JP2001273381A - Method and device for supporting fashion prediction, and storing medium with fashion prediction supporting program stored thereon - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a method/device for predicting which candidate of a word or concept in fashion comes true in the case of preparing the candidates of words or concepts in fashion and for numerically clarifying what kind of social background can make the word in fashion come true. SOLUTION: A distance between a word dictionary and the candidate of a word in fashion is quantized by a natural language processing technique such as a vector space method. Especially as the feature of this invention, the higher order concept of an index word in a dictionary is stored and what is the higher order concept of the index word close to the candidate of the word or concept in fashion is investigated. Thus, a social background for forming the word or concept in fashion can be clarified. Furthermore, not only the higher order concept is investigated but its ratio, correlation coefficient and annual change can be utilized for predicting the fashion.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for supporting a fashion prediction and a storage medium storing a program for supporting a fashion prediction, and more particularly to a copy writer for making a copy, at the time of concept development and after making the copy. The present invention relates to a fashion prediction support method and apparatus for supporting a copywriter with a support system based on numerical values at the time of verification, and a storage medium storing a fashion prediction support program.

[0002]

2. Description of the Related Art Buzzwords are used in CF (Commercial Film) or play a role like slogan. In terms of sales promotion, the quality of buzzwords has a great influence on the sales of the product itself and on the product image. Creating “good buzzwords” is the biggest challenge and goal for copywriters.

However, even if you say that you make good buzzwords,
It is not easy to know which of the buzzwords newly created by copywriters is “hit”.

[0004] In such a situation, the copywriter has no choice but to choose a catch phrase "somehow likely to hit" by his own intuition, and there is no method to objectively judge the good or bad. On the other hand, the orderer (client) also has no choice but to rely on intuition to predict what is likely to be hit from the presented catch phrase. As a result, you bet on company luck with the intuition of a lay person in charge and a feeling that you are likely to suit your favorite actress.

As described above, even when a copywriter comes up with a plurality of buzzword candidates, there is no criterion for judging which of them is presented to the client. Therefore, the copywriter cannot present any objective basis when presenting it to the client. on the other hand,
The presented client also has to rely on intuition as a means of determining which one is better, and cannot make an objective judgment. As a result, a buzzword is selected only by feeling, and the result is delighted.

It is not automatically possible to come up with a buzzword. For example, buzzwords such as "Dachu-no" remain in people's hearts because they have never existed before. Buzzwords can only be created by humans. Therefore, what is required is
It is a means of predicting to some extent whether the buzzword you created will hit.

[0007] For this reason, new buzzwords created by copywriters need to be ordered in some way, and a framework that allows the user to understand the reason why the ranking is given is required. It is. We need a tool to order new buzzwords. With such a tool, copywriters can not only select promising buzzwords from their own ideas, but also explain to clients why they are promising.

As an invention of such a trend prediction support technology, a technology (Japanese Patent Application No. 11-2002) which is an invention of the inventors is disclosed.
No. 08, "Method and Apparatus for Supporting Epidemic Prediction and a Storage Medium in Which a Prediction Prediction Support Program is Stored", filed on July 14, 1999). In the technology of the prior application, words including buzzwords are grouped as headwords for each year and stored in a headword storage unit, and a description of the word is stored in a description storage unit. Is stored in the buzzword storage means, and the distance between the word of a certain year stored in the buzzword storage means and the word in the year before that year stored in the headword storage means is stored in the explanatory note storage. The distance can be calculated based on the explanation sentence stored in the means position.

In Japanese Patent Application No. 11-200208, words including buzzwords are grouped as headwords for each year and stored in the headword storage means, and buzzwords of that year are stored in the buzzword storage means. The distance between the word of a certain year stored in the buzzword storage means and the word in the year before that year stored in the headword storage means is calculated as the vowel / consonant of both the headword and buzzword. And sound-repellent flights
The distance can be calculated based on the degree of coincidence of the prompting sound.

[0010]

However, according to the prior art, it is only possible to measure the distance between the buzzword and the headword and to know the distance. At that time, you know which word is “close”,
The economic background cannot be known.

[0011] This technique alone is not sufficient as a basis for the ordering of buzzwords invented by copywriters. When explaining to a client the reason for the ordering and why prominent words are promising, there is no social need.・ It is not possible to explain including the economic background.

The present invention has been made in view of the above points, and it is necessary to arrange some sort of new buzzword, and to determine why the assigned rank is higher. It is an object of the present invention to provide a trend prediction support method and apparatus, and a storage medium storing a trend prediction support program, which can be understood by a user as well as the social and economic background that the buzzword may have. I do.

[0013]

FIG. 1 is a diagram for explaining the principle of the present invention.

