JP2000194745A - Trend evaluating device and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a trend evaluating device which can estimate a trend with improved trend evaluating accuracy by calculating the relative value from the mean value of time series data set in a standard range and the mean time of time series data set in an evaluating object period. SOLUTION: A standard range deciding means 105 decides a standard range based on the inputted evaluating object time. A mean value calculating means 103 calculates the mean value of occurrence frequency of every evaluating object word in an evaluating object period among the time series data which are accumulated in a time series data accumulator 102. A mean value calculating means 104 calculates the mean value of the evaluating object words which are inputted by an input means 101 in the standard range decided by the means 105 among the time series data accumulated in the accumulator 102 and defines this calculated mean value as the reference value of relative occurrence frequency. Then an output means 108 outputs the words accumulated in a trend accumulator 107 as the results.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a trend evaluation apparatus and a trend evaluation method, and more particularly to a trend evaluation apparatus and a trend evaluation method capable of performing trend evaluation and extraction specifying a time range and predicting a trend.

[0002]

BACKGROUND OF THE INVENTION Trends are sensitive to the times,
Information or objects that increase the number of users or the number of recognized people.

[0003] In recent years, new product planning and R & D competition have intensified, and it has become important to grasp information and things that are trending in the world. On the other hand, the period during which an object or information becomes a trend is decreasing year by year, and therefore, it is becoming increasingly important to investigate and analyze the trend quickly and accurately. Also, if it is possible to predict what kind of information and things will become trends based on past and present trends, it will be helpful for new product planning and R & D.

[0004]

2. Description of the Related Art As a technique related to trend evaluation, conventionally, for example, a search request sentence of a user in a text search apparatus is accumulated as a search log, and the appearance frequency of a word in the search log is counted at regular intervals. And techniques for ranking are known.

[0005] However, if only words having a high frequency of appearance are ranked, words often used as search words (for example,
“Free”, “Image”, “Photo” etc.) always occupy the top, and it is hard to say that it reflects the times.

[0006] For example, Japanese Patent Application Laid-Open No. 6-11835 proposes a method that enables a search for a word or information that has become a trend during a specific period or field. This is shown in FIG.
As shown in FIG. 8, with respect to the time-series data of the appearance frequency of the word, an average value Px of the appearance frequency in the evaluation target period and an average value P0 of the appearance frequency in the entire time range of the time-series data are obtained.
The trend is evaluated using the ratio Px / P0.

Here, the evaluation target period is the time shown in FIG.
The period is from T00 to T01, and a specific time (1 point) T00 = T01 can be specified.

[0008] The trend evaluation apparatus described in Japanese Patent Application Laid-Open No. 6-11835 will be further described with reference to FIG. This trend evaluation method includes input means 101 such as a keyboard, a time-series data accumulator 102, an average value calculation means 103 in the evaluation target period, an average value calculation means 120 in the entire time range, and a relative appearance frequency calculation. It comprises means 106, trend word accumulator 107, and output means 108.

The input unit 101 is a unit for inputting various conditions such as a period to be evaluated, a word to be evaluated, and a threshold value of a trend. In the time-series data storage 102,
Time series data of the appearance frequency of each word is stored.

The average value calculation means 103 for the evaluation period is
On the basis of the evaluation target word and the evaluation target period input by the input means 101, an average value of the time-series data stored in the time-series data storage 102 in the evaluation target period is obtained.

The average value calculating means 120 over the entire time range calculates the word to be evaluated inputted by the input means 101,
An average value of the time-series data in the time-series data storage 102 over the entire time range is obtained.

The relative appearance frequency calculating means 103 calculates the average value over the entire time range calculated by the average value calculating means 120 over the entire time range and the evaluation target calculated by the average value calculating means 103 during the evaluation period. The ratio with the average time in the period is calculated, and the ratio exceeding the threshold is stored in the trend word storage 107.

The trend word accumulator 107 accumulates the relative appearance frequency obtained by the relative appearance frequency calculation means 106, the evaluation target period, and the word.

The output means 108 outputs the words stored in the trend storage 012 so that the words can be referred to by a user or an application system.

FIG. 17 is a flowchart showing the processing flow of the conventional method. The operation of the conventional device will be described with reference to FIGS.

In order to evaluate the trend of a certain word W1 in a certain evaluation period, the word W1 and the evaluation period are input from the input means 101 as an end condition (step S1).
1).

Next, the average value Px in the evaluation target period is calculated by the average value calculation unit 103 in the evaluation target period from the time-series data of the appearance frequency of the word W1 stored in the time-series data storage unit 102. (Step S12).

On the other hand, the average value P0 of the appearance frequency of the word W1 stored in the time-series data storage 102 in the entire time range.
Is calculated by the average value calculation means 120 over the entire time range (step S13).

Average value of frequency of occurrence of word W1 over the entire time range
Ratio Px of average value Px of appearance frequency in evaluation period to P0
/ P0 is calculated by the relative appearance frequency calculation means 106 (step S14).

The larger the value of Px / P0 is, the higher the trend is. The relative appearance frequency calculation means 106 determines the words exceeding a predetermined threshold, the evaluation target period, and the relative appearance period. The frequency is stored in the trend word storage 107.

If the number of words to be evaluated is one and the number of periods to be evaluated is one (step S15), the output means 108 outputs the words stored in the trend word storage 107 (step S15).
16).

According to the above-described method, it is possible to evaluate not only a general word having a high appearance frequency but also a trend word whose appearance frequency changes in accordance with the times.

If the end condition is input to the input means 101 as a specific evaluation target period and all words, the trend evaluation in the specific evaluation target period is repeatedly performed for all words (step S15), thereby outputting the output means 108. In, it is also possible to extract a word that has become a trend in a specific evaluation target period.

Further, the input means 101 inputs the word W1 and a plurality of evaluation target periods as end conditions, and repeatedly calculates the relative appearance frequency of the word W1 in the plurality of evaluation target periods (step S15), whereby the output means 108 In, it is also possible to search for a period in which the word W1 has become a trend.

[0025]

However, the trend evaluation technique described in Japanese Patent Application Laid-Open No.
It includes the following problems.

