JP2020137005A - Data processing device and data processing method - Google Patents

Abstract

To appropriately classify a distributed program, or the like on the basis of an evaluation of a contactor.SOLUTION: A data processing device stores evaluation data indicating an evaluation for a distributed content of a contactor who contacts the distributed content in each of the distributed content and in each contactor, stores meta data indicating explanatory information for explaining the distributed content in each distributed content, specifies a corresponding relationship between an index led from the evaluation and the explanatory information on the basis of the evaluation data in each distributed content and each contactor and the meta data in each distributed content, adopts the explanatory information indicated by an objective meta data to the corresponding relationship as an object of each of a plurality of distributed contents, and thereby identifying the index of the object and classifying each of the plurality of distributed contents into any one of a plurality of groups on the basis of an identified index.SELECTED DRAWING: Figure 7

Description

The present invention relates to a data processing device and a data processing method for processing data related to the delivered contents delivered in the past, and in particular, a metadata showing explanatory information of the delivered contents and a contact person who has come into contact with the delivered contents. The present invention relates to an evaluation data indicating an evaluation of the delivered contents, a data processing device for processing, and a data processing method.

Programs and commercials broadcast on television or radio, and digital contents and the like (hereinafter, programs and the like) distributed on the Internet may be classified based on genres and attributes. The classification of programs, etc. in this way makes it easier for contacts of the programs, etc. (specifically, TV listeners, radio listeners, Internet users, etc.) to find programs, etc. that match their tastes. It is useful in. In addition, the classification of programs, etc. is useful for advertisers when planning advertisements. For example, if an appropriate classification is selected and the program is placed in the broadcast time zone of the program corresponding to the classification, the advertisement can be efficiently placed. Can be done.

In addition, as a classification of programs and the like, a classification targeting programs and the like scheduled to be distributed in the future can be mentioned, and a classification targeting programs and the like already distributed can be considered. The latter classification is required, for example, when analyzing the contact status of distributed programs (specifically, the audience rating, listening rate, number of Internet users, etc.), and specifically, the contact status is classified. It is used when analyzing separately.

By the way, the classification of programs and the like is generally performed according to general attribute items, specifically, genre, performers, contents, broadcast (delivery) date and time, distribution source such as broadcasting station, and the like. (See, for example, Patent Document 1). In Patent Document 1, the metadata of a TV program is analyzed, attribute items such as the broadcast time, genre, and performers of the TV program are specified, and the TV program is classified based on the specified attribute items. .. Further, Patent Document 1 also discloses that a program title included in program metadata is morphologically analyzed and a television program is classified based on the analysis result.

Japanese Unexamined Patent Publication No. 2007-20309

Examples of the information used when classifying the delivered programs and the like include explanatory information (meta information) of the programs and the like as described above. On the other hand, as the information for classification, the evaluation of the program or the like by the contact person of the program or the like (for example, the impression and the viewing feeling of the program or the like) can be considered. If programs and the like can be classified based on the evaluation of the contact person, the utility value of the classification result will increase, and for example, it will be possible to find a plurality of programs and the like whose psychological states of the contact person at the time of contact are similar to each other. ..

However, if the contact person changes, the evaluation may change even if the program or the like is the same. Therefore, when classifying the program or the like based on the contact person's evaluation, it is necessary to consider the above circumstances. That is, regarding the classification of programs and the like based on the evaluation of the contact person, it is required to appropriately classify the programs and the like that have already been distributed in consideration of the various evaluations of the contact person.

Therefore, the present invention has been made in view of the above circumstances, and an object of the present invention is to solve the following object.
Specifically, the present invention provides a data processing apparatus and a data processing method capable of solving the above-mentioned problems of the prior art and appropriately classifying the distributed programs and the like based on the evaluation of the contact person. The purpose is to do.