The fashion prediction support system of the present invention (claim 1) (1 in FIG. 1)

A set of words including a buzzword is used as a headword, and the time lapse including each year (hereinafter, this time lapse is referred to as “year”, but the period is not necessarily limited to a fiscal year or a calendar year). A headword storage step (11 in FIG. 1) for grouping and storing according to

An explanatory sentence storing step (12 in FIG. 1) for storing, as an explanatory sentence, an explanation for each of the headwords stored in the headword storing step;

An upper concept storage step (13 in FIG. 1) for storing a higher concept for each headword stored in the headword storage step;

A buzzword storage step (14 in FIG. 1) for selecting a word that has become popular and noticed during the year from among the headwords stored in the headword storage step and storing the word as a buzzword. ,

The distance between the word of a certain year stored in the trendy word storing step and the word in the year before the corresponding year stored in the headword storing step is stored in the explanatory note storing step. Calculating the inter-word distance based on the explanation sentence (15 in FIG. 1);

[0020] A certain number of superordinate concepts for each word stored in the headword storage step, in which the distance between the words stored in the buzzword storage step is determined to be short by the inter-word distance calculation step. Is selected from the higher-level concepts stored in the higher-level concept storage step (step 16 in FIG. 1).
When,

An upper concept displaying step (1 in FIG. 1) for displaying the upper concept selected in the upper concept selecting step.
7).

Further, in the fashion prediction support system according to the present invention (claim 2), in the trend calculation step, the distance between words is calculated by using a vector space method. This is a prediction support method.

FIG. 4 is another diagram for explaining the principle of the present invention. The trend prediction support device (4 in FIG. 4) of the present invention (claim 3)

A set of words including buzzwords is used as a headword, and the time progress including the year (hereinafter, this time progress is referred to as “year”, but the period is not necessarily limited to the fiscal year or calendar year). Headword storage means 41 for grouping and storing according to

An explanation sentence storage means 42 for storing the explanation for each entry word stored in the entry word storage means 41 as an explanation sentence;

An upper concept storage means 43 for storing an upper concept for each headword stored in the headword storage means 41;

From among the headwords stored in the headword storage means 41, a word that has become popular and noticed during the year is selected and the word is stored as a buzzword.
4 and

The distance between the word of a certain year stored in the buzzword storage means 44 and the word in the year before the corresponding year stored in the headword storage means 41 is expressed by:
An inter-word distance calculating unit 45 that calculates based on the explanatory sentence stored in the explanatory sentence storage unit 42;

The headword storage unit 41 determines that the distance is short from each word stored in the buzzword storage unit 44 by calculation using the inter-word distance calculation unit 45.
A high-level concept selecting unit 46 that selects a certain number of high-level concepts for each word stored in the high-level concept stored in the high-level concept storage unit 43;

The present invention is characterized in that it comprises a superordinate concept display means 47 for displaying the superordinate concept selected by the superordinate concept selecting means 46.

Further, the fashion prediction support apparatus of the present invention (claim 4) is characterized in that the inter-word distance calculation means (45 in FIG. 4)
4. The fashion prediction support apparatus according to claim 3, wherein a vector space method is used when obtaining a distance from a word.

Further, the present invention (claim 5) is a storage medium storing a trend prediction support program for predicting a trend from a trend word candidate and selecting a trend word,

A set of words including buzzwords is used as a headword, and the time lapse including each year (hereinafter, this time lapse is referred to as “year”, but the period is not necessarily limited to the fiscal year or calendar year). A headword storage process for grouping and storing according to

An explanatory sentence storing process for storing, as an explanatory sentence, an explanation for each of the headwords stored by the headword storing process;

An upper concept storage process for storing a higher concept for each headword stored by the headword storage process;

A buzzword storage process of selecting a word that has become popular and noticed during the year among the headwords stored by the headword storage process, and storing the word as a buzzword;

The distance between the word of a certain year stored by the buzzword storage process and the word of the year before the corresponding year stored by the headword storage process is stored by the explanatory sentence storage process. A word-to-word distance calculation process to calculate based on the description sentence,

The superordinate concept for each word stored in the headword storage process, which is determined to have a short distance from each word stored in the buzzword storage process by the calculation using the inter-word distance calculation process. A high-level concept selection process of selecting a certain number of them from the high-level concepts stored by the high-level concept storage process;

And a superordinate concept display process for displaying the superordinate concept selected by the superordinate concept selecting process.

According to a sixth aspect of the present invention, in the process of calculating the distance between words, a vector space method is used to determine the distance to a word, and the fashion support program according to the fifth aspect is stored. This is a storage medium.

FIG. 2 is a diagram for explaining the principle of another configuration method of the present invention.

The present invention (claim 7) relates to a fashion prediction support method (2 in FIG. 2) for predicting a fashion from buzzword candidates and selecting a buzzword.

The method of claim 1 (11, 12, 1 in FIG. 1)
A superordinate concept ratio calculating step (3, 14, 15, 16) for calculating the ratio of the superordinate concept for each buzzword to each buzzword according to claim 1 selected for each superordinate concept. 2) in FIG.

A fashion prediction support method characterized by having a rate calculation result display step (21 in FIG. 2) for displaying a rate calculated by the higher concept rate calculation step.

The present invention (claim 8) shown in FIG.
A trend prediction support device 9 for predicting a trend from buzzword candidates and selecting a buzzword,

The means of claim 3 (41, 42, 4 in FIG. 4)
A superordinate concept ratio calculating means for calculating, for each superordinate concept, the ratio of each superordinate concept to each buzzword selected with the use of (3, 44, 45, 46). (90 in FIG. 9),

The present invention is characterized by having a ratio calculation result display means 91 for displaying the ratio obtained by the superordinate concept ratio calculation 90.