The first problem is to correctly evaluate the trend characteristic of a tendency to “periodically fluctuate in frequency” as shown at Tx in FIG. 19A showing an example of time-series data of the word appearance frequency. It is difficult. For example, the appearance frequency at Tx and the average value over the entire time range in FIG. 19A match the appearance frequency at Ty and the average value over the entire time range in FIG. 19B, respectively.

For this reason, although FIG. 19A clearly shows a tendency to increase in Tx, and FIG. 19B clearly shows a tendency to decrease, it cannot be distinguished. The reason is that the time range for obtaining the relative appearance frequency is not taken into account.

The second problem is that, in the case of "gradually penetrating" as shown at the time T2 in FIG. 20, the average value of the whole gradually increases. That is the tendency. The reason is that the period in which the time change is stable is not set as the reference period.

Next, a third problem will be described with reference to FIG. The appearance frequency and the average value of the entirety of the word Wx in FIG. 21A at the point of time Tx and the average frequency of the word Wy in FIG. .

It is clear that the word Wx fluctuates from normal and shows a noise-like change, and it is obvious that the word Wx only appears at the time of Tx by chance. And Wy have the same trend. This is because stability is not taken into account when evaluating the trend.

The fourth problem is that the present or past trend of a certain word can be evaluated, but at what time in the future the word will become the trend, or what word will become the trend in the future. Is difficult to predict. This is because the above-described conventional device does not include any prediction means.

Accordingly, the present invention has been made in view of the above problems, and an object of the present invention is to provide a trend evaluation apparatus which can improve the accuracy of trend evaluation and can make prediction.

More specifically, a first object of the present invention is to provide a trend evaluation device capable of correctly evaluating a tendency that the frequency of appearance periodically rises and falls.

[0034] A second object of the present invention is to provide a trend evaluation apparatus capable of correctly evaluating a tendency to gradually increase.

A third object of the present invention is to provide a trend evaluation device capable of distinguishing a noise from a true trend.

A fourth object of the present invention is to provide a trend evaluation device capable of predicting a trend.

[0037]

A trend evaluation apparatus according to the present invention that achieves the above object has a standard range determining means for limiting a range for obtaining a reference value of a relative frequency of appearance based on an input evaluation target period. 1), and an average value calculating means (106 in FIG. 1) in the standard range for obtaining an average value in the standard range determined by the standard range determining means and using this as a reference value of the relative frequency of appearance. . By adopting such a configuration, it is possible to evaluate the tendency of the appearance frequency to fluctuate periodically when calculating the relative appearance frequency.

Further, the trend evaluation apparatus of the present invention uses a scanning period length and a reference period length determining means (10 in FIG. 4) for determining the scanning period length and the reference period length based on the input evaluation target period.
9), reference period determining means (110 in FIG. 4) for determining the reference period based on the scanning period length and the reference period length determined by the scanning period length and the reference period length determining means, and the reference period determining means. And an average value calculating means (111 in FIG. 4) for calculating the average value in the reference period.
By adopting such a configuration and setting a stable period in the time-series data as the reference period, the second object of the present invention can be achieved.

Further, the third trend evaluation device of the present invention may be a conventional technology, or a stability calculation means for calculating a stability in a reference period in addition to the first or second trend evaluation device. (112 in FIG. 8) and a trend evaluation unit (113 in FIG. 8) for evaluating the trend based on the relative appearance frequency and the stability in the reference period. The third object of the present invention can be achieved by adopting such a configuration and taking into account the stability in the reference period in the trend evaluation.

A fourth trend evaluation device according to the present invention is an event information storage device for storing event information (see FIG. 10) in addition to the prior art, the first, second, or third trend evaluation device. 114) and a trend predicting means (1 in FIG. 10) for extracting a word common to the trend word and searching for an event including the common word from the event information storage.
15). The fourth object of the present invention can be achieved by adopting such a configuration and searching for event information based on a common word of trend words.

[0041]

Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of the first embodiment of the present invention. In FIG. 1, the same or similar elements as those in FIG. 16 are denoted by the same reference numerals.

Referring to FIG. 1, a trend evaluation apparatus according to the present invention has a first embodiment in which
1, time series data accumulator 102, average value calculation means 103 for evaluation period, relative appearance frequency calculation means 106, trend accumulator
107, an output unit 108, a standard range determination unit 105, and a standard range average value calculation unit 104. The average value calculation means 103 and the relative appearance frequency calculation means 1
06, the standard range determination unit 105, and the average value calculation unit 104 in the standard range realize their respective functions and processes by executing programs on a computer constituting the trend evaluation apparatus. In this case, the first embodiment of the present invention can be implemented by loading the program from the readable recording medium storing the program into the main storage and executing the program. Specific processing steps for realizing the processing in each means will be described later with reference to flowcharts.

In the first embodiment of the present invention, means for calculating an average value over the entire time range in the conventional apparatus shown in FIG.
120 is replaced with a standard range average value calculating means 104, and a standard range determining means 105 is further provided. Other means have the same configuration as that shown in FIG.

The standard range determining means 105 determines a standard range based on the evaluation target period input by the input means 101. For example, the standard range may be n times the length of the evaluation target period before the evaluation target period. Specifically, if the evaluation period is one week, n weeks and the evaluation period is 1
If it is a year, the standard range is an appropriate period for trend evaluation, such as n years.

The standard range can be appropriately designated to another value by the input means 101. For example,
This is because if it is desired to perform a trend evaluation on a change expected to have a cycle of one year, the standard range must be within one year.

FIG. 3 is an explanatory diagram showing an example of a standard range and a period to be evaluated in the first embodiment of the present invention. In FIG. 3, the evaluation target range is T00 to T01, and the standard range is T10 to T00.

The standard range average value calculating means 104 determines the evaluation target word inputted by the input means 101 from the time series data stored in the time series data storage 102 by the standard range determining means 105. Calculate the average value in the standard range.

The input unit 101 is a unit for inputting various conditions such as a period to be evaluated, a word to be evaluated, and a threshold value of a trend. The time-series data storage 102 stores time-series data of the appearance frequency of each word.

