JP2006260052A - Keyword analysis system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a keyword analysis system correcting input data when loss is generated in input data showing a keyword inputted by a consumer and an input frequency of the keyword. <P>SOLUTION: The input data showing the inputted keyword and the input frequency of the keyword are acquired (step S10), and are stored. The keyword having small fluctuation of the input frequency is set to a stable word from the stored input data (step S11), the input frequency of the stable word is compared (step S13), and it is decided whether a loss decision target day is a loss day or not (step S14). When it is decided that it is the loss day (step S15), a total value of the input frequency of the keyword in each day of four weeks including a week wherein the loss day is present is calculated (step S16), and a correction coefficient is calculated by use of the calculated total value (step S17). By multiplying the input frequency of each the keyword in the loss day by the correction coefficient, the input frequency is corrected (step S18). <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a keyword analysis system that obtains information for marketing support by analyzing a keyword input to a search engine site on a network by a consumer.

  With the spread of information communication equipment, consumers obtain data on products and services using a network such as the Internet. For example, by inputting a product name or the like on a search engine site on the Internet, detailed data regarding the product is obtained. Therefore, a large amount of search keywords are input to the search engine site every day. The search keyword input by the consumer and the data indicating the input frequency of the keyword are valuable data reflecting the consumer's interest, potential customer needs, and the like.

  A keyword analysis system that obtains information for marketing support by analyzing a keyword input by a consumer has been filed by the applicant of the present application (see Patent Document 1). According to the keyword analysis system described in Patent Literature 1, marketing support is performed by determining whether or not the keyword is a rapidly increasing keyword that is rapidly attracting high consumer interest based on fluctuations in the number of keyword inputs. I'm getting information for.

Japanese Patent Laid-Open No. 2003-196301

  By the way, when information for marketing support is obtained by analyzing a keyword for search, data indicating the exact number of times the keyword has been input for searching is required. In other words, if the keyword is not analyzed based on the exact number of inputs, there will be an error in the data being analyzed, and accurate keyword analysis cannot be performed for useful marketing support. Can't get the information.

  However, the search keyword and the data (input data) indicating the number of input of the keyword are missing due to a system failure that occurred in the server for storing the log data indicating the number of search data etc. in the search engine site. May occur. In other words, in the server, when a large amount of data exceeding the server's processing capacity is input, for example, when the server's function is temporarily stopped, or when the server's function is stopped due to a failure of the hardware equipment that constitutes the server. Etc. may occur. In such a case, for example, a keyword input while the server is in a function stop state is in a state where the input count is not counted, the input data is not accurately stored in the server, and the input data is lost. Occurs. The input data during the occurrence of the system failure in the server is data input by a consumer or the like when the failure occurs, and cannot be processed again such as later acquisition. On the other hand, for example, when the input data for each day is accumulated, there is a day (missing day) when the input data is lost due to the occurrence of the system failure, and the continuity of the input data is increased. There is a problem that the keyword analysis using the input data that is obstructed and has continuity cannot be performed accurately.

  The subject of this invention is the keyword which correct | amends input data so that an exact keyword analysis can be performed even if it is a case where the defect | deletion arises in the data which show the keyword inputted by the consumer, and the input frequency of this keyword. To provide an analysis system.

  The keyword analysis system according to claim 1 includes: an input data storage unit that stores an input keyword and the number of times the keyword is input every first predetermined period; and a keyword stored in the input data storage unit. The stable word setting means for setting a predetermined keyword as a stable word and an input for comparing the number of times the stable word is input for each first predetermined period within a second predetermined period including the first predetermined period. As a result of comparing the number of times of input by the number of times comparison means and the number of times of input comparison, the number of times of input of the stable word is compared with the number of times of input in the other first predetermined period included in the second predetermined period. Input missing period determining means for determining the first predetermined period, which has decreased by a predetermined percentage or more, as an input missing period, and a third including the input missing period Total value calculation means for calculating a total value of the number of times each keyword is input for each first predetermined period within a predetermined period, and the total for each first predetermined period calculated by the total value calculation means Correction coefficient calculation means for calculating a correction coefficient for the number of times of input in the input deficiency period using the value, the number of times of input of each keyword input in the input deficiency period, and the correction calculated by the correction coefficient calculation means Correction means for correcting the number of input times of each keyword in the input deficient period using a coefficient.