The present invention (claim 9) is a storage medium storing a trend prediction support program for predicting a trend from a trend word candidate and selecting a trend word,

For the superordinate concept for each buzzword of claim 5 selected using the superordinate concept selection process of claim 5, the superordinate concept ratio for calculating the ratio of each superordinate concept to each buzzword of claim 5 Calculation process,

A ratio calculation result display process for displaying a ratio obtained by the calculation in the superordinate concept ratio calculation process.

FIG. 3 is a diagram for explaining another principle of the present invention.

The present invention (claim 10) relates to a fashion prediction support method (3 in FIG. 3) for predicting a fashion from buzzword candidates and selecting a buzzword.

A superordinate concept ratio calculation for calculating a ratio of each superordinate concept to each buzzword, which is selected by using the method according to claim 1, for each superordinate concept. Step (20 in FIG. 3);

A multi-year correlation calculating step (30 in FIG. 3) for calculating a correlation with respect to the ratios for a plurality of years determined by the superordinate concept ratio calculating step;

Correlation calculation result display step for displaying the correlation coefficient obtained in the multi-year correlation calculation step (FIG. 3)
31).

Another embodiment of the present invention shown in FIG. 11 (claim 11) is a trend prediction support apparatus (10 in FIG. 11) for predicting a trend from a trend word candidate and selecting a trend word.

For the superordinate concept for each buzzword of claim 3 selected by the superordinate concept selecting means of claim 3, a superordinate concept for calculating a ratio of each superordinate concept to each buzzword of claim 3 Ratio calculation means (90 in FIG. 11)
When,

A multi-year correlation calculating means (111 in FIG. 11) for calculating a correlation with respect to the ratio for a plurality of years obtained by the high-level concept ratio calculating means 90;

Correlation calculation result display means 11 for displaying the correlation coefficient obtained by the multi-year correlation calculation means 111
2 is provided.

The present invention (claim 12) is a storage medium storing a trend prediction support program for predicting a trend from a trend word candidate and selecting a trend word,

For the superordinate concept for each buzzword of claim 5 selected using the superordinate concept selection process of claim 5, the superordinate concept ratio for calculating the ratio of each superordinate concept to each buzzword of claim 5 Calculation process,

A multi-year correlation calculation process for calculating a correlation with respect to the ratios for a plurality of years obtained in the superordinate concept ratio calculation step;

A correlation calculation result display process for displaying the correlation obtained by the multi-year correlation calculation process is provided.

As described above, in the present invention, the headword stored in the headword storage step (headword storage means) or the buzzword stored in the buzzword storage step (buzzword storage means) is It is characterized by being put together. For example, as a headword, there is a headword of “Basic knowledge of modern terminology” issued by Free Kokuminsha.
The headwords of "Basic knowledge of modern terminology" are partially replaced every year, new words and coined words that attract people's interest are registered, and old words are deleted from the headwords. In this case, the explanation of the headword stored in the explanation sentence storage step (explanatory note storage means) is “basic knowledge of modern terminology”.
Is an explanatory note for each headword. On the other hand, examples of buzzwords stored in the buzzword storage step (buzzword storage means) are buzzwords that have been selected for the “buzzword award” also established by Free Kokuminsha. The details of the word-to-word distance calculation step (word-to-word distance calculation means) will be described later, but the closeness between words is expressed numerically.

First, a user of the present invention needs to consider a number of buzzwords on his own. Then, the imagined buzzword is stored as a buzzword in the next year and a buzzword in a buzzword storage step. Then, an inter-word distance calculation step (inter-word distance calculation means) determines to which superordinate concept the headword determined to be close to the buzzword that has come to mind.
And a higher concept display step (higher concept display means). In addition, it is possible to display the transition of the superordinate concept of a word considered to be close to the past buzzword for each year. Then, the correlation of each superordinate concept can be calculated and displayed from the ratio of the superordinate concept after the year.

For example, the buzzword “Shomuni”, which was announced as a buzzword award in the fall of 1998 (from Jiyokumin Kokuminsha), included “Kusanagi-kun” and “Energy” among the headwords of the previous year. I know it's close to what I said. It is also known that the superordinate concepts of these close headwords include “customs / fashion” and “culture / art”. According to the results of actual analysis, the general concept of "economy and management"
It is clear that it is close to "Shomni". All of this suggests that the word "shomni" was not sold only in the short skirt of the leading actress.
It was a buzzword with a social and economic background. In this way, while looking at the distance between the past buzzword and the previous buzzword, the new buzzword proposed this time is examined to see which is closer to the superordinate concept of the past buzzword. For example, a buzzword I came up with
If it is close to a word that belongs to “customs / fashion” or “culture / art”, will it sell? Also, if the new word is close to a word belonging to a superordinate concept that the public is not likely to be interested in, the buzzword can be determined to be difficult to sell.

At least, in the present invention, buzzwords that make the user think “now” can be eliminated as described above. And
It is clear how copywriters or clients should act when the system shows that new buzzwords are surprisingly relevant to recent social situations. Of course, it is not known if this system will hit it. It is a fact of the future. But when it hits, it is hypothesized what deep psychology produced this hit. According to the present invention, by presenting the social situation behind such a buzzword, it is possible to confirm whether the buzzword considered by the user conforms to the social situation.