The average value calculation means 103 during the evaluation period is
On the basis of the evaluation target word and the evaluation target period input by the input means 101, an average value of the time-series data stored in the time-series data storage 102 in the evaluation target period is obtained.

The relative appearance frequency calculation means 106 calculates the average value in the standard time range calculated by the average value calculation means 104 in the standard range and the evaluation target period calculated by the average value calculation means 103 in the evaluation target period. Is calculated, and those exceeding the threshold are stored in the trend word storage 107.

The trend word accumulator 107 accumulates the relative appearance frequency obtained by the relative appearance frequency calculation means 106, the evaluation target period, and the words.

The output means 108 outputs the words stored in the trend storage 107 so that the user or the application system can refer to the words.

FIG. 2 is a flowchart showing the operation of the first embodiment of the present invention. The operation of the first embodiment of the present invention will be described with reference to FIGS.

In the first embodiment, the input unit 101 inputs an evaluation target period and an evaluation target word as an end condition (step S101).

The standard range determining means 106 determines a standard range based on the input evaluation target period (step S10).
7).

On the other hand, the average value calculating means 10 during the evaluation target period
3 calculates the average value of the appearance frequency of the evaluation target word in the evaluation target period in the time-series data stored in the time-series data storage device 102 (step S102).

Mean value calculating means 104 in the standard range
Calculates the average value of the evaluation target words input by the input means 101 in the standard range determined by the standard range determination means 105 in the time-series data stored in the time-series data storage 102 (step S103), which is a reference value of the relative appearance frequency.

Next, the relative appearance frequency calculating means 106 calculates the average value Pc in the standard range calculated by the average value calculating means 104 in the standard range and the average value calculating means 10 in the evaluation period.
The ratio Px of the average value Px in the evaluation period calculated by 3
/ Pc is calculated (step S104), and if it exceeds the threshold, the word, the evaluation target period, and the relative appearance frequency are stored in the trend word storage.

The above steps S107, S102, S103, S104
Is repeated for all the evaluation target periods and evaluation target word pairs input by the input means 101 (step S105).

When the end condition is satisfied in step S105, the output means 108 outputs the word stored in the trend storage 107 as a result (step S106).

If the end condition is input to the input means 107 as a specific evaluation target period and all words, the trend evaluation in the specific evaluation target period is repeatedly performed for all words (step S105). It is also possible to extract a trending word in the evaluation target period.

Further, the input means 101 inputs the word W1 and the plurality of evaluation target periods as end conditions, and repeatedly calculates the relative appearance frequency of the word W1 in the plurality of evaluation target periods (step S105), whereby the output means 108 In, it is also possible to search for a period in which the word W1 has become a trend.

Next, the operation and effect of the first embodiment of the present invention will be described.

In the first embodiment of the present invention, the reference value in the relative appearance frequency calculation is calculated by limiting the reference value to an appropriate range based on the evaluation target period. Thus, correct evaluation can be performed even on time-series data having periodicity as shown in FIG.

Next, a second embodiment of the present invention will be described.

FIG. 4 is a block diagram showing the configuration of the second embodiment of the present invention. In FIG. 4, the same reference numerals are given to the same or the same elements as those in FIG. FIG. 7 is an explanatory diagram illustrating an example of a scanning period and a reference period according to the second embodiment of the present invention.

Referring to FIG. 4, according to a second embodiment of the present invention, the average value calculating means 104 in the standard range in the first embodiment shown in FIG. The average value calculating unit 111 is configured by replacing the standard range determining unit 105 with the reference period determining unit 110, and further includes a scanning period length and a reference period length determining unit 109. In addition,
Average value calculation means 111 in the reference period, reference period determination means 11
The functions and processes of the scanning period length and reference period length determining means 109 are realized by executing a program on a computer constituting the trend evaluation apparatus. In this case, by loading the program from the readable recording medium storing the program into the main storage and executing the program,
A second embodiment of the present invention can be implemented. Specific processing steps for realizing the processing in each means will be described later with reference to flowcharts.

Scanning period length and reference period length determining means 109
Determines the scanning period length and the reference period length based on the evaluation period input by the input unit 101. Here, the reference period length is the length of a period that is a reference for calculating the relative appearance frequency, and the scanning period length is a time range for determining the reference period.

FIG. 7 showing an example of the scanning period and the reference period length.
, The scanning period length is T00−T10, and the reference time length is T12−T
It is 11. As an example, the scanning period length is n in the evaluation target period.
The reference time length may be twice as long as the evaluation target period. This is appropriately set by the input unit 101 according to the purpose of the trend evaluation.

The reference period determining means 110 determines the reference period based on the scanning period length and the reference period length determined by the scanning period length and the reference period length determining means 109. here,
The reference period is a period in which the time-series data stored in the time-series data storage unit 102 is most stable in the scanning period. As the stability index, for example, the variance, the standard deviation, and the sum of the absolute values of the differences from the average value in the reference period can be used.

The average value calculating means 111 in the reference period calculates the average value in the reference period determined by the reference period determining means 110. The average value in this reference period becomes the reference value in the relative appearance frequency calculation.

FIG. 5 is a flowchart showing the operation of the second embodiment of the present invention. 5, the same processing steps as those shown in FIG. 2 are denoted by the same reference numerals. Referring to FIG. 4 and FIG.
Will be described in detail. Steps in FIG.
S101, shown in S102 and S104-S106, input means 101,
The processing operations of the average value calculation unit 103, the relative appearance frequency calculation unit 106, and the output unit 108 during the evaluation target period are the same as the operations of steps S101, S102, and S104 to S106 shown in FIG. Description is omitted.

In the first embodiment described above, there is no means for setting a time period in which the change is stable as a reference in the time-series data.

In the second embodiment of the present invention, the input means
The scanning period length and the reference period length determining means 109 determine the scanning period length and the reference period length based on the evaluation target period input by 101 (step S201).

Next, based on the scanning period length and the reference period length determined by the scanning period length and the reference period length determining unit 109, the reference period determining unit 110 determines a reference period by an operation described later (step S202). ).