In addition, when the delivered programs, etc. are grouped (classified) based on the evaluation of the contact person, if it is possible to determine which group the program, etc. scheduled to be delivered in the future belongs to, the contact person for the program, etc. scheduled to be delivered. (Strictly speaking, the prospective contact person) can be inferred. A data processing device having such a function is particularly useful in the broadcasting industry, the advertising industry, and the like, and its development is expected.
Therefore, another object of the present invention is a data processing device capable of estimating evaluations (for example, impressions) of programs scheduled to be distributed in the future by using classification (grouping) of programs or the like that have already been distributed. It also aims to realize.

In order to achieve the above object, the data processing apparatus of the present invention stores evaluation data indicating the evaluation of the delivered content by the contact person who has come into contact with the delivered content for each of the delivered contents and for each of the contacts. The evaluation data storage unit to be used, the metadata storage unit that stores the metadata indicating the explanatory information for explaining the delivered contents for each of the delivered contents, and the said for each of the delivered contents and the contact person. Based on the evaluation data and the metadata for each delivered content, the specific unit that specifies the correspondence between the index derived from the evaluation and the explanatory information, and each of the plurality of delivered contents are targeted. By applying the explanatory information indicated by the metadata of the target to the correspondence, a plurality of the delivered items have been distributed based on the indexing unit for calculating the index of the target and the index calculated by the indexing unit. It is characterized by having a classification unit that classifies each of the contents into any of a plurality of groups.

The data processing apparatus of the present invention configured as described above classifies the delivered contents based on the evaluation of the contact person with respect to the delivered contents (programs and the like). More specifically, the data processing apparatus of the present invention identifies the correspondence between the index derived from the evaluation and the explanatory information based on the metadata indicating the explanatory information of the delivered contents and the evaluation data for each contact. To do. This correspondence reflects the evaluation of each contact person for each delivered content.
Then, in the data processing content of the present invention, an index is calculated for each of the plurality of delivered contents according to the above correspondence, and the plurality of delivered contents are grouped based on the calculated index. According to the above configuration, since the delivered contents are classified by applying the above correspondence, each delivered content is based on the contact's evaluation while considering that the contact's evaluation may change in various ways. It becomes possible to classify appropriately.

Further, in the above data processing device, the specific unit performs machine learning using the evaluation data for each of the delivered contents and the contact, and the metadata for each of the delivered contents. , It is preferable to specify the correspondence by constructing a mathematical model showing the correspondence.
With the above configuration, it is possible to accurately identify the above-mentioned correspondence through the construction of a mathematical model by machine learning.

Further, in the above data processing device, the delivered content may be at least one of a broadcasted program and a broadcasted advertisement that have been broadcast in the past.
With the above configuration, at least one of the broadcasted programs and broadcasted advertisements broadcasted in the past can be classified based on the evaluation of the contact person (specifically, the viewer, the listener, etc.). It will be possible.

Further, in the data processing device, a classification category is preset for the delivered content, and the classification unit sets each of the plurality of delivered contents belonging to the classification category for each classification category. , It is more preferable to classify into one of the plurality of groups based on the index calculated by the indexing unit.
With the above configuration, since a plurality of delivered contents belonging to the classification category are classified for each classification category, systematic classification with the classification category as a major classification becomes possible, and the convenience of the classification result is improved. ..

Further, in the above data processing device, it is more preferable that the evaluation data is data showing the result of a questionnaire regarding the impression of the contact person with respect to the delivered content. With such a configuration, it is possible to classify each of the plurality of delivered contents based on the impressions of the contact person.
Further, it is more preferable that a part of the evaluation data showing the result of the questionnaire is used as verification data for the validity of the correspondence. With such a configuration, the validity of the correspondence can be verified, so that the credibility of the grouping using the correspondence is high.

Further, in the above data processing device, an input information acquisition unit that acquires input information input as the explanatory information of the distribution schedule content to be distributed in the future, and a group to which the distribution schedule content belongs among the plurality of groups. The determination unit has a determination unit for determining, and the indexing unit calculates the index of the distribution schedule content by applying the input information to the correspondence, and the determination unit calculates the indexed distribution schedule. It is even more preferable to determine the group to which the scheduled delivery content belongs based on the index of the content.
With the above configuration, an index of the scheduled delivery content is calculated from the explanatory information of the scheduled delivery content to be delivered in the future, and the group to which the scheduled delivery content belongs is determined based on the index. As described above, according to the above configuration, it is possible to grasp in advance which group the program or the like scheduled to be distributed in the future belongs to.