[0068]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an actual embodiment of the present invention will be described.

FIG. 1 shows a first embodiment of a fashion prediction support system according to the present invention. The present invention shown in FIG. 1 (claim 1) includes the following steps.

First, the headword storage step 11 is executed. In this step, a set of buzzwords is stored as “headwords” in groups according to the passage of time, such as for each year. Here, the term "new" is, for example, "financial big bang", "computer virus" or the like. Further, by storing information such as a word number and a pointer together with a description sentence to be described later and a superordinate concept, it is possible to expect a higher processing speed. It should be noted that the headwords are stored separately for each year, and the method of realizing the headwords may be selected according to the preference of those skilled in the art. For example, a plurality of files can be created for each year on the hard disk, and the headword for each year can be stored for each file.

Next, an explanation storing step 12 is executed.
Here, the description for each headword stored in the headword storage step is stored as “explanatory text”. The description may be stored in, for example, an SGML format on a hard disk.
An area such as an array on M may be secured and stored by a program. Since it is necessary to correspond to the headword stored in the headword storage step 11, a number may be given to the headword.
Also, the headword and the description may be described in the same SGML and stored in one memory. Note that the description need not always be stored for each year, but a plurality of files may be created for each year on the hard disk, and the headword for each year may be stored for each file.

Next, the upper concept storage step 13 is executed. Here, the semantic superordinate concept of each headword stored in the headword storage step 11 is stored.
The high-level concept may be stored in a hard disk drive in the SGML format, or an area such as an array may be secured in a DRAM for display by a program for display as described later. Since it is necessary to correspond to the headword stored in the headword storage step, a number may be given to the headword. Also, the headword and the superordinate concept may be described in the same SGML and stored in one memory.

Next, a buzzword storage step 14 is executed.
Here, among the headwords stored in the headword storage step, words that have become popular or attracted attention during the year are selected and stored as “buzzwords” for each year. For example, in “Basic knowledge of modern terminology”, a term representing a particular year in the fall of a certain year is selected as a “buzzword award”. The above-mentioned "Shomuni" and "Dachu-no"
Is an example. The modern terms included in these "Buzzword Awards" will be included as headwords in the "Basic Knowledge of Modern Terms" issued the following year. In this sense, in the following description, “buzzwords” are thus a subset of “headwords”. However, "Dachu-no" is registered as a headword in the 1999 edition of "Basic Knowledge of Modern Terms", but does not exist in the "Basic Knowledge of Modern Terms" before the 1999 edition. Therefore, it is necessary to extract and store only headwords that have hit the word, especially as buzzwords, among headwords that exist in a certain year. In the following explanation, buzzwords are described as being selected from headwords in order to unify the description, but future buzzwords newly created by copywriters are not necessarily registered as headwords for the year. Even if it does not, it does not affect the gist of the present invention.

The four steps described above (11, 12,
13, 14) basically only store data. FIG. 5 shows a data format of a headword and a description actually stored. The example shown in the figure is one format of data that the above-mentioned headword storage step, explanation sentence storage step, and upper concept storage step can handle. However, the data format is the SGML format for convenience of explanation. Here, the text surrounded by the tag <LEVEL_4> is “explanatory text” and the tag <LEVEL_4>
_4> is a “entry word”, and the content described as a parameter “name” of the tag <LEVEL_1> is a “superordinate concept”. However, in this example, although there are three superordinate concepts, the outermost (that is, the uppermost) concept will be described below as being a superordinate concept. In the headword storage step 11 for storing the headword, only the headword part of FIG. 5 is stored. However,
It is necessary to provide some kind of link between the high-level concept and this, and various realization methods such as a headword number and a pointer can be easily assumed. Description sentence storage step 12,
The same applies to the upper concept storage step 13.

Next, the inter-word distance calculation step 15 (FIG. 1) is executed. A typical distance calculation method is a vector space method. In this method, the distance between words included in a headword set is calculated as follows. However, it is assumed that all headwords are accompanied by explanatory sentences.

FIG. 8 is a flowchart showing the operation of the inter-word distance calculation step according to the first embodiment of the present invention.

Step: S171) The explanatory sentences of all the headwords are morphologically analyzed and divided into words for each headword.

Step: S172) Label words appearing in all headwords (however, particles or auxiliary verbs are usually excluded) and label them W1,..., Wn. here,
n is the type of the word that has appeared. According to the basic knowledge of modern terminology, the number of words exceeds tens of thousands even for one year.

Step: S173) For each headword, the vector space of the headword is determined as follows. That is, it is an n-dimensional binary vector (an n-dimensional vector whose element is either "0" or "1"), and the j-th element includes the word Wj in the description of the headword. If there is, it becomes "1", and if it is not included, it becomes "0".

Step: S174) The mutual distance between any two headwords in the headword set S is defined as the inner product of the vector space of each headword.

Various variations can be considered in the distance calculation by the above-described vector space method. For example,
The length of the vector may be normalized to 1 for each headword. Further, in the above description, the weight is limited to “0” and “1”, but more generally, it may be a floating point number. For example, when the same word appears many times in a certain explanatory sentence, it is conceivable that the weight of the element is made larger than that of the other words. For details of the natural language processing technology including the morphological analysis technology, see, for example, Iwanami Shoten Publishing, Makoto Nagao, “Introduction to Natural Language Processing”.