The average value calculation means 111 for the reference period calculates the average value for the reference period determined by the reference period determination means 110 (step S203). The average value in this reference period becomes the reference value for calculating the relative appearance frequency.

Therefore, assuming that the average value in the reference period is Ps and the average value in the evaluation period is Px, the relative appearance frequency is Px
/ Ps.

FIG. 6 is a flowchart showing the operation of the reference period determining means 110 according to the second embodiment of the present invention. With reference to FIG. 6, the operation of the reference period determination unit 110 will be described.

The reference period determining unit 110 receives the scanning period length and the reference period length from the scanning period length and the reference period length determining unit 109 on condition, and starts scanning before the evaluation target period. For example, in FIG.
110 starts scanning in the reference period from T00.

That is, the average value and the stability in the reference period T are calculated with T being T00>T> T00− (T12−T11) as the first reference period T (step S21).

If the variance calculated in step S21 is the minimum value, the reference period T is stored (step S22),
Steps R101 to R103 are repeated for the entire scanning period while the reference period is slightly shifted (step S24) (step S23).

When the processing is completed, the most stable period in the scanning period is determined as the reference period.

Next, the operation and effect of the second embodiment of the present invention will be described.

In the second embodiment of the present invention, of the time-series data, the most stable reference period length in the scanning period is set as the reference period. Thereby, as shown in FIG. 20, even for the tendency that the appearance frequency gradually increases,
The trend can be evaluated with the period in which the change is always stable as the reference period.

Further, since the period for obtaining the reference period is limited by the scanning period length, as in the first embodiment, as shown in FIG. Also, correct evaluation can be performed.

Next, a third embodiment of the present invention will be described with reference to the drawings.

FIG. 8 is a block diagram showing the configuration of the third embodiment of the present invention. In FIG. 8, the same or similar elements as those in FIG. 4 are denoted by the same reference numerals.

Referring to FIG. 8, in the third embodiment of the present invention, in addition to the configuration of the second embodiment shown in FIG. An evaluation unit 113 is further provided. Note that the stability calculation means 112 and the trend evaluation means 113 in the reference period are:
By executing a program on a computer constituting the trend evaluation device, the respective functions and processes are realized. In this case, the third embodiment of the present invention can be implemented by loading the program from the readable recording medium storing the program into the main storage and executing the program. Specific processing steps for realizing the processing in each means will be described later with reference to flowcharts.

The stability calculation means 112 for the reference period determines the reference period determined by the reference period determination means 110 based on the reference period.
Calculate the stability of time series data in the reference period.

The trend evaluation means 113 calculates the trend based on the relative appearance frequency calculated by the relative appearance frequency calculation means 106 and the stability in the reference period obtained by the stability calculation means 113 in the reference period. For the word that has been evaluated and exceeds the threshold value, the evaluation target period and the trend property are stored in the trend word storage 012.

FIG. 9 is a flowchart showing the operation of the third embodiment of the present invention. In FIG. 9, the same reference numerals are given to the same processing steps as those in FIG. The operation of the third exemplary embodiment of the present invention will be described with reference to FIGS.

In FIG. 9, steps S101, S102, S104
-Input means 101, average value calculation means 103 in the evaluation target period, relative appearance frequency calculation means 106, average value calculation means in the reference period shown in S106, S201, and S202 according to the third embodiment of the present invention. 111, the reference period determining means 110, the scanning period length and the reference period length determining means 109, each operation of the output means 108,
Since they are the same as the respective units of the second embodiment, the description will be omitted. That is, in FIG. 9, steps S204 and S205 are newly added to the flowchart shown in FIG. 5, and the differences from the second embodiment will be explained below.

In the second embodiment, the stability during the reference period is not taken into account in the trend evaluation.

In the third embodiment of the present invention, based on the reference period determined by the reference period determination unit 110, the stability calculation unit 112 in the reference period determines the stability of the time-series data in the reference period. Calculation is performed (step S204).

The trend evaluation means 113 determines the trend based on the relative appearance frequency calculated by the relative appearance frequency calculation means 106 and the stability in the reference period obtained by the stability calculation means 112 in the reference period. Evaluate (Step S20
Five).

Next, the operation and effect of the third embodiment of the present invention will be described.

In the third embodiment of the present invention, the stability in the reference period is taken into account in the trend evaluation. Thereby, it is possible to distinguish between a mere noise as shown in FIGS. 21A and 21B and a true trend.

Further, in the third embodiment of the present invention, the stability calculation means 112 in the reference period and the trend evaluation means 113 are combined with the conventional apparatus shown in FIG. Sex calculation means 112 calculates the stability over the entire time range, the trend property evaluation means 113, the relative appearance frequency calculated by the relative appearance frequency calculation means 106,
By evaluating the trend based on the stability over the entire time range obtained by the stability calculation means 112 in the reference period, it is expected that the same effect as in the third embodiment of the present invention can be obtained.

Further, as a modification of the third embodiment of the present invention, the stability calculating means 112 and the trend evaluation means 113 in the reference period in the third embodiment of the present invention are shown in FIG. In combination with the configuration of the first embodiment, the stability calculation means 112 for the reference period calculates the stability in the standard range, and the trend evaluation means 113 calculates the relative frequency calculated by the relative appearance frequency calculation means 116. The same effect as in the third embodiment of the present invention can be obtained by evaluating the trend based on the frequency of appearance and the stability in the standard range obtained by the stability calculation means 112 in the reference period. Can be expected.

Next, a fourth embodiment of the present invention will be described.

FIG. 10 is a block diagram showing the configuration of the fourth embodiment of the present invention. In FIG. 10, the same or similar elements as those in FIG. 8 are denoted by the same reference numerals.
Referring to FIG. 10, a fourth embodiment of the present invention has a trend predicting unit 11 in addition to the configuration of the third embodiment.
5 and an event information accumulator 114. The functions and processes of the trend prediction means 115 are realized by executing a program on a computer constituting the trend evaluation device. In this case, the fourth embodiment of the present invention can be implemented by loading the program from the readable recording medium storing the program into the main storage and executing the program. Specific processing steps for realizing the processing in each means will be described later with reference to flowcharts.