Further, in the above data processing device, the input information acquisition unit acquires the input information for each of the first distribution schedule contents and the second distribution schedule contents that are different from each other, and the indexing unit obtains the first distribution schedule. For each of the content and the second delivery schedule content, the input information is applied to the correspondence to determine the index, and the determination unit divides each of the first distribution schedule content and the second distribution schedule content. Based on the issued index, the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs are determined, and the group to which the first scheduled delivery content belongs and the second scheduled delivery content It is even more preferable to have a similarity degree calculation unit for calculating the similarity degree with the group to which the member belongs.
According to the above configuration, the group can be determined for each of the first delivery scheduled content and the second delivery scheduled content to be delivered in the future. Further, in the above configuration, the degree of similarity between the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs can be calculated. As a result, matching between groups becomes possible. For example, to a person who distributes a certain first distribution schedule content, a second distribution schedule content belonging to a group having a high degree of similarity to the group to which the first distribution schedule content belongs. Can be presented.

Further, in the above data processing device, it is more preferable that the first scheduled distribution content is an advertisement to be broadcast in the future and the second scheduled distribution content is a program to be broadcast in the future.
In the above configuration, groups are determined for each of the advertisements and programs to be broadcast in the future, and the degree of similarity between the groups is calculated. This makes it possible to match the group to which the advertisement to be broadcast in the future belongs and the group to which the program to be broadcast in the future belongs. For example, for a person who plans to place an advertisement, the group to which the advertisement belongs It is possible to present programs that belong to a group with a high degree of similarity. Then, if the above advertisement is broadcast during the broadcast time of the presented program, the advertising effect can be improved.

Further, the data processing apparatus may include a quantification processing unit that performs a process of quantifying the explanatory information indicated by the metadata. In such a configuration, the similarity degree calculation unit includes the explanatory information of the first distribution schedule content quantified by the quantification processing unit and the second distribution schedule content quantified by the quantification processing unit. It is more preferable to calculate the degree of similarity between the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs based on the explanatory information.
In the above configuration, when calculating the degree of similarity between the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs, the numerical explanation information of each delivered scheduled content is used, so that the degree of similarity is determined. It becomes possible to calculate appropriately.

Further, in the above data processing apparatus, it is more preferable that the specific unit specifies the corresponding relationship between the index and the numerically quantified explanatory information.
In the above configuration, it is possible to more appropriately identify the correspondence relationship by using the numerical explanatory information.

Further, in order to solve the above-mentioned problems, in the data processing method of the present invention, the storage device provides evaluation data indicating the evaluation of the delivered contents by the contact person who has come into contact with the delivered contents for each delivered content and. Storage is performed for each contact, the storage device stores metadata indicating explanatory information for explaining the delivered content for each delivered content, and the computer stores each of the delivered content and the contact. Based on the evaluation data for each of the delivered contents and the metadata for each delivered content, the correspondence relationship between the index derived from the evaluation and the explanatory information is specified, and the computer determines the correspondence between the plurality of delivered contents. By applying the explanatory information indicated by the metadata of the target to the correspondence, the index of the target is calculated, and the computer determines the index, and the computer determines the plurality of distributions based on the index. It is characterized in that each of the completed contents is classified into one of a plurality of groups.
According to the above method, it is possible to appropriately classify each of the plurality of delivered contents based on the contact's evaluation while considering that the contact's evaluation may change.

According to the present invention, it is possible to appropriately classify each of a plurality of delivered contents based on the evaluation of the contact, while considering that the evaluation of the contact may change.
Further, according to the present invention, the contents scheduled to be delivered in the future can be grouped in the same manner as the delivered contents, so that the evaluation of the contact person (strictly speaking, the contact person) with respect to the delivered contents is predicted. It is possible.