  According to the keyword analysis system of claim 1, a predetermined keyword is a stable word, that is, a keyword that is empirically known to have a small variation in the number of inputs and that the number of inputs is stable, for example, Set words that are familiar to network users such as the Internet as stable words. Next, other days included in the second predetermined period including the number of times of input of the stable word in any first predetermined period, for example, any one day (missing determination target day) and the missing determination target day, for example, In the three days including the defect determination target date, the number of times the stable word is input on the days before and after the defect determination target date is compared. Next, if the number of stable word inputs on the missing date is less than a predetermined percentage compared to the number of stable words entered on the days before and after the missing date, enter the missing date as the missing date. It is determined as a period (missing date), and a total value of the number of keyword inputs per day within a third predetermined period including the missing input period, for example, four weeks including the missing date is calculated. Then, using the calculated total value, a correction coefficient used when correcting the number of keyword inputs in the input deficit period is calculated, and using the calculated correction coefficient and the number of keyword inputs in the input deficit period, Correct the number of inputs. Accordingly, the number of keyword inputs in the input deficit period is appropriately corrected using the total value of the keyword input counts for each day of the four weeks including the week in which the missing date exists, for example, the days around the input deficit period. Therefore, accurate keyword analysis can be performed even when the input data is deficient due to a system failure or the like in the server.

  The keyword analysis system according to claim 2, wherein the stable word setting unit is configured to perform a fourth predetermined determination based on the number of times the keyword is input every first predetermined period stored in the input data storage unit. A moving average calculating means for calculating a moving average of the number of times of input for each period; and a moving average storing means for storing the moving average calculated by the moving average calculating means, and stored in the moving average storing means A keyword having a moving average variation equal to or less than a predetermined ratio is set as the stable word.

  According to the keyword analysis system of the second aspect, keywords whose moving average fluctuations are not more than a predetermined ratio are set as stable words. In other words, a moving average is calculated every fourth predetermined period, for example, every seven days, to eliminate fluctuations in the number of times of input depending on the day of the week, and keywords whose fluctuations in the calculated moving average are less than a predetermined ratio are set as stable words. can do. Therefore, for example, when past input data is sufficiently accumulated, an appropriate keyword is set as a stable word using the past input data, and it is determined whether or not the deficiency determination target date is an input deficiency period. Can be done appropriately.

  The keyword analysis system according to claim 3, wherein when the plurality of stable words are set by the stable word setting unit, the input missing period determination unit has at least two input times of the stable word. A first predetermined period that decreases by a predetermined ratio or more is determined as the input deficient period.

  According to the keyword analysis system of the third aspect, the first predetermined period in which the number of times of input of at least two or more stable words is decreased by a predetermined ratio or more is determined as the input missing period. That is, there are many cases where the number of times of input of a certain keyword increases abruptly due to media reports or the like, but the number of times of input of a keyword (stable word) in which the number of input is stable rarely decreases rapidly. Furthermore, a sudden decrease in the number of inputs of a plurality of stable words is a situation that can hardly occur when the server is functioning normally. Therefore, by determining the input missing period based on whether or not the number of times of inputting a plurality of stable words is simultaneously decreased by a predetermined ratio or more, it is possible to determine whether or not the input missing period is highly accurate.

  The keyword analysis system according to claim 4, wherein the total value calculation means includes a total value of the number of times of input among the keywords input within the first predetermined period stored in the input data storage means. Threshold value setting means for setting a threshold value for determining a keyword to be excluded when calculating the threshold value is stored in the input data storage means except for the keyword determined to be excluded based on the threshold value set by the threshold value setting means. The total value is calculated by adding together the number of times of input of each of the keywords input within the first predetermined period.

  According to the keyword analysis system of the fourth aspect, the total number of times of input is calculated using keywords other than the excluded keywords based on the set threshold value. For example, as input data, it is assumed that keywords input during a first predetermined period (for example, one day) and the number of input times of each keyword are stored up to the top 10,000 in descending order of input times. In this case, if 101 is set as the threshold value, the total number of times of input is calculated by excluding the top 1 to 100 keywords. In other words, the top 1 to 100 keywords are keywords whose number of inputs increases rapidly due to media reports or the occurrence of events, and are likely to be temporarily focused keywords. Therefore, in order to calculate the total value by excluding keywords that are temporarily attracted by media reports, etc., the total value of the number of times of input is calculated using keywords that are ranked below the threshold set as the threshold. . Therefore, it is possible to calculate the correction coefficient with high accuracy by calculating the total value of the input frequency by eliminating the fluctuation of the total value of the input frequency due to the keyword that is temporarily attracting attention. The number of inputs can be corrected.

  The keyword analysis system according to claim 5, wherein the keyword and the number of times the keyword is input are data that is input to a server by a consumer via electronic means and stored in the server. The method further comprises input data acquisition means for acquiring the keyword and the number of times the keyword is input every first predetermined period.