Next, returning to FIG. 1, an upper concept selection step 16 is executed. This is because, as shown in FIG. 5, a superordinate concept is given to each headword in advance, and in the above-described inter-word distance calculation step, a superordinate concept corresponding to the headword determined to be close is determined. is there.

Finally, a broader concept display step 17 is executed to display the broader concept of the headword determined to be close to the user. In general, by looking at the broader concepts, it is possible to look at headwords that are close to the broader and social aspects, such as "economics / management" and "customs / fashion", at a different level from the headword level. You can understand the social background. The biggest feature of the present invention is to display the superordinate concept and assist the user in making a decision.

In order to make it easier to grasp the structure of the entire system, the above steps (11, 12, 13, 1)
4) shows an image in which data is actually stored in FIG.
Shown in FIG. 7 is a flowchart showing the operation of fashion creation support. The step of predicting the fashion based on the file arrangement in FIG. 6 is performed as follows (FIG. 7).

Step: S110) Copywriter 20
0 generates a plurality of buzzword candidates 210 (FIG. 6) and stores them in the memory. This is hereinafter referred to as “buzzword candidate”.

Step: S120) Copywriter 20
0 creates a description 220 for the buzzword candidate 210 and stores it on the memory.

Step: S130) The headwords are taken out from the dictionary 290 (for example, “Basic knowledge of modern terminology” issued by Jiyokuminsha) which is issued every year, and stored as a headword list in the memory. However, since the "basic knowledge of modern terms" is issued every year, the storage work of this headword is executed separately for each year. FIG. 6 shows headwords 211 and 2 stored for two years.
It is shown as 12. At this time, the above-mentioned buzzword candidate 210 is also registered as the headword 219 for the next fiscal year.

Step: S140) Also, the explanation of each headword written in “Basic knowledge of modern terminology” is written in explanation 2
21 and 222 are stored. The explanations may be stored collectively regardless of the year, but since the explanations may be different depending on the year, the explanations may be stored for each year as shown in FIG.

Step: S150) The superordinate concept for the headword is stored as superordinate concept 230. The superordinate concept is irrelevant to the year and may be stored collectively.

Step: S160) A buzzword is stored for each year. The buzzword is, for example, an award-winning work of the buzzword award selected by Jiyu Kokuminsha. Store for each year. The buzzword candidates will be buzzwords for the next fiscal year. Note that the buzzword does not necessarily have to be included in the headword. In FIG. 6, existing buzzwords are stored (241, 242) for each year.

Step: S170) The distance between the buzzword and the previous year is calculated by the aforementioned vector space method or the like, and several headwords closest to each buzzword are extracted. Specifically, the target of the distance calculation from the buzzword 210 next year is, for example, the headword 212 of the current year, which is the previous year. For the buzzword 242 of this year, last year's headword 2
Calculate the distance between 11 and.

Step: S180) The superordinate concept to which the several headwords thus extracted belong is extracted from the superordinate concept 230 as being close.

Step: S190) By outputting the above calculation results, the copywriter 200 determines which headword 212 of this year in which each of the created "next year's buzzwords" 210 is close to which superordinate concept 230 Can be checked. If the superordinate concept 230 to which the close headword belongs reflects the social situation of this year, it can be determined that the “buzzword of next year” may be hit. The past buzzwords 242 and 241 are compared with the headwords of the previous year. This allows you to see what is behind the past epidemic.

Next, an embodiment of the fashion prediction method of the present invention shown in FIG. 2 will be described.

FIG. 2 is the same as FIG. 1 up to the upper concept selection step 16. However, the present embodiment is different from the embodiment of FIG. 1 in that the ratio of the superordinate concepts to which some words determined to be close from the calculation result belong is displayed. FIG.
Shows the configuration of the fashion prediction support method 2 of the second embodiment of the present invention. However, the steps that perform the same processing as the invention of FIG. 1 are given the same numbers.

In the trend prediction support method 2 shown in FIG. 2, a set of words such as buzzwords is defined as a “headword”, and the time lapse of each year or the like (hereinafter, this time lapse is referred to as “year”. The period is not limited to the fiscal year or calendar year)
A headword storage step 11 for storing in groups according to

An explanation sentence storage step 12 for storing the explanation for each entry word stored in the entry word storage step 11 as an “explanation sentence”;

An upper concept storing step 13 for storing the category to which the headword belongs as a “higher concept”;

[0099] Among the headwords, words that have become popular and noticed in the year are selected and stored as "buzzwords" for each year.

The distance between the word (buzzword) stored in the buzzword storage step 14 and the word (headword) stored in the headword storage step 11 is calculated using the description sentence described in the description storage step 12. Step 15 for calculating the distance between words to be calculated using

The upper-ranked word calculated in the inter-word distance calculation step 15 that is close to the word of a certain year stored in the buzzword storage step 14 and stored in the headword storage step 11 in the previous year Higher concept selection step 1 for selecting a certain number of concepts from the higher concept stored in the higher concept storage step 13
6,

For each superordinate concept selected in the superordinate concept selecting step 16, a superordinate concept ratio calculating step 20 for calculating a ratio for each word stored in the buzzword storage step 14,

The present invention comprises a ratio calculation result displaying step 21 for displaying the ratio of the superordinate concept for each word stored in the buzzword storage step 14 calculated in the superordinate concept ratio calculating step 20.