The event information storage 114 stores date and detailed related words as data for yearly events and various events.

The trend predicting means 115 predicts a trend based on the event information stored in the event information storage 114 and the trend words stored in the trend word storage 107.

FIG. 11 is a flowchart showing the operation of the fourth embodiment of the present invention. Referring to FIGS. 10 and 11,
The operation of the fourth embodiment of the present invention will be described in detail.

In the fourth embodiment of the present invention shown in FIG. 10, input means 101, average value calculating means 103 in the evaluation target period, relative appearance frequency calculating means 106, average value calculating means 111 in the reference period. The operation of each of the reference period determining unit 110, the scanning period length and the reference period length determining unit 109, the output unit 108, the stability calculating unit 112 in the reference period, and the trend evaluating unit 113 is performed by the first, second, and third operations. Since these units are the same as the units 101, 103, 106, 111, 110, 109, 108, 112, and 113 in the third embodiment, description thereof will be omitted.

Since the first, second and third embodiments do not have a function for predicting a trend, it is difficult to predict a word which will become a trend in the future. .

In the fourth embodiment of the present invention, the trend predicting means 115 refers to the trend word accumulator 114 and extracts a word satisfying the condition as a trend word (step).
S301). As the extraction condition, an evaluation period or the like can be used in addition to the total or maximum value of the appearance frequency in the scanning period of the value calculated by the trend evaluation unit 113.

Next, from the event information accumulated in the event information accumulator 114, an event including each trend word and having a date close to the date of the event and the evaluation target period of the trend word is searched (step S302).

For each piece of the searched event information, words that are often included in common among the related words stored in the event information storage unit 104 are set as common words, and trend words are grouped by the common words ( Step S30
3).

Next, the common words used for grouping are included in the related words, and events scheduled in the future are searched from the event information storage 114 (step S304), and the output means 108 outputs the result (step S305). .

Next, the operation and effect of the fourth embodiment of the present invention will be described.

In the fourth embodiment of the present invention, words common to trend words are extracted, and event information is searched based on the extracted words. As a result, a trend can be predicted.

As a modification of the fourth embodiment of the present invention, the trend predicting means 115 and the event information accumulator 114 of the fourth embodiment of the present invention are replaced by the conventional device shown in FIG. By performing the same operation as that of the fourth embodiment of the present invention in combination with the above, it is expected that the same effect as that of the fourth embodiment of the present invention can be obtained.

Further, as another modification, the trend predicting means 115 and the event information accumulator 114 of the fourth embodiment of the present invention are combined with the first embodiment shown in FIG. By performing the same operation as that of the fourth embodiment, it is expected that the same effect as that of the fourth embodiment of the present invention can be obtained.

Further, the trend prediction means 421 and the event information storage 411 of the fourth embodiment of the present invention
By combining with the device of the third embodiment shown in FIG. 4 and performing the same operation as this embodiment,
An effect similar to that of the fourth embodiment of the present invention can be expected. Examples of the present invention will be described in order to further describe the above-described embodiments of the present invention with specific examples.

[0117]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, a first embodiment of the present invention will be described. This example corresponds to the above-described first embodiment of the present invention.

Referring to FIG. 1, this embodiment employs input means.
Keyboard functioning as 101 and standard range determining means 10
5, a central processing unit of a personal computer functioning as an average value calculating means 104 in the standard range, an average value determining means 103 in the evaluation target period, and a relative appearance frequency calculating means 106,
It includes a time-series data storage device 102, a magnetic disk storage device functioning as a trend word storage device 107, and a display device functioning as an output unit 108.

Here, it is assumed that time-series data as shown in FIG. 3 is stored in a time-series data storage in a magnetic disk storage device for a word W having a one-month periodicity. About this word W, "11/1" (November 1)-"1
The explanation will proceed with an example of evaluating the trend, with “1/3” (November 3) as the evaluation target period. In FIG. 3, T01 = “11/3” and T00 = “11/1”. In this example, a word W is input from the keyboard as an evaluation target word, "11/1" to "11/3" as an evaluation target period, and 20 days as a standard range length. The standard range determination means 105 included in the central processing unit determines the standard range for 20 days before “11/1” because the standard range length is specified by the input. That is, T10 in FIG. 3 is “10/12”.

Next, the average value calculation means 103 in the evaluation target period included in the central processing unit performs processing from “11/1” to “11 /
The average value Px of the appearance frequency of the word W in “3” is calculated. Further, the average value calculation means 104 in the standard range included in the central processing unit calculates the word W in “10/12” to “10/31”.
The average value of the appearance frequency Pc of is calculated.

Finally, the relative appearance frequency calculating means 106 included in the central processing unit calculates the ratio Px / Pc of the average value Pc in the standard range and the average value Px in the evaluation target period, and sets the value as a threshold. If the value exceeds the limit, the result is stored in the trend word storage 107 included in the magnetic disk storage device, and the result can be confirmed on the display serving as the output means 108. In this case, the larger the relative appearance frequency Px / Pc is, the higher the trend is, and Px / Pc
It can be seen that> 1 indicates an upward trend, and Px / Pc <1 indicates a downward trend. In FIG. 3, Pc> 1, and it can be seen that there is an upward trend.

In the above-described first embodiment, the standard range is specified by the input means 101. However, the standard range is only required to be sufficiently shorter than the periodicity of the time-series data of the word W, and is included in the central processing unit. The standard range determination means 105 may automatically quadruple the evaluation target period to 12 days.

Next, a second embodiment of the present invention will be described. This embodiment corresponds to the above-described second embodiment of the present invention. In this embodiment, the configuration is the same as that of the first embodiment, but the central processing unit of the personal computer is provided with the standard period determining means 105 as shown in FIG.
Instead of the reference period determining means 110, the average value calculating means 1 in the reference period is used instead of the average value determining means 104 in the standard range.
This embodiment differs from the first embodiment in that it has the functions of the scanning period length and the reference period determining means 109 in addition to functioning as 11.