  According to the keyword analysis system of the fifth aspect, since the keyword input to the server by the consumer and the number of times the keyword is input are acquired every predetermined period, the consumer's interests and needs are accurately determined. The reflected data can be obtained, and if the acquired input data is deficient, the keyword input count is corrected, so appropriate keyword analysis can be performed using accurate data. it can.

  According to the keyword analysis system of the present invention, even if the input data is deficient due to the occurrence of a system failure or the like in the server, the keyword input count is appropriately corrected. It can be performed. Therefore, by performing accurate keyword analysis using appropriately corrected input data, we analyze consumer needs and consumer interests, for example, product sales strategy, sales start time, product purchases, etc. It is possible to obtain useful information for marketing support, such as determining the etc.

  Hereinafter, a keyword analysis system according to an embodiment of the present invention will be described with reference to the drawings. In the following, an example will be described in which a search keyword input by a consumer and the number of times the keyword is input are processed in a search engine site established on the Internet.

  FIG. 1 is a block diagram of the entire system including the keyword analysis system according to the embodiment. The keyword analysis system 2 is connected to a Web server 6 of a company that has established a search engine site on the network via the network 4, and the Web server 6 is consumed by each consumer via the network 4. Connected to a search engine site on a network such as a personal computer, a mobile terminal, or a mobile phone 8a, 8b, 8c.

  As shown in FIG. 1, the keyword analysis system 2 includes a data control unit 20 that controls processing in the keyword analysis system 2, and the data control unit 20 includes a keyword acquired from the Web server 6 and the number of times the keyword is input. Is connected to the input data storage unit 22. The data control unit 20 includes a communication control unit 24 that controls communication with the Web server 6, a display unit 26 that displays keyword analysis results and the like, and a data storage unit 28 that stores keyword analysis results and the like. It is connected.

  The input data storage unit 22 stores input data, that is, data indicating a keyword and the number of times the keyword is input. The input data is data acquired from the Web server 6 via the communication control unit 24 every predetermined period (first predetermined period), for example, every day. Each keyword and the number of times the keyword is input are Is stored in correspondence with the input date. For example, as shown in FIG. 2, “typhoon information” input times “168, 260”, “typhoon” input times “121,013”, and “meteorological agency” input times “85, 881 "and the like. Note that “word A” to “word I” in FIG. 2 are keywords indicating proper nouns and the like.

  The “stable word” shown in FIG. 2 indicates whether or not an arbitrary date is a date on which a loss of input data has occurred, that is, whether an arbitrary day (a date for determining a loss) is a lost date (an input loss period). It is a keyword used when determining whether or not. This “stable word” is a keyword that is empirically known from the past input data that the number of input changes is small, and is generally a word familiar to network users such as the Internet, For example, the name of a specific Web site, the name of a specific search engine, the name of a transportation facility, and the like are applicable. The missing date indicates a date when the keyword input by the consumer and the number of times the keyword is input are not normally recorded due to a system failure or the like occurring in the server 6.

  Next, with reference to the flowchart of FIG. 3, it is determined whether or not the date is a missing date, and when it is determined that the date is a missing date, a process for correcting the number of times the keyword entered on the missing date will be described. . The keyword is input by a consumer who uses a search engine site managed by the server 6. That is, when a consumer accesses a search engine site through the network 4 using the consumer systems 8a, 8b, and 8c and inputs a search keyword, the input search keyword is stored in the Web server. 6 is stored. That is, when the keyword for search input by the consumer is input once, it is counted as the number of times of input, and the keyword and the number of times of input of each keyword are stored in the Web server 6 as input data.

  In the keyword analysis system 2, the input data stored in the Web server 6 is acquired every day (every first predetermined period) via the communication control unit 24 (see step S10, FIG. 3). That is, among the input data stored in the Web server 6, the input data up to the top 100,000 are input from the Web server 6 to the keyword analysis system 2 once a day in the descending order of the number of inputs per day. By transmitting the keyword, the keyword analysis system 2 acquires the keyword and the input count of each keyword. The acquired keywords and the number of times each keyword is input are stored as input data in the input data storage unit 22 (see FIG. 2).

  Next, a stable word is set based on the input data stored in the input data storage unit 22 (step S11). Hereinafter, processing for setting a stable word will be described with reference to the drawings. In the following, a process for setting a stable word will be described by taking as an example a case in which input data for each day from September 1, 2001 to February 1, 2005 is stored in the input data storage unit 22. .

  First, keywords (stable word candidates) that are candidates for stable words are extracted from the input data stored in the input data storage unit 22 (see step S20, FIG. 4). This stable word candidate is extracted by setting a condition such as being a keyword that always shows a rank within the top 100 in a period of one year or more. In the following, a case where four keywords “word 1”, “word 2”, “word 3”, and “word 4” are extracted as stable word candidates will be described as an example.

