. The normalization number for every fixed period is calculated.

  In step S3 (fluctuation score calculation step), the control unit 10 (variation score calculation unit 12) sets the number of times of use in a certain period normalized by multiplying the number of times used in step S1 by the normalized number. The fluctuation score for each fixed period is calculated for each query by multiplying the fluctuation rate and the fluctuation number for the number of times of use in the fixed period immediately before the fixed period. The control unit 10 (variation score calculation unit 12) stores the calculated variation score for each fixed period in the variation score DB 22.

  In step S4 (burst query extraction step), the control unit 10 (burst query extraction unit 13) stores a predetermined value for the variation score for each query calculated in step S3 and stored in the variation score DB 22. Exceeding queries are extracted as burst queries, which are queries that are rapidly increasing in use frequency. Here, the control part 10 (burst query extraction part 13) may extract the query corresponding to this fluctuation score as a burst query, when exceeding a predetermined value in the newest fixed period.

  In step S5 (adjustment step), the control unit 10 (adjustment unit 14) adjusts the scale of the variation score stored in the variation score DB 22 for the burst query extracted in step S4. Here, the control part 10 (adjustment part 14) may adjust the fluctuation score of the continuous fixed period retroactively from the newest fixed period and the fluctuation score exceeding predetermined value. The control unit 10 (adjustment unit 14) stores the adjusted variation score in the adjusted variation score DB 23.

  In step S6 (query set extraction step), the control unit 10 (query set extraction unit 15) is similar in the transition of the variation score after the scale adjustment in the adjustment step by analyzing the time-series correlation of the variation score for each fixed period. A combination of a plurality of burst queries is extracted as a burst query set.

  In step S7 (output step), the control unit 10 (output unit 16) outputs the burst query set extracted in step S6.

  As described above, according to the present embodiment, the related burst query extraction device 1 is determined by the query set extraction unit 15 to determine that the transition of the variation score is similar by analyzing the time series of the correlation of the variation score. Are extracted as a burst query set. Subsequently, the related burst query extraction device 1 outputs the burst query set extracted by the query set extraction unit 15 by the output unit 16.

  Therefore, the related burst query extraction device 1 can extract a plurality of burst queries having similar transition scores as a combination of a plurality of burst queries that may have burst due to the same cause. The query can be presented to the user of the related burst query extraction device 1 as candidate information that can share the rich snippet.

  Further, the related burst query extraction device 1 calculates a normalization number for each fixed period, and multiplies the normalization number for each fixed period to normalize the number of times of use for the fixed period before the fixed period. By multiplying the number of uses for the period by the rate of change and the number of fluctuations, the change score for each fixed period is calculated for each query, so the change score can be calculated when the total number of uses for each fixed period has not changed. . Thereby, the related burst query extraction device 1 can calculate the variation score based on the normalized number of times of use of the query in consideration of the influence on the variation score due to the variation of the total number of times of use for a plurality of fixed periods.

  Further, the related burst query extraction device 1 extracts a plurality of burst queries having similar transitions of the adjusted fluctuation score except for a combination of burst queries in which the difference of the adjusted fluctuation score is a predetermined value or more. Thereby, the related burst query extraction device 1 further selects only the combinations of burst queries whose difference in the variation score is less than the predetermined value in addition to the variation of the variation score being similar, and the query set Can be extracted as

  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.

  For example, the query set extraction unit 15 extracts, as a burst query set, a combination of a plurality of burst queries with similar transitions of the fluctuation score after adjustment by the adjustment unit 14, but is not limited thereto. For example, the query set extraction unit 15 may extract a combination of burst queries common to burst queries common to a plurality of burst query sets as a burst query set.

  In the above description, the transition of the adjusted fluctuation scores of the two burst queries is similar, but the query set extraction unit 15 can also extract three or more burst queries as burst query sets. For example, when there are three burst queries A, B, and C, the query set extraction unit 15 only applies when the transition of the adjusted variation score is similar for all combinations of A and B, A and C, and B and C. These three queries A, B, and C may be extracted as burst query sets, or even if A and B, A and C are similar, and B and C are not similar, The queries A, B, and C may be extracted as a burst query set. In this way, it is possible to consider the possibility of bursting due to the same cause via the burst query A.