In FIG. 2, headword storage step 1
1, description storage step 12, upper concept storage step 1
3, the buzzword storage step 14, the inter-word distance calculation step 15, and the higher concept selection step 16 are the same as those in FIG.

The operation of the upper concept ratio calculating step 20 will be described below. FIG. 10 is a flowchart showing the operation of calculating the higher-level concept ratio.

Step: S201) In the broader concept selection step 16, the broader concepts to which words determined to be close to each buzzword belong are designated as c1 to cn. Here, n is the number of close words. In general, n is determined to be a predetermined number in advance.

Step: S202) The j-th element cj is
Since it corresponds to one of the superordinate concepts D1 to Dm stored in the superordinate concept 230, the number of appearances of each D is counted.

Step: S203) The number of appearances of each D is set to n
Divide by.

As described above, by outputting the result determined in this manner on the ratio calculation result display unit 21, the copywriter 200 determines that each of the “buzzwords of the next year” 210 created is close. Headword 21 of this year
The ratio of the superordinate concept to which 2 belongs can be checked. If the headword that is considered to be close has a high percentage of high-level concepts that reflect the social situation of this year, it can be determined that the “buzzword of next year” may be hit. The past buzzwords 242 and 241 are compared with the headwords of the previous year. This allows you to see what ’s behind the past fads,
You can know how the social situation that is the background is going up and down in the year.

FIG. 3 is a diagram showing another embodiment 3 of the fashion prediction support system of the present invention. This embodiment is the same as the embodiment of FIG. 2 up to the ratio calculation processing. However, the present embodiment is different from the embodiment of FIG. 2 in that the ratio is calculated over a plurality of years and the correlation between the superordinate concepts is calculated.

FIG. 3 shows the processing steps of the fashion prediction support system according to the third embodiment of the present invention, and the same parts as those in FIGS. 1 and 2 are denoted by the same reference numerals.

The fashion prediction support system 3 shown in FIG.

A set of words such as buzzwords is referred to as a “headword” and the time lapse of each year or the like (hereinafter, this time lapse is referred to as the “year”, but the period is not necessarily limited to the fiscal year or calendar year). Heading storage step 11 for storing in a grouped manner according to

An explanation sentence storage step 12 for storing the explanation for each entry word stored in the entry word storage step 11 as “explanatory sentence”;

An upper concept storing step 13 for storing a word (category) corresponding to a semantic upper concept to which the headword belongs as an “upper concept”;

A buzzword storage step 14 for selecting a word that has become popular and noticed in the year among the headwords and storing it as a “buzzword” for each year;

The distance between the word (buzzword) stored in the buzzword storage step 14 and the word (headword) stored in the headword storage step 11 is stored in the description sentence stored in the description storage step 12. An inter-word distance calculation step 15 to be calculated using;

The headword storage step 11 calculated in the inter-word distance calculation step 15 and determined to be close to the word of a certain year stored in the buzzword storage step 14
A superordinate concept selecting step 16 for selecting a certain number of superordinate concepts for words in the previous year stored from the superordinate concept stored in the superordinate concept storing step 13;

For each superordinate concept selected in the uppermost concept selecting step 16, a superordinate concept ratio calculating step 20 for calculating a ratio to each word stored in the buzzword storage step 14,

A multi-year correlation calculating step 30 for calculating a correlation for each superordinate concept with respect to the ratio calculated over a plurality of years by the superordinate concept ratio calculating step 20;

A correlation calculation result display step 31 for displaying the result calculated in the multi-year correlation calculation step 30 is provided. The operation of the multi-year correlation calculation step 30 will be described later.

In FIG. 11, the headword storage means 41,
The explanation sentence storage unit 42, the upper concept storage unit 43, the buzzword storage unit 44, the inter-word distance calculation unit 45, the upper concept selection unit 46, and the higher concept ratio calculation unit 90 are the same as those in FIG.

The multi-year correlation calculation step 30 in FIG. 9 and the multi-year correlation calculation means 111 in FIG.
It has a function of calculating the correlation coefficient between the superordinate concepts by calculating the annual change of the ratio value for each superordinate concept.

FIG. 12 is a flowchart showing the operation of multi-year correlation coefficient calculation.

Step: S301) Superordinate concept ratio calculating step (20) or superordinate concept ratio calculating means (9)
According to 0), the ratio of the superordinate concept for the headword in the previous year is calculated for the buzzword in each year. As a result, for example, the superordinate concept “economy and management”
From 9 to 1999, it is possible to calculate what percentage of each year (out of the 10 superordinate concepts) it had.

Step: S302) Regarding the transition of the ratio of each superordinate concept calculated in this way, the correlation between the superordinate concepts is calculated by a multi-year correlation calculating step (30) or a multi-year correlation calculating means (111). Is calculated by
It should be noted that if there are ten superordinate concepts, there are as many correlation coefficients as there are combinations taking two out of ten.