  Next, a daily moving average is calculated for a fourth predetermined period of stable word candidates extracted in step S20, for example, every 7 days (step S21). That is, since it is known that the keyword inputted by the consumer and the number of times of the input are generally changed depending on the day of the week, in order to eliminate the periodic fluctuation for each day of the week, 7 for each stable word candidate. The daily moving average for each day (7-day daily moving average) is calculated. For example, by calculating the total number of inputs for 7 days from September 25th to October 1st, 2001, and dividing the calculated total value by 7, the daily movement for 7 days on October 1, 2001 Calculate the average. Similarly, the 7-day daily moving average on October 2, 2001, using the number of inputs for the seven days from September 26, 2001 to October 2, 2001, is set to 7 on September 27, 2001 to October 3, 2001. Calculate the 7-day daily moving average on October 3, 2001 using the number of inputs in the day, and calculate the 7-day daily moving average of each stable word candidate for the period from October 1, 2001 to February 1, 2005. calculate. Note that the calculated 7-day daily moving average value of each stable word candidate is stored in the data storage unit 28 in association with the date (the date of the seventh day).

  Next, using the value of the 7-day daily moving average calculated in step S21, the minimum value of the 7-day daily moving average in a period that goes back one year from a predetermined date is obtained (step S22). That is, when the predetermined date is October 1, 2002, the minimum value of the 7-day daily moving average in the period from October 2, 2001 to October 1, 2002 is obtained. Similarly, when the predetermined date is October 2, 2002, the predetermined date is October 3, 2002, and so on, from October 1, 2002 to February 1, 2005. For each stable word candidate, the lowest value of the 7-day daily moving average in a period that goes back one year from the date is obtained. Note that the minimum value of the 7-day daily moving average of each stable word candidate obtained is associated with the date of a predetermined day for each stable word candidate, for example, in a period that goes back one year from October 1, 2002. The minimum value is stored in the data storage unit 28 in correspondence with October 1, 2002.

  Next, using the value of the 7-day daily moving average calculated in step S21, the highest value of the 7-day daily moving average in a period that goes back one year from the predetermined date is obtained (step S23). That is, by performing the same process as the process in step S22 described above, the maximum value of the 7-day daily moving average is obtained for each stable word in the period that goes back one year from each day from October 1, 2002 to February 1, 2005. Ask for candidates. The maximum value of the 7-day daily moving average of each stable word candidate obtained is stored in the data storage unit 28 in association with the date of a predetermined day for each stable word candidate together with the minimum value.

  Next, a stable word determination index is calculated using the lowest value of the 7-day daily moving average calculated in step S22 and the highest value of the 7-day daily moving average calculated in step S23 (step S24). That is, in the data storage unit 28, by dividing the highest value of the moving average for 7 days stored corresponding to the same date for each stable word candidate by the lowest value, each stable word candidate for each day is divided. A stable word determination index is calculated. The calculated stable word determination index of each stable word candidate is stored in the data storage unit 28 in association with the date.

  Next, based on the stable word determination index calculated in step S24, it is determined whether the stable word candidate is suitable for the stable word (step S25). That is, the determination is made based on whether or not the fluctuation of the stable word determination index stored in the data storage unit 28 is within a predetermined range.

  FIG. 5 is a diagram illustrating an example of a graph showing fluctuations in the stable word determination index. FIG. 5 shows a graph showing fluctuations in the stable word determination indexes of the four stable word candidates “word 1” to “word 4”. Here, the larger the value of the stable word determination index, the larger the difference between the minimum value and the maximum value of the moving average for 7 days, indicating that the fluctuation of the number of inputs is large. Therefore, the smaller the value of the stable word determination index, the smaller the difference between the minimum value and the maximum value of the moving average for 7 days, indicating that the fluctuation in the number of inputs is small and stable. Therefore, from the graph shown in FIG. 5, for example, depending on whether or not the value of the stable word determination index is held in advance as a threshold, for example, a value smaller than “2” is held for a predetermined period or more, It is determined whether the stable word candidate is suitable for the stable word.