  FIG. 9 shows a predetermined period (2010/5/6 17: 15-18: 44) of a plurality of burst queries (“◎ Surin”, “× Emperor ▽ child”, “○ River Δ flower”) according to the present embodiment. It is a figure which shows another example of the difference of the fluctuation score after adjustment for every fixed period (15 minutes). In the example shown in FIG. 9, the burst query combination in which the time-series transition of the difference calculated every certain period falls within the range of the first value (for example, “0.5”) is “◎ Surin”. The query set extraction unit 15 uses the combination of “○ River △ Hana” and “○ River △ Hana” and “× Teiji ▽ Child”. And “◎ Surin” may be extracted as a burst query set.

DESCRIPTION OF SYMBOLS 1 Related burst query extraction apparatus 10 Control part 11 Aggregation part 12 Fluctuation score calculation part 13 Burst query extraction part 14 Adjustment part 15 Query set extraction part 16 Output part 20 Storage part 21 Search log DB
22 Fluctuation score DB
23 Adjusted Fluctuation Score DB

Claims (4)

Search log storage means for associating the query used for the search with the time when the query was used and storing it as a search log;
Based on the search log stored in the search log storage unit, for each of the queries, a totaling unit that counts the number of uses for a certain period;
A variation that calculates a variation score for each fixed period for each query based on a variation rate and a variation number with respect to the number of uses for a certain period immediately before the certain period of the number of times counted by the aggregation unit. A score calculation means;
A burst that extracts, as a burst query that is a query whose usage frequency is rapidly increasing, a query in which the variation score for each certain period of each query calculated by the variation score calculating unit exceeds a predetermined value in the latest certain period Query extraction means;
For each of the burst queries extracted by the burst query extraction means, a scale of each variation score calculated in each fixed period from a first fixed period retroactive to the latest fixed period from the latest fixed period Adjusting means for adjusting; and
Query set extraction means for extracting, as a burst query set, a combination of a plurality of the burst queries having similar transitions of the scaled variation score between the first predetermined period and the latest fixed period ;
An associated burst query extraction device comprising: output means for outputting the burst query set extracted by the query set extraction means. The variation score calculating means includes
For each fixed period, the value obtained by dividing the number of uses of all queries in the fixed period immediately before the fixed period by the number of uses of all queries in the fixed period is calculated as a normalized number,
Multiplying the number of uses for the certain period normalized by multiplying the normalized number for each certain period, the rate of change and the number of fluctuations for the number of uses for the certain period immediately before the certain period, and The related burst query extraction device according to claim 1 , wherein the fluctuation score for each fixed period is calculated for each query. The query set extraction unit includes a combination of burst queries in which a difference between the variation scores is equal to or greater than a predetermined value among a plurality of combinations of the burst queries whose transition scores after the scale adjustment by the adjustment unit are similar. except for the associated burst query extractor device according to a combination of the burst query to claim 1 or 2 is extracted as the burst query set. A method executed by a burst query extraction device comprising search log storage means for associating and storing a query used for a search and a time when the query was used as a search log,
Based on the search log stored in the search log storage means, for each query, a totaling step of totalizing the number of uses for a certain period;
A variation that calculates a variation score for each fixed period for each query based on the variation rate and the number of variations of the number of uses counted in the aggregation step with respect to the number of uses for a certain period immediately before the certain period. A score calculating step;
A burst query that extracts a query in which the variation score for each certain period of each query calculated in the variation score calculation step exceeds a predetermined value in the latest certain period as a burst query that is a query whose use frequency is rapidly increasing An extraction step;
For each of the burst queries extracted in the burst query extraction step, the scale of each variation score calculated in each fixed period from the first fixed period retroactive to the latest fixed period from the latest fixed period An adjustment step for making adjustments;
A query set extraction step of extracting a plurality of combinations of the burst queries having similar transitions of the scaled variation score between the first predetermined period and the latest fixed period as a burst query set;
An output step of outputting the burst query set extracted in the query set extraction step.

Images