The correlation coefficient calculated by the multi-year correlation calculation step 30 or the multi-year correlation calculation means 111 is output to the correlation calculation result display step (31) or the correlation calculation result display means 112. Accordingly, the copywriter 200 can examine the transition of the ratio of the superordinate concepts and the correlation between the superordinate concepts.
If each of the created “next year's buzzwords” 249 shows an increasing trend in the percentage of the superordinate concepts to which the headword 212 of this year that is considered to be close is increasing, the “next year's buzzword” is It can be determined that there is a possibility. This allows
Learn what's behind the past epidemic.

Although the first, second and third embodiments have been described based on the configurations shown in FIGS. 4, 9 and 11, the headword storage means 41 and the description storage means 4
2. Upper concept storage means 43 and buzzword storage means 44
Can be realized by a program for registering a headword, an explanatory sentence, a general concept, and a buzzword, respectively. Also,
The operations of the word-to-word distance calculation means 45, the high-level concept selection means 46, the high-level concept display means 47, the high-level concept ratio calculation means 90, the ratio calculation result display means 91, the multi-year correlation calculation means 111, and the correlation calculation result display section 112 are programmed. And a floppy (registered trademark) connected to a computer used as a buzzword prediction support device
The present invention can be easily implemented by storing it in a portable storage medium such as a disk or a CD-ROM and installing it when implementing the present invention.

Note that the present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

[0130]

As described above, according to the present invention, it is possible to order new buzzwords created by copywriters.

In addition, since the result of the ordering is reported in a form such as what high level concept is close, the user can understand why the score is high.

Further, the copywriter can select a promising buzzword from his own idea.

In addition, a concept that is likely to become popular in the future can be submitted as a prototype, and whether it is “hit” can be evaluated numerically, and can be used as product development support.

[Brief description of the drawings]

FIG. 1 is a diagram (1) for explaining the principle of the present invention.

FIG. 2 is a diagram (2) for explaining the principle of the present invention;

FIG. 3 is a diagram (3) for explaining the principle of the present invention;

FIG. 4 is a diagram (4) for explaining the principle of the present invention;

FIG. 5 is an example of a headword and an explanatory sentence.

FIG. 6 is a diagram for explaining the entire data structure.

FIG. 7 is a flowchart showing the operation of the present invention.

FIG. 8 is a flowchart illustrating an operation of calculating a distance between words.

FIG. 9 is a diagram (5) for explaining the principle of the present invention;

FIG. 10 is a flowchart of a calculation of a superordinate concept ratio.

FIG. 11 is a diagram (6) for explaining the principle of the present invention;

FIG. 12 is a flowchart of a correlation coefficient calculation.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 epidemic support system of the present invention (1) 2 epidemic support system of the present invention (2) 3 epidemic support system of the present invention (3) 4 epidemic prediction device of the present invention (1) 9 epidemic prediction device of the present invention (2) 10 Fashion Prediction Device of the Present Invention (3) 11 Headword Storage Step 12 Explanation Sentence Storage Step 13 Top Concept Storage Step 14 Vogue Word Storage Step 15 Inter-word Distance Calculation Step 16 Top Concept Selection Step 17 Top Concept Display Step 20 Top Concept Ratio Calculation step 21 Ratio calculation result display step 30 Multi-year correlation calculation step 31 Correlation calculation result display step 41 Headword storage means 42 Explanation sentence storage means 43 Superior concept storage means 44 Vogue word storage means 45 Inter-word distance calculation means 46 Superior concept selection Means 47 Higher concept display means 90 Higher concept ratio calculation means 91 Ratio calculation result display means 111 Multi-year correlation calculation means 112 Correlation calculation result display means

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shigeo Kaneda Karasuma Higashiiri Le, Imadegawa-dori, Kamigyo-ku, Kyoto Higashiiru School Corporation, Doshisha University Graduate School of Policy Science (72) Inventor Sadahiro Ikeda Imadegawa-dori, Kyoto City, Karasuma Higashiiriru School Corporation and Doshisha University Graduate School of Policy Science F-term (reference) 5B009 QA01 QA12 5B049 AA06 BB49 CC11 DD01 DD05 EE01 FF03 FF09 GG04 GG07 5B075 MM11 ND03 NK32 NR12

Claims (12)