  For example, the value of the stable word determination index of “word 1” is smaller than the threshold value “2” throughout the period from October 1, 2002 to February 1, 2005. Is determined to be suitable. Also, the value of the stable word determination index of “Word 2” is smaller than the threshold value “2” throughout the period from October 1, 2002 to February 1, 2005, and is the same as “Word 1”. Therefore, it is determined that the number of times of input is small and suitable for a stable word. On the other hand, the stable word determination index value of “Word 3” is smaller than the threshold value “2” from October 1, 2002 to around July 2003, but since July 2003, 2004 The value up to about July is a value equal to or higher than the threshold value “2”, and thereafter, from July 2004 to February 1, 2005, the value is again smaller than the threshold value “2”. Therefore, “word 3” was not suitable for a stable word since July 2003 and until around July 2004, and was not suitable for a stable word. However, at least after July 2004, the fluctuation of the number of inputs was small. It is determined that the keyword can be used as a stable word. Similarly, the value of the stable word determination index of “Word 4” is a threshold value “2” or more from October 1, 2002 to around February 2003, and from February 2003 to around June 2004. The interval is a value smaller than the threshold “2”, and from June 2004 to February 1, 2005, the value is again greater than the threshold “2”. Therefore, “word 4” had a small variation in the number of inputs from February 2003 to around June 2004, but before and after that, the variation in the number of inputs was large. It is determined that it is not suitable.

  Next, based on the determination result in step S25, any keyword in the stable word candidates is set as a stable word (step S26). For example, when “2” is set as the threshold value, “word 1” and “word” in which the value of the stable word determination index always indicates a value smaller than the threshold value “2” based on the determination result in step S25. 2 ”is stored in the data storage unit 28 as a stable word, and the stable word is set.

  Next, the stable word set in step S11 (steps S20 to S26 in the flowchart shown in FIG. 4) is extracted from the input data stored in the input data storage unit 22 (see step S12 and FIG. 3). ).

  Next, a predetermined period (second predetermined period) including a defect determination target date that is a determination target of whether or not it is a defect date, for example, a defect determination target day and days before and after the defect determination target date 3 The number of times the stable word is input on each day of the day is compared (step S13). For example, when the defect determination target date is November 17, 2004, the number of stable word inputs on the day before the defect determination target date (November 16, 2004, hereinafter referred to as “16 days”), The number of times the stable word is input on the determination target date (November 17, 2004, hereinafter referred to as “17th”), and the next day (November 18, 2004, hereinafter referred to as “18th”). .) Are compared with the number of times the stable word is input. Hereinafter, the input count comparison process will be described with reference to FIG.

  FIG. 2 is a diagram illustrating an example of input data on each day from November 16 to 18, 2004, that is, an example of a keyword input by a consumer and an input count of the keyword. In FIG. 2, the top 15 keywords and the number of input times of the keywords are shown in a list format in descending order of the number of input times on each day. Here, there are two stable words, “Stable word 1” and “Stable word 2”, which are in the top 15 in all three days from the 16th to the 18th, as shown in FIG. Therefore, in this case, the number of inputs of “Stable Word 1” on the 16th, the number of inputs of “Stable Word 1” on the 17th, and the number of inputs of “Stable Word 1” on the 18th are respectively compared. Similarly, for “stable word 2”, the number of inputs on the 16th, the number of inputs on the 17th, and the number of inputs on the 18th are compared.

  Here, as shown in FIG. 2, the number of inputs of “Stable Word 1” on the 16th is “105,500”, the number of inputs on the 17th is “28,380”, and the number of inputs on the 18th is “85,353”. When the number of inputs on the 16th is 100%, the number of inputs on the 17th is about 27% and the number of inputs on the 18th is about 81%. In addition, the number of inputs of “Stable Word 2” on the 16th is “49,732”, the number of inputs on the 17th is “14,968”, and the number of inputs on the 18th is “49,890”, which is 16 days. Assuming that the number of inputs in 100 is 100%, the number of inputs on 17th is about 30%, and the number of inputs on 18th is about 100%. Therefore, the number of times of inputting “stable word 1” and “stable word 2” on the 17th is drastically reduced compared to the number of times of inputting “stable word 1” and “stable word 2” on the 16th and 18th. This is shown as a comparison result.

  Next, based on the comparison result of the number of stable word inputs in step S13, it is determined whether or not the missing determination target date is a missing date (step S14). That is, whether or not the number of times the stable word is input on the deficiency determination target day is rapidly decreased by a predetermined rate or more, for example, 10% or more, compared to the surrounding days (for example, the days before and after the deficiency determination target day), or Then, it is determined whether or not the deficiency determination target date is a deficiency date based on whether or not the number of times of input of a plurality of stable words is rapidly decreasing simultaneously on the deficiency determination date. For example, as shown in FIG. 2, the number of times “stable word 1” and “stable word 2” are input on the missing determination target date (17th) is the days before and after the defect determination target date (16th and 18th). Compared to the number of inputs, it is significantly reduced. Accordingly, it is determined that the 17th, which is the date for determination of deficiency, is a deficit date in which deficiency has occurred in the input data.