It is a figure which shows the provider of the service using the data processing apparatus of this invention, and the person concerned. It is explanatory drawing of the service using the data processing apparatus of this invention. It is a figure which shows an example of the structure of the system including the data processing apparatus of this invention. It is a block diagram which shows the function of a data processing apparatus. It is a conceptual diagram which shows the image of grouping. It is explanatory drawing about the processing with respect to the explanatory information of the program / advertisement belonging to a group. It is a figure which shows the flow of a grouping process. It is a figure which shows the flow of a group determination process. It is a figure which shows the flow of the similarity degree calculation process. It is a figure which shows the contact screen of the recommended advertising space.

The investigation result processing apparatus and the investigation result processing method according to one embodiment of the present invention (the present embodiment) will be described in detail below with reference to the accompanying drawings.
It should be noted that the embodiments described below are merely examples for facilitating the understanding of the present invention, and do not limit the present invention. That is, the present invention may be modified or improved from the embodiments described below without departing from the spirit of the present invention. Moreover, as a matter of course, the present invention includes an equivalent thereof.

Further, in the present specification, the term "device" refers to one device that independently exerts a specific function, or a plurality of devices that exist in a dispersed manner but cooperate to exert a specific function. Also includes.

Further, in the following description, the “content” distributed / distributed is the content distributed / distributed to the public through the media, and specifically, a television (including Internet television) program. And advertisements, radio (including IP simul radio) programs and advertisements, and Web contents and Web advertisements.

In the following, as an example of the “content” to be distributed / distributed, a TV program (corresponding to a program) and a TV commercial (corresponding to an advertisement) will be described. However, the structure and effect of the present invention can, of course, be applied to contents other than TV programs and TV commercials.

Further, in the following description, "contact" corresponds to the act of watching a television program and a television commercial, and "contact" means a viewer of the television program and the television commercial. Here, "viewing" means not only viewing broadcast programs and advertisements in real time, but also recording programs and advertisements, playing them back within a certain period of time, and viewing those distributed on the Web. Including so-called time viewing.

It should be noted that, for example, in the case of radio programs and radio commercials, the person who listens to the contents other than the television program and the television commercial corresponds to the contact person. In addition, for Web contents and Web advertisements, the person who uses (views) them corresponds to the contact person.

Further, in the following description, the viewer who is the "contact person" may be an individual or a group (for example, a household) to which the individual belongs.

<< About the service using the data processing device of this embodiment >>
Prior to the description of the data processing device and the data processing method of the present embodiment, the service using the data processing device of the present embodiment will be described with reference to FIGS. 1 and 2. FIG. 1 is a diagram showing a service provider using the data processing device of the present embodiment and related persons thereof. FIG. 2 is an explanatory diagram of a service using the data processing device of the present embodiment.

The data processing apparatus of this embodiment is used by the research company R illustrated in FIG. The research company R is a person who provides a service using the data processing device of the present embodiment, and specifically provides a recommendation service. The recommendation service is a service that presents recommended advertising space to a person who is considering purchasing an advertising space (for example, advertising company C shown in FIG. 1).

Here, the "recommended advertising space" is an advertising space for effectively hitting an advertisement scheduled to be broadcast in the future (hereinafter referred to as an advertisement scheduled to be broadcast). More specifically, the recommended ad space is the broadcast time of a program that is expected to be evaluated by viewers similar to the scheduled broadcast advertisement among the programs to be broadcast in the future (for example, the time zone with hatching in FIG. 2). ) Is the spot CM frame set in. The recommended advertising space is not limited to the spot CM frame, but may be a time CM frame (so-called sponsored frame).

In addition, in providing the recommendation service, the research company R classifies TV programs and TV commercials (hereinafter, broadcasted programs and broadcasted advertisements) that have been broadcast in the past, as shown in FIG. Specifically, the research company R acquires metadata (hereinafter, program metadata) related to the broadcasted program. The program metadata is explanatory information for explaining a broadcasted program, specifically data indicating the genre, performers, contents, title, etc. of the program, and can be obtained from, for example, the TV station T that broadcast the program. Is. The explanatory information indicated by the program metadata and the acquisition route of the program metadata are not particularly limited, and can be acquired from, for example, a program production company or the like.