Here, a certain word W having a one-month periodicity
It is assumed that time-series data as shown in FIG. 7 is stored in the time-series data storage 102 in the magnetic disk storage device. About this word W, "11/1"-"11/7"
The description will be made by taking the evaluation of the trend as an example of the evaluation target period. Here, in FIG. 7, T01 =
“11/7”, T00 = “11/1”. In this example, the word “W” is input from the keyboard as the evaluation target word, “11/1” to “11/7” as the evaluation target period, and “20 days” as the scanning period length. Since the scanning period length and the reference period length determining means 109 included in the central processing unit specify the scanning period length by input, the scanning period is set to “11 /
Decide 20 days before "1". That is, T10 in FIG. 7 is “10/12”.

Since the reference period length is not specified, the reference period is automatically set to “7 days” which is the same as the evaluation target period.
And Next, the central processing unit determines, as the reference period, “7 days” at which the variance is minimum between “10/12” and “10/31” by the reference period determining means. The variance is calculated as the sum of squares of the difference from the mean. For example, in FIG. 7, it is assumed that T11 = “10/15” and T12 = “10/21” are the periods in which the variance in the seven days is the minimum.

Next, the average value calculation means 103 in the evaluation target period included in the central processing unit performs processing from “11/1” to “11 /
The average value Px of the appearance frequency of the word W in “3” is calculated. Further, the average value calculation means 111 for the reference period included in the central processing unit is configured to output the word W in “10/15” to “10/21”
The average value of the appearance frequency Ps of is calculated.

Finally, the relative appearance frequency calculation means 106 included in the central processing unit calculates the ratio Px / Ps of the average value Ps in the reference period and the average value Px in the evaluation period, and the value is calculated. If it exceeds the threshold value, it can be stored in the trend word storage 107 included in the magnetic disk storage device, and the result can be confirmed on the output means 108 such as a display. in this case,
The larger the relative frequency of appearance Px / Ps, the higher the trend, and Px
It can be seen that if / Ps> 1, there is an upward trend, and if Px / Ps <1, there is a downward trend. In FIG. 7, Pc> 1, and it can be seen that there is an upward trend.

In the above-described second embodiment, the standard range is specified by the input means 101, but the standard range may be sufficiently shorter than the periodicity of the time series data of the word W, and is therefore included in the central processing unit. The standard range determination unit 110 may automatically quadruple the evaluation target period to 28 days. Also,
The reference period length may be specified by the input means 101.

Assume that a trend evaluation is performed on the time-series data as shown in FIG. 12 by designating the evaluation target period as “11/4”, the scanning period as 30 days, and the reference period as 5 days. The reference period determination means 110 included in the central processing unit performs the same operation as described above, and determines the most stable reference period as “10 /
9 "to" 10/13 ".

Therefore, compared with the trend evaluation based on the relative appearance frequency based on the average over the entire time, the trend characteristic of “11/4” is not underestimated. Also,
Even if there is a time change from "11/4" to "11/11" and the average value in the entire time range changes, the scan period is extended by 8 days, and the "11/4""And the trend at" 11/11 "can be evaluated.

Next, a third embodiment of the present invention will be described. This embodiment corresponds to the above-described third embodiment of the present invention.

This embodiment has the same configuration as that of the second embodiment. However, as shown in FIG. 8, the central processing unit of the personal computer uses the stability calculation means 11 in the reference period.
2 and a function as trend evaluation means 113.

Here, an example will be described in which the tendency of a word Wx whose appearance frequency changes drastically as shown in FIG. 13 and a word Wy which suddenly becomes a trend from a stable change is evaluated. In this case, Wx and Wy are used as evaluation target words from the keyboard, and “12/3” is used as the evaluation target period.
And 7 days as the reference period length. The central processing unit operates in the same manner as in the second embodiment, and sets the most stable period in the time-series data of the appearance frequency of the word Wx as the reference period.

In FIG. 13, the reference period is “11/13”
"11/19". Next, the central processing unit sets the stability G of the change of the word Wx in the reference period as the reciprocal of the variance V, and G = 1
Calculate / V.

Next, the central processing unit obtains the average value Ps in the reference period and the average value Px in the evaluation period by the same operation as in the second embodiment, and the ratio H = (Px−Ps ) / Ps is the relative frequency of appearance.

Further, the trend evaluation means 113 included in the central processing unit calculates the product of the relative appearance frequency H and the stability G in the reference period as the trend F = G × H. Here, the relative appearance frequency is calculated as Px /
The reason why (Px−Ps) / Ps is used instead of Ps will be described. When stability is incorporated into the trend evaluation, as in the second embodiment, if Px / Ps> 1, the rising trend, Px
If / Ps <1, it is impossible to judge a tendency such as a downward trend.

Therefore, by setting (Px−Ps) / Ps as the relative frequency of appearance and defining the product of the relative frequency of appearance H and the stability G as the trend F, if F> 0, then increase, if F <0, decrease. This is because such a tendency can be determined.

For the word Wx, the variance in the reference period is Vx,
The trend is Fx.

Next, the central processing unit calculates the variance Vy and the trend Fy in the reference period in the same procedure for the word Wy. At this time, as shown in FIG. 13, even if the relative appearance frequencies are equal for both the words Wx and Wy, Fy is greater than or equal to Fx because Vx> Vy.

In this embodiment, by adding the functions of the stability calculation means 112 and the trend evaluation means 113 in the reference period to the central processing unit of the personal computer of the second embodiment, noise and trends can be reduced. In contrast to the prior art that enables distinction, the central processing unit of a personal computer is provided with a function of a stability calculation means 112 and a trend evaluation means 113 during a reference period, thereby enabling the relative frequency of appearance between two words to be increased. Even if are the same, the noise and the trend can be distinguished as in the present embodiment by taking the stability over the entire period into consideration for the trend evaluation.

Also, in the first embodiment, by adding the functions of the stability calculation means 112 and the trend evaluation means 113 in the reference period to the central processing unit of the personal computer, two words can be used. Even if the relative appearance frequencies are the same, the noise and the trend can be distinguished as in the present embodiment by taking the stability in the standard range into consideration for the evaluation of the trend.

Next, a fourth embodiment of the present invention will be described. This embodiment corresponds to the above-described fourth embodiment of the present invention.