  If it is determined that the missing date is not a missing date (step S15), it is not necessary to correct the number of times of input, and thus the processing for correcting the number of times of input in the keyword analysis system 2 is terminated.

  On the other hand, when it is determined that the missing date is a missing date (step S15), in order to correct the number of keywords input on the missing date, within a predetermined period (third predetermined period) including the missing date. The total value of the number of keyword inputs per day is calculated (step S16). That is, for a predetermined period including the missing date, for example, every day of four weeks including the week in which November 17, 2004 determined as the missing date exists, specifically, from November 1, 2004 to November 2004. The total value of the number of inputs per day on month 28 is calculated. This total value is calculated by adding the number of times of input of each keyword input on the basis of the input data stored in the input data storage unit 22.

  Here, for example, in the input data storage unit 22, it is assumed that the keywords up to the top 100,000 and the number of input times of the keywords are stored as input data in the descending order of the number of input times of the keywords input to the server 6. . In this case, the total number of times of input is determined by excluding the top 1 to 100 keywords from the input data stored in the input data storage unit 22, and the remaining keywords, that is, 101 to 100,000. It is calculated by adding the number of times each keyword is input.

  The reason why the top 1 to 100th keywords are excluded is to prevent the occurrence of errors. In other words, the top 1 to 100 keywords are those related to things or people that the consumer has suddenly become interested in due to some reason, such as media reports, or those that are intentionally input in large quantities. Or, there is a high possibility that it shows a change (seasonal change) due to an event such as Christmas or Valentine. Accordingly, by calculating the total number of input times by excluding the number of times of inputting the keywords from the top 1 to the 100th, it is possible to calculate the total value by eliminating variations in the number of times of input due to media reports or seasonal fluctuations. Can do.

  Note that keywords to be excluded can be determined by, for example, setting the ranking of keywords to be excluded as a threshold value in advance. Here, the threshold value can be set, for example, by analyzing input data on an arbitrary day in the keyword analysis system 2. That is, as shown in FIG. 6, the number of input times of a keyword in an arbitrary day within the period (the third predetermined period) for which the total value is calculated is as follows. Graph. Then, when the rank is lowered by 10 (increased by 10), the rank corresponding to the point where the decrease in the number of inputs is 1000 or less, that is, the point where the slope of the graph shown in FIG. It can be set as a threshold.

  Next, using the total value of the number of times of input calculated in step S16, a correction coefficient used when correcting the number of times of keyword input on the missing date is calculated (step S17). First, the total number of input times on an arbitrary day, for example, October 1, 2004, is set to “1”, and the number of input times per day from November 1 to 28, 2004 is indexed (see FIG. 7). ). That is, the total value of the number of inputs is indexed by dividing the total value of the number of inputs on each day from November 1st to 28th, 2004 by the total value of the number of inputs on October 1, 2004.

  Next, each week shown in FIG. 7, that is, the week of November 1 to 7, 2004 (first week), the week of November 8 to 14 (second week), and the days of November 15 to 21 The average value (7-day average) of the indexed input count (input count index) in the week (third week) and the week of November 22 to 28 (fourth week) is calculated. That is, using the input frequency index shown in FIG. 7, the 7-day average “1.016” in the first week, the 7-day average “0.971” in the second week, and the seven-day average “0.899” in the third week Then, an average of “0.974” for 7 days in the fourth week is calculated.

  Next, the average value (6 day average) of the number of times of input for 6 days in each week is calculated except for the missing day and the day of the same day of the week as the missing day. That is, when the missing date is 17, since 17th is Wednesday (see FIG. 7), the average of 6 days in each week is calculated using the input number index of the month, Tuesday, and Thursday to Sunday excluding Wednesday. Accordingly, from the input frequency index shown in FIG. 7, the average of 6 days in the first week “1.008”, the average of 6 days in the second week “0.960”, the average of 6 days in the third week “0.993”, An average of “0.968” for the 6th day of the fourth week is calculated.

  Next, for each week other than the week including the missing day (third week), the difference between the 7-day average and the 6-day average is calculated, and the average value of the differences (difference average) is calculated. That is, the difference “0.007” in the first week, the difference “0.011” in the second week, and the difference “0.006” in the fourth week are calculated, and the average value of the calculated differences, that is, the missing date The difference average “0.008” indicating the difference between Wednesday, which is the day of the week, and other days other than Wednesday is calculated. Next, the difference value is added to the 6-day average of the third week to calculate a desired value “1.001” of the 7-day average in the third week.