In addition, the research company R acquires evaluation data (hereinafter, program evaluation data) indicating the evaluation of the broadcasted program. The program evaluation data is data indicating the evaluation of the broadcasted program by the viewer who has watched the broadcasted program. More specifically, the research company R conducts a questionnaire survey targeting the plurality of monitors M shown in FIG. In this questionnaire survey, each monitor M answers the impression of the TV program (broadcasted program) actually watched among the TV programs broadcasted during the survey period. The research company R acquires program evaluation data showing the questionnaire results of each monitor M.

Here, the "impression" is the emotional state of the viewer when the viewer actually watches the TV program (broadcast program), that is, the so-called context, and more specifically, the impression and impression of the TV program. The image, the situation and mood when watching the TV program, the emotional desire to expect from the TV program, and the feelings toward the performers of the TV program are mentioned. Specifically, the degree of favor, friendliness, and harassment. Excitement, excitement, uplifting feeling (feeling fun), freshness, usefulness, informativeness, spectacularness, topicality, and whether the content is funny (suitable for entertainment), etc. are applicable.
For example, as the above-mentioned "context", a change in feelings (emotions) that a viewer has when he / she repeats viewing behavior, for example, a TV program and a TV commercial that are continuously broadcast on a certain broadcasting station, or continuously. You may deal with the flow of feelings when you watch a plurality of TV programs broadcasted on TV.

In the present embodiment, the above-mentioned questionnaire survey is conducted in a net research format, and each monitor M answers the questionnaire using a terminal owned by each monitor M (for example, the monitor terminal 22 shown in FIG. 3). , The program evaluation data showing the result of the questionnaire is sent to the research company R through the above terminal. However, the present invention is not limited to this, and for example, each monitor M collects an entry form in which the answers to the questionnaire are entered, or obtains the questionnaire results by interviewing the investigator, and the obtained questionnaire results are used on a keyboard, touch panel, etc. The program evaluation data may be acquired by inputting through the input device of.

In addition, the research company R also acquires metadata (hereinafter referred to as advertisement metadata) and evaluation data (hereinafter referred to as advertisement evaluation data) for the broadcasted advertisement. The advertisement metadata is explanatory information for explaining the broadcasted advertisement, specifically, data indicating the category of the product handled by the broadcasted advertisement, the performers and contents of the broadcasted advertisement, and the like, for example, producing the advertisement. It can be obtained from the advertising company C. The explanatory information indicated by the advertisement metadata and the acquisition route of the advertisement metadata are not particularly limited.

The advertisement evaluation data is data showing impressions of the TV commercials actually viewed by each monitor M among the TV commercials (broadcasted advertisements) broadcast during the survey period. As with the program evaluation data, the advertisement evaluation data can be obtained from the terminal owned by each monitor M via the network, or the completed questionnaire and the results of the interview survey collected from each monitor M can be input. It can be obtained.

By the way, the broadcasted program and the broadcasted CM correspond to the "delivered content" in the present invention.

Then, the research company R classifies each of the plurality of TV programs broadcast in the past predetermined period (for example, the above-mentioned questionnaire survey period) into a plurality of groups based on the acquired program metadata and program evaluation data. To do. Similarly, the research company R divides each of the plurality of TV commercials broadcast in the past predetermined period (for example, the above-mentioned questionnaire survey period) into a plurality of groups based on the acquired advertisement metadata and advertisement evaluation data. Classify.

In addition, the research company R obtains explanatory information about a TV commercial (that is, a scheduled broadcast advertisement) scheduled to be broadcast in the future. Specifically, the research company R obtains the explanation information of the broadcast schedule advertisement from, for example, the advertising company C that handles the broadcast schedule advertisement, and inputs the information on the research company R side (that is, the explanation of the broadcast schedule advertisement). Get input information about information). The source of the explanatory information of the scheduled broadcast advertisement is not limited to the advertising company C, and may be, for example, an advertiser.