[Claims] 1. A set of words including a buzzword is used as an entry word, and the time lapse including each year (hereinafter, this time lapse is called “year”, but the period is not necessarily limited to the fiscal year and the calendar year) The headword storage step stores the headword stored in the headword storage step as an explanatory sentence. A high-level concept storing step of storing a high-level concept for each of the stored headwords, and selecting a word that has become popular and noticed during the year among the headwords stored in the headword storage step,
A buzzword storage step of storing the word as a buzzword, a distance between a word of a certain year stored in the buzzword storage step, and a word in a year before the corresponding year stored in the headword storage step. Is calculated based on the explanation sentence stored in the explanation sentence storage step, and the distance is shorter by the inter-word distance calculation step from each word stored in the buzzword storage step. A superordinate concept selecting step of selecting a fixed number of superordinate concepts for each word stored in the headword storing step from the superordinate concepts stored in the superordinate concept storing step. A superordinate concept display step of displaying the superordinate concept selected by the step. And wherein the epidemic prediction support method. 2. In the step of calculating a distance between words,
The method according to claim 1, wherein the vector space method is used. 3. A set of words including a buzzword is used as an entry word, and the time lapse including each year (hereinafter, this time lapse is referred to as “year”, but the period is not necessarily limited to the fiscal year and the calendar year) Are not grouped)
A headword storage means for storing a headword, an explanation sentence storage means for storing an explanation for each headword stored in the headword storage means as an explanatory sentence, and each of the headwords stored in the headword storage means A high-level concept storage unit that stores a high-level concept for the headword, and a word that has become popular and noticed during the year among the headwords stored in the headword storage unit, and stores the word as a buzzword A buzzword storage means, a distance between a word of a certain year stored in the buzzword storage means and a word in a year before the corresponding year stored in the headword storage means, An inter-word distance calculating means for calculating based on the stored explanatory sentence, and a total using the inter-word distance calculating means from each word stored in the buzzword storage means. The upper-level concept selecting means for selecting a fixed number of higher-level concepts for each word stored in the headword storage means from the higher-level concepts stored in the higher-level concept storage means, wherein An epidemiological prediction support device, comprising: a superordinate concept display means for displaying the superordinate concept selected by the superordinate concept selecting means. 4. The fashion prediction support apparatus according to claim 3, wherein a vector space method is used for said inter-word distance calculation means. 5. A set of words including a buzzword is used as an entry word, and a time lapse including each year (hereinafter, this time lapse is referred to as a “year”, but the period is not necessarily limited to a fiscal year and a calendar year). The headword storage process stores the headword stored in the headword storage process as an explanatory sentence. The headword storage process stores the description of each headword stored in the headword storage process as an explanatory sentence. Storing a high-level concept for each of the stored headwords, a high-level concept storage process, and selecting a word that has become popular and noticed during the year among the headwords stored by the headword storage process;
A buzzword storage process for storing the word as a buzzword, a distance between a word of a certain year stored by the buzzword storage process, and a word stored in the headword storage process in a year before the relevant year. Is calculated based on the explanation sentence stored by the explanation sentence storage process, and using the inter-word distance calculation process from each word stored by the buzzword storage process. A high-level concept selection process of selecting a fixed number of high-level concepts for each word stored in the headword storage process, the distance being determined to be short by the calculation, from the high-level concepts stored in the high-level concept storage means; A superordinate concept display process for displaying the superordinate concept selected by the superordinate concept selection process. Storage medium storing the epidemic prediction support program characterized by. 6. A storage medium storing a fashion prediction support program according to claim 5, wherein a vector space method is used for the inter-word distance calculation process. 7. A trend prediction support method for predicting a trend from buzzword candidates and selecting a buzzword, wherein a superordinate concept for each buzzword according to claim 1 selected using the method of claim 1. , A superordinate concept ratio calculating step of calculating a ratio to each buzzword of claim 1 for each superordinate concept, and a ratio calculation result displaying step of displaying a ratio calculated by the superordinate concept ratio calculating step. A fashion prediction support method characterized by having: 8. A trend prediction support apparatus for predicting a trend from a trend word candidate and selecting a trend word, wherein each trend word is selected by using the means of claim 3. Calculating a ratio of each superordinate concept to each buzzword according to claim 3;
An epidemic prediction support device comprising: a high-level concept ratio calculating unit; and a ratio calculation result display unit for displaying a ratio obtained by calculation. 9. A trend prediction support program for predicting a trend from buzzword candidates and selecting a buzzword, for each buzzword according to claim 5 selected using the superordinate concept selection process according to claim 5. A superordinate concept ratio calculating process for calculating a ratio of each buzzword to each buzzword for each superordinate concept, and a ratio calculation result displaying process for displaying a ratio obtained by the calculation. A storage medium that stores a fashion prediction support program. 10. A fashion prediction support method for selecting a buzzword by predicting a buzzword from buzzword candidates, wherein a superordinate concept for each buzzword according to claim 1 selected using the method of claim 1. A high-level concept ratio calculating step of calculating a ratio for each buzzword according to claim 1 for each high-level concept; and calculating a correlation with respect to the ratio for a plurality of years obtained by the high-level concept ratio calculating step, for a plurality of years. A fashion prediction support method, comprising: a correlation calculation step; and a correlation calculation result display step of displaying a correlation coefficient obtained in the multi-year correlation calculation step. 11. A trend prediction support apparatus for predicting a trend from a trend word candidate and selecting a trend word, wherein each trend according to claim 3 is selected by using the high-level concept selecting means according to claim 3. For the superordinate concept for each word,
4. A high-level concept ratio calculating means for calculating a ratio of each buzzword to each buzzword for each high-level concept, and a multi-year correlation calculation for calculating a correlation with respect to a plurality of years obtained by the high-level concept ratio calculating means. Means, and a correlation coefficient display means for displaying a correlation coefficient obtained by the plurality of correlation calculation means. 12. A trend prediction support program for predicting a trend from buzzword candidates and selecting a buzzword, for each buzzword according to claim 5, selected using the superordinate concept selection process according to claim 5. A superordinate concept ratio calculating process for calculating a ratio to each buzzword of each of the superordinate concepts, and a correlation with respect to a plurality of years determined by the superordinate concept ratio calculating process. A multi-year correlation calculation process; and a correlation calculation result display process for displaying a correlation coefficient obtained by the multi-year correlation calculation process.