  Then, the value “1.001” for the 7-day average in the third week and the 6-day average for the third week “7” By substituting “0.993”, the value “1.029” that should be the index of the number of times of missing day input is calculated. Therefore, the correction coefficient “3.12” is calculated by dividing the calculated number-of-inputs index “1.029” that should be the missing date by the actual number-of-inputs index “0.33” of the missing day. The calculated correction coefficient is stored in the data storage unit 28.

  Next, using the calculated correction coefficient, the number of keyword inputs on the missing date is corrected (step S18). That is, the number of inputs on the missing date is corrected by multiplying the number of input times of each keyword on the missing date by the correction coefficient calculated in step S17. For example, the keyword “Typhoon information” input number “54,495” input on November 17, 2004 shown in FIG. 2 is multiplied by the correction coefficient “3.12”, and “ The number of input of “typhoon information” is corrected to “170,024”. Similarly, the number of input times of other keywords input on November 17, 2004 is corrected. Then, using the corrected input data, for example, a keyword analysis process is performed such as specifying a keyword whose number of inputs has suddenly increased or a keyword whose number of inputs has been decreasing.

  According to the keyword analysis system according to this embodiment, it is determined whether or not the missing date is a missing date based on the degree of decrease in the number of times of input of a stable word with a small variation in the number of times, and is determined as a missing date. The number of keyword inputs on a given day is corrected. In other words, it has been empirically known that the number of times the keyword is input increases rapidly due to media reports or the like, but rarely decreases rapidly. Therefore, whether or not the number of times of input of the stable word on the deficiency determination target day is reduced by a predetermined ratio or more with respect to the number of inputs on the peripheral determination day (for example, days before and after the deficiency determination target date), Whether or not there is a deficiency in the input data, that is, the number of keyword inputs, is accurately determined with high accuracy. If it is determined as a deficient date, the number of keyword inputs on the deficient date is corrected. Therefore, even if a system failure or the like occurs and the input data is lost, keyword analysis can be performed using the corrected number of inputs, and trends in consumer interests can be accurately analyzed. Analyze and obtain useful and useful information for marketing support.

  Further, according to the keyword analysis system according to this embodiment, the correction coefficient is calculated using the total value of the number of input times of each keyword input every day within a predetermined period including the missing date, and the calculated correction The number of inputs is corrected using a coefficient. In other words, the input number index is calculated by indexing the number of input times per day using the total value of the number of keyword inputs per day for four weeks including the week in which the missing date exists. Next, the average value of the input frequency index for one week including the missing date (7-day average) and the average value of the input frequency index for 6 days excluding the missing day or the day of the same day of the week (average for 6 days) Is calculated to calculate the correction coefficient by dividing the calculated value by the actual input frequency index on the missing date. Then, the number of inputs on the missing date is corrected using the calculated correction coefficient. Therefore, the number of times of input can be corrected with high accuracy by using the correction coefficient calculated with high accuracy in consideration of the characteristics (day-of-week characteristics) of the missing day of the week. Therefore, accurate keyword analysis can be performed. For example, information useful for planning appropriate sales strategies, such as accurately determining the sales start time, sales period, purchase period, purchase amount, etc. of products, etc. is obtained. be able to.

  In the keyword analysis system described above, keywords whose fluctuations in the past number of inputs are within a predetermined range are set as stable words, but predetermined keywords may be set as stable words in advance. In other words, when past input data is not sufficiently stored, keywords that are known to have little variation in the number of inputs and that are familiar to users of the network such as the Internet are identified as stable words. May be set in advance.

  In the keyword analysis system described above, the moving average is calculated every day for 7 days when setting the stable word, but the daily moving average is not limited to 7 days. That is, since the daily moving average is calculated for 7 days so as to eliminate the periodic fluctuations for each day of the week, if the periodic fluctuations for each day of the week can be eliminated, the daily moving averages such as every 14 days or every 21 days are calculated. It may be.

  Moreover, in the above keyword analysis system, when setting a stable word, the minimum and maximum values of the 7-day daily moving average for one year that goes back from an arbitrary day are obtained. The required period is not limited to one year. That is, the minimum value and the maximum value may be obtained based on a period such as three months or half a year. Further, although the threshold value is “2”, the threshold value is not limited to “2” and can be arbitrarily set.

  In the keyword analysis system described above, a stable word candidate that always shows a value smaller than the threshold is set as the stable word. However, when the value is less than the threshold for a predetermined period, for example, one year or more, It may be determined that the keyword is suitable for a stable word. That is, the condition for determining whether or not the word is a stable word may be arbitrarily set.