Then, as shown in FIG. 2, the research company R determines to which group among the plurality of advertisement groups that classify the broadcasted advertisements the broadcast scheduled advertisement belongs based on the acquired input information (strictly speaking). ,Predict. Hereinafter, the group to which the broadcast scheduled advertisement belongs will be referred to as an "advertisement group".

Further, the research company R obtains explanatory information about each of the TV programs scheduled to be broadcast in the future (hereinafter referred to as broadcast programs). Specifically, the research company R obtains, for example, explanatory information of programs scheduled to be broadcast after N months (N is an arbitrary natural number), more specifically, information on future TV program listings from each TV station T. , The information is input on the research company R side (that is, the input information regarding the explanation information of the program scheduled to be broadcast is acquired). The source of the explanatory information of the program scheduled to be broadcast is not limited to the TV station, and may be, for example, a program production company.

Then, as shown in FIG. 2, the research company R determines (strictly speaking) to which group among the plurality of program groups in which the broadcasted programs are classified belongs to each broadcast scheduled program based on the acquired input information. Predicts). Hereinafter, the group to which the program scheduled to be broadcast belongs will be referred to as a "program group".

By the way, the broadcast schedule advertisement and the broadcast schedule program correspond to the "delivery schedule content" in the present invention, and also correspond to the first distribution schedule content and the second distribution schedule content which are different from each other. More specifically, the scheduled broadcast advertisement is the first scheduled distribution content, and the scheduled broadcast program is the second scheduled distribution content.

Furthermore, the research company R performs matching between the advertisement group and the program group as shown in FIG. More specifically, the research company R identifies a program group similar to the advertising group (strictly speaking, the best match group described later). Here, the specified program group is a group in which the viewer's evaluation (strictly speaking, an index described later) for the broadcast scheduled program belonging to the program group is expected to be most similar to the broadcast scheduled advertisement.

After going through the series of processes up to the above, the research company R provides the recommendation service to the advertising company C. Specifically, the research company R identifies a program scheduled to be broadcast that belongs to a program group similar to the advertising group, and plans to broadcast the advertising space (spot CM frame) set during the broadcasting time of the program scheduled to be broadcast. It is proposed to advertising company C as a "recommended advertising space" for advertisements.

In the present embodiment, the implementing body of the above-mentioned recommendation service is the research company R, but the present invention is not limited to this, and for example, the television station T may implement the recommendation service, or It may be implemented by a third-party organization other than the advertising company C, the research company R and the television station T, or another service provider. Alternatively, the advertising company C may perform data processing corresponding to the recommendation service (specifically, data processing for specifying the "recommended advertising space") by itself.

In addition, the recommendation service is months, weeks, days, hours, minutes, or seconds before the date and time when the advertisement is broadcast using the "recommended inventory" proposed by the service. It is available before.

<< About the configuration of the data processing device according to this embodiment >>
Next, the configuration of the data processing device (hereinafter, data processing device 10) according to the present embodiment will be described with reference to FIG. FIG. 3 is a diagram showing an example of the configuration of a recommendation service providing system including the data processing device 10.

The data processing device 10 uses various data related to TV programs and TV CMs, specifically, program metadata, program evaluation data, advertisement metadata, and advertisement evaluation data, and carries out a series of data processing related to the recommendation service. In the present embodiment, the data processing device 10 is composed of a computer owned by the research company R (specifically, the server computer 11 shown in FIG. 3) and a storage device 12 connected to the server computer 11. Has been done.

The server computer 11 is a computer that forms the main body of the data processing device 10, and performs a series of data processing related to the recommendation service. As shown in FIG. 3, the server computer 11 includes a CPU 11a, a memory 11b including a ROM and a RAM, a communication interface 11c, a hard disk drive 11d, an input device 11e including a mouse and a keyboard, and a display and a printer. The output device 11f and the like.

Further, a data processing program (hereinafter referred to as a data processing program) is installed in the server computer 11. When this data processing program is read and executed by the CPU, a series of data processing related to the recommendation service is executed in the server computer 11.