This embodiment has the same configuration as that of the third embodiment, but the central processing unit of the personal computer has the function of the trend predicting means 115 as shown in FIG. Is the event information storage 114
The function is further provided.

Here, it is assumed that words as shown in FIG. 14 are extracted as trend words and stored in the trend word storage of the magnetic disk storage device. The central processing unit searches the event information accumulator in the magnetic disk storage device for an event including these trend words and having a near date.

As shown in FIG. 15, the event information is data including various related events such as genre, venue, organizer, performer, in addition to the date and event name of each event.

For the trend word in FIG. 14, "10/25
"Akihana Prize", "11/1 F1 Japan Grand Prix", "11/1 Emperor Prize / Autumn", "11/8 Kikuhana Prize" are searched.

Next, the central processing unit finds common words from the related words of the searched event, and groups trend words.

In this embodiment, "horse racing" and "G1" are common words, "Akihana Prize", "Emperor Prize" and "Kikuhana Prize" belong to one group, and "F1" belongs to another group. .

Next, the central processing unit searches for events including common words from events scheduled in the future. In this example, "11/15 Queen Elizabeth's Cup" is searched. By outputting this search result on a display, it is possible to obtain information that will become a trend word in the future.

As described above, in the fourth embodiment, in addition to the configuration of the third embodiment, the function of the trend predicting means 115 is added to the central processing unit of the personal computer, and the event information is stored in the magnetic disk storage device. By adding the function of the device 114, it is possible to predict the trend as in the present embodiment, but in addition to the configuration of the prior art, the function of the trend prediction means 115 is added to the central processing unit of the personal computer, By adding the function of the event information accumulator 114 to the magnetic disk storage device, it is possible to predict the trend as in the present embodiment.

In addition to the configuration of the first embodiment, the function of the trend predicting means 115 is added to the central processing unit of the personal computer, and the function of the event information storage 114 is added to the magnetic disk storage device. Thus, a trend can be predicted as in the present embodiment. Further, in addition to the configuration of the second embodiment, the central processing unit of the personal computer includes trend predicting means 11.
By adding the function of 5, and adding the function of the event information storage 114 to the magnetic disk storage device, it is possible to predict a trend in the same manner as in the present embodiment.

As described above, in the first to fourth embodiments, the frequency of appearance of a word has been described as an example. However, as the time-series data, the probability of appearance for the entire word may be used.

[0153]

As described above, according to the present invention, the following effects can be obtained.

A first effect of the present invention is that a trend can be evaluated even for a change having a periodicity as shown in FIG.

The reason is that, in the present invention, the reference value in the relative appearance frequency calculation is calculated within an appropriate range based on the evaluation target period.

The second effect of the present invention is that, even if the frequency of appearance gradually increases as shown in FIG. 20, the trend can be evaluated using a period in which the change is always stable as a reference period. ,That's what it means.

The reason is that, in the present invention, the reference period having the most stable reference period length in the scanning period in the time-series data is used as the reference period.

A third effect of the present invention is that mere noise and a true trend can be distinguished from the time-series data as shown in FIGS. 21 (a) and 21 (b). is there.

The reason is that, in the present invention, the stability in the reference period is taken into account in the trend evaluation.

A fourth effect of the present invention is that a trend can be predicted.

The reason is that in the present invention, words common to trend words are extracted, and event information is searched based on the extracted words.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a flowchart showing an operation of the first exemplary embodiment of the present invention.

FIG. 3 is an explanatory diagram illustrating an example of a standard range and an evaluation target period according to the first embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 5 is a flowchart showing the operation of the second exemplary embodiment of the present invention.

FIG. 6 is a flowchart showing an operation of a reference period determining unit according to the second embodiment of the present invention.

FIG. 7 is an explanatory diagram illustrating an example of a scanning period and a reference period according to the second embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 9 is a flowchart showing the operation of the third exemplary embodiment of the present invention.

FIG. 10 is a block diagram showing a configuration of a fourth exemplary embodiment of the present invention.

FIG. 11 is a flowchart showing an operation of the fourth exemplary embodiment of the present invention.

FIG. 12 is an explanatory diagram showing an example of time-series data in the second example of the present invention.

FIG. 13 is an explanatory diagram showing an example of time-series data in a third example of the present invention.

FIG. 14 is an explanatory diagram showing an example of a trend word stored in a trend word storage according to the fourth embodiment of the present invention.

FIG. 15 is an explanatory diagram showing an example of event information stored in an event information storage according to the fourth embodiment of the present invention.

FIG. 16 is a block diagram showing a configuration of a conventional technique.

FIG. 17 is a flowchart showing the operation of the conventional technique.

FIG. 18 is an explanatory diagram showing an example of an evaluation target period and an entire time range in the related art.

FIG. 19 is an explanatory diagram showing an example of a first problem in the related art.

FIG. 20 is an explanatory diagram showing an example of a second problem in the related art.

FIG. 21 is an explanatory diagram showing an example of a third problem in the related art.

[Explanation of symbols]

 101 input means 102 time series data storage 103 average value calculation means in evaluation period 104 average value calculation means in standard range 105 standard range determination means 106 relative frequency calculation means 107 trend word storage 108 output means 109 scanning period Length and reference period determination means 110 Reference period determination means 111 Average value calculation means for reference period 112 Stability calculation means for reference period 113 Trend evaluation means 114 Event information storage 115 Trend prediction means 120 Average over the entire time range Value calculation means

Claims (20)