  Further, when there are many stable word candidates that show a value less than the threshold value for a predetermined period, the threshold value may be adjusted. That is, when there are many stable word candidates that maintain a value smaller than the threshold “2” for a predetermined period or longer, the threshold value and the length of the predetermined period are set so that an appropriate number of stable words are set. Etc. may be adjusted.

  Further, in the above keyword analysis system, when determining whether or not the deficiency determination target date is a deficient date, the comparison result with the number of times of inputting stable words on the days before and after the deficient date is used. Is not limited to the number of times stable words are input on the preceding and following days. For example, using the comparison result with the stable word input count on the same day of the previous week and the next week considering the periodicity of each week, or the comparison result with the stable word input count in the same period of the previous year considering the annual periodicity, etc. It may be determined whether it is a missing date.

  In the keyword analysis system described above, when calculating the correction coefficient, the case where a holiday or a holiday exists is not considered, but the correction coefficient is calculated considering the case where a holiday or holiday exists. May be. In other words, if there are holidays or holidays in the four weeks including the week in which the missing date exists, the calculation may be performed without the holidays. Further, the 7-day average value or the 6-day average value of the week in which the holiday exists may be corrected, and the correction coefficient may be calculated using the corrected 7-day average and the corrected 6-day average. .

It is a block block diagram of the whole system containing the keyword analysis system which concerns on embodiment of this invention. It is a figure which shows an example of the input data which concerns on embodiment of this invention. It is a flowchart for demonstrating the correction process of the frequency | count of input in the keyword analysis system which concerns on embodiment of this invention. It is a flowchart for demonstrating the setting process of the stable word in the keyword analysis system which concerns on embodiment of this invention. It is a figure for demonstrating the setting process of the stable word which concerns on embodiment of this invention. It is a figure for demonstrating the threshold value of the keyword excluded when calculating the total value of the frequency | count of input which concerns on embodiment of this invention. It is a figure which shows an example of the input frequency index | exponent which concerns on embodiment of this invention.

Explanation of symbols

  2 ... Keyword analysis system, 4 ... Network, 6 ... Web server, 8a, 8b, 8c ... Consumer system, 20 ... Data control unit, 22 ... Input data storage unit, 24... Communication control unit, 26... Display unit, 28.

Claims (5)

Input data storage means for storing the input keyword and the number of input times of the keyword for each first predetermined period;
Among the keywords stored in the input data storage means, stable word setting means for setting a predetermined keyword as a stable word;
An input frequency comparison means for comparing the input frequency of the stable word for each first predetermined period within a second predetermined period including the first predetermined period;
As a result of comparing the number of inputs by the number-of-inputs comparison means, the number of times of input of the stable word is equal to or greater than a predetermined ratio compared to the number of times of input in the other first predetermined period included in the second predetermined period. An input deficiency period determination means for determining the first predetermined period that is decreasing as an input deficiency period;
A total value calculating means for calculating a total value of the number of times of input of each keyword for each of the first predetermined periods within a third predetermined period including the input loss period;
Correction coefficient calculation means for calculating a correction coefficient of the number of times of input in the input deficient period using the total value for each of the first predetermined periods calculated by the total value calculation means;
Correction means for correcting the number of input times of each keyword in the input deficiency period using the number of input times of each keyword input in the input loss period and the correction coefficient calculated by the correction coefficient calculation means. A keyword analysis system characterized by that. The stable word setting means includes
A moving average calculating means for calculating a moving average of the number of times of input for each fourth predetermined period based on the number of times of input of the keyword for each of the first predetermined periods stored in the input data storage means;
Moving average storage means for storing the moving average calculated by the moving average calculation means,
The keyword analysis system according to claim 1, wherein a keyword whose fluctuation in the moving average stored in the moving average storage means is a predetermined ratio or less is set as the stable word. The input deficiency period determining means includes
When a plurality of the stable words are set by the stable word setting means, a first predetermined period in which the number of inputs of at least two or more stable words is decreased by a predetermined ratio or more is determined as the input missing period. The keyword analysis system according to claim 1 or 2, wherein The total value calculating means includes
Threshold setting means for setting a threshold value for determining a keyword to be excluded when calculating the total value of the number of times of input among the keywords input within the first predetermined period stored in the input data storage means. Prepared,
The number of input times of each keyword input within the first predetermined period stored in the input data storage means, excluding keywords determined to be excluded based on the threshold value set by the threshold value setting means. The keyword analysis system according to claim 1, wherein the total value is calculated by adding together. The keyword and the number of times the keyword has been input are data that are input to the server via electronic means by a consumer and stored in the server,
The keyword according to any one of claims 1 to 4, further comprising input data acquisition means for acquiring the keyword and the number of times the keyword has been input from the server for each of the first predetermined periods. Analysis system.

Images