[Claims] 1. A means for determining a standard range for defining a range for obtaining a reference value of a relative value of time-series data based on an input evaluation target period, and averaging the time-series data in the determined standard range. Means for calculating a value, wherein the relative value is calculated from an average value of the time-series data in the standard range and an average time of the time-series data in the evaluation period. . 2. A means for determining a scanning period length and a reference period length of time-series data based on an input evaluation target period, and time-series data based on the determined scanning period length and reference period length. Means for determining a reference period, and means for calculating an average value of the time-series data in the determined reference period, comprising: an average value of the time-series data in the reference period; and And calculating a relative value from an average value of the time series data. 3. A means for calculating the stability of the time-series data in a reference period, and a trend evaluation for evaluating a trend of the time-series data based on the relative value and the stability in the reference period. The trend evaluation device according to claim 1 or 2, further comprising: 4. The trend evaluation device according to claim 1, wherein said time-series data comprises appearance frequency information of sampled words, and said relative value is a relative appearance frequency. 5. A trend evaluation device for evaluating a trend from time-series data of the appearance frequency of a word, wherein the trend evaluation device according to any one of claims 1 to 3, further comprising: means for storing event information; And a trend predicting means for extracting a word common to the trend word and searching for an event including the common word from the event information accumulator. 6. A standard range determining means for determining a standard range based on the input evaluation target period and the evaluation target word based on the evaluation target period of the evaluation target word, wherein the standard range determining means determines the standard range. Means for calculating an average value in the standard range for obtaining an average value in the standard range, an average value of the frequency of occurrence of the evaluation target word in the standard range, and an appearance frequency of the evaluation target word in the input evaluation target period And a relative appearance frequency calculating means for calculating a relative appearance frequency from an average value of the trend evaluation device. 7. A scanning period length and a reference period length determining means for determining a scanning period length and a reference period length based on an evaluation target period for an input evaluation target period and an evaluation target word; A reference period determining unit that determines a reference period based on the scanning period length and the reference period length, an average value calculating unit that calculates an average value in the determined reference period, and an average value calculating unit in the reference period. Average value of the appearance frequency in the reference period, and relative appearance frequency calculation means for calculating a relative appearance frequency from the average value of the appearance frequency of the evaluation target word in the input evaluation target period, Trend evaluation equipment. 8. A stability calculation means in a reference period for calculating stability in the standard range determined by the reference range determination means, and a relative frequency of occurrence calculated by the stability calculation means in the reference period. The trend evaluation device according to claim 6, further comprising: a trend evaluation unit that evaluates the trend based on the stability in the standard range. 9. A stability calculation means in a reference period for calculating stability in the reference period determined by the reference period determination means; a relative appearance frequency calculated by the relative occurrence frequency calculation means and a reference frequency in the reference period. 9. The trend evaluation device according to claim 8, further comprising: a trend evaluation unit that evaluates the trend based on the stability in the reference period calculated by the stability calculation unit. 10. An average value of the appearance frequency of the evaluation target word in the entire time range of the time-series data with respect to the input evaluation target period and the evaluation target word, and the appearance of the evaluation target word in the evaluation target period A trend evaluation apparatus comprising a relative frequency calculating means for calculating a relative frequency of occurrence from an average value of the frequency; a stability calculating means for a reference period for calculating the stability of the time-series data over the entire time range; And a trend evaluation means for evaluating the trend based on the relative appearance frequency calculated by the frequency calculation means and the stability over the entire time range calculated by the stability calculation means in the reference period. A trend evaluation device characterized by the following. 11. The average value of the frequency of appearance of the evaluation target word in the entire time range of the time-series data with respect to the input evaluation target period and evaluation target word, and the appearance of the evaluation target word in the evaluation target period In the trend evaluation device, there is provided a means for calculating a relative appearance frequency from an average value of the frequency; an event information storage means for storing event name, date, and related words as data; and a trend word stored in the trend word storage. Extract common words from the event accumulator,
Trend forecasting means for forecasting trends based on common words,
A trend evaluation device comprising: 12. An event information storage means for storing the name, date, and related words of an event as data, and a common word of a trend word stored in the trend word storage means is extracted from the event storage means. The trend evaluation device according to any one of claims 6 to 11, further comprising: a trend prediction unit that predicts a trend based on: 13. A standard range for limiting a range for obtaining a reference value of relative frequency of appearance is determined based on the input evaluation target period, an average value of frequency of appearance in the standard range is determined, and the average A trend evaluation method characterized by using a reference value of an appearance frequency. 14. A scanning period length and a reference period length are determined based on the input evaluation target period, and a reference period is determined based on the determined scanning period length and the reference period length. The average value of the appearance frequency in the reference period
A trend evaluation method, wherein the average value is used as a reference value of a relative appearance frequency. 15. The method according to claim 13, wherein the stability of the appearance frequency in the reference period is calculated, and the trend characteristic is evaluated based on the relative appearance frequency and the stability in the reference period. Trend evaluation method. 16. The method according to claim 1, wherein event information is stored in a storage unit in advance, a word common to the trend word is extracted, and an event including the common word is retrieved from the event information to predict the trend. Item 16. The trend evaluation method according to any one of Items 13 to 15. 17. A process for determining a standard range based on the evaluation target period of the evaluation target word with respect to the input evaluation target period and the evaluation target word, and (b) determining the standard range. (C) calculating the average value of the appearance frequency of the evaluation target word in the standard range and the average value of the appearance frequency of the evaluation target word in the input evaluation target period. A recording medium for recording a program for causing a computer to execute each of the above-described processes (a) to (c) of calculating the appearance frequency. 18. A process for determining a scanning period length and a reference period length based on the evaluation target period for the input evaluation target period and evaluation target word; and (b) the determined scanning. A process of determining a reference period based on the period length and the reference period length; (c) a process of calculating an average value in the determined reference period; and (d) an appearance frequency of the evaluation target word in the reference period. Calculating the relative frequency of occurrence from the average value of and the average value of the frequency of appearance of the evaluation target word in the input evaluation target period. Causing the computer to execute each of the above processes (a) to (c) Recording medium on which the program of the above is recorded. 19. The recording medium according to claim 16, wherein (e) processing for calculating stability in the standard range determined by said standard range determining means; and (f) calculating the calculated relative appearance frequency and A recording medium for recording a program for causing a computer to execute each of the above-mentioned processes (e) to (f) on the basis of the evaluation of the trend based on the stability in the standard range. 20. The recording medium according to claim 17, wherein: (g) processing for calculating the stability in the determined reference period; and (h) calculation by the relative appearance frequency and stability calculation means in the reference period. A recording medium on which is recorded a program for causing a computer to execute the processes of (g) to (h) for evaluating the trend based on the stability obtained in the reference period.