In the present embodiment, the server computer 11 owned by the research company R constitutes the data processing device 10, but the present invention is not limited to this. That is, the data processing device 10 may be configured by a computer owned by an organization other than the research company R (for example, a television station T, an advertising company C, or a third-party organization). For example, an ASP (Application Service Provider) server may configure the data processing device 10, and the ASP server may perform a series of data processing related to the recommendation service as an ASP service.

The storage device 12 stores various data required for the recommendation service. Specifically, the storage device 12 stores the program metadata of the broadcasted program broadcasted in the past for each program, and evaluates the broadcasted program of the monitor M that has watched the broadcasted program (strictly speaking). Stores program evaluation data indicating (results of a questionnaire regarding impressions at the time of viewing) for each program and for each monitor M (viewer).

In the present embodiment, the program metadata is provided from the data distribution server 21 owned by each television station T via the network and stored in the storage device 12. Further, in the configuration shown in FIG. 3, the program evaluation data is acquired from the monitor terminal 22 owned by each monitor M via the network and stored in the storage device 12.

In addition, the storage device 12 stores the advertisement metadata of the broadcasted advertisement broadcasted in the past for each advertisement, and evaluates the broadcasted advertisement of the monitor M that has viewed the broadcasted advertisement (strictly speaking, at the time of viewing). It is stored for each advertisement and each monitor M (viewer) showing the result of the questionnaire regarding the impression of.

In the present embodiment, the advertising metadata is provided from the data distribution server 23 owned by the advertising company C via the network and stored in the storage device 12. Further, in the configuration shown in FIG. 3, the advertisement evaluation data is acquired from the monitor terminal 22 owned by each monitor M via the network and stored in the storage device 12. By the way, the source of the advertisement metadata is not limited to the advertising company C, but may be another person (for example, an advertiser).

In the present embodiment, the storage device 12 is communicably connected to the server computer 11. The server computer 11 accesses the storage device 12 when providing the recommendation service, reads various data in the storage device 12, and uses the storage device 12 for predetermined data processing.

In the configuration shown in FIG. 3, the storage device 12 is separated from the server computer 11, and is configured by, for example, a database server other than the server computer 11. However, the present invention is not limited to this, and the storage device 12 may be configured by an auxiliary storage device (for example, the hard disk drive 11d of the server computer 11) built in or externally attached to the server computer 11.

Next, the configuration of the data processing device 10 will be described again from the functional aspect with reference to FIG. FIG. 4 is a block diagram showing the functions of the data processing device 10.

As shown in FIG. 4, the data processing device 10 includes an evaluation data storage unit 31, a metadata storage unit 32, a digitization processing unit 33, a specific unit 34, an indexing unit 35, a classification unit 36, an input information acquisition unit 37, and a determination. It has a unit 38 and a similarity calculation unit 39 as functional units. Among these functional units, the evaluation data storage unit 31 and the metadata storage unit 32 are configured by the above-mentioned storage device 12. The other functional units (seven functional units) are realized by the cooperation of the hardware device of the server computer 11 and the data processing program installed in the server computer 11. In the present embodiment, all of the above-mentioned seven functional units are realized by one server computer 11, but the present invention is not limited to this, and some of the seven functional units are not limited to this. May be realized by another computer.

Hereinafter, each functional unit of the data processing device 10 will be described in detail.
(Evaluation data storage unit)
The evaluation data storage unit 31 stores the program evaluation data for the broadcasted program for each program and for each monitor M (that is, the viewer of the broadcasted program). Further, the evaluation data storage unit 31 stores the advertisement evaluation data for the broadcasted advertisement for each advertisement and for each monitor M (that is, the viewer of the broadcasted advertisement).

In addition, regarding the program evaluation data and the advertisement evaluation data, as described above, these evaluation data are data showing the results of a questionnaire regarding the impressions of the viewer (monitor M) on the broadcasted program / advertisement. is there. To explain in more detail, the questionnaire provides multiple answer candidates regarding impressions (for example, "exciting", "as expected", "mannered content", "laughing", and "I don't want to see it anymore"). There is. The monitor M, which has actually watched the broadcasted program, responds to the impression of the broadcasted program by selecting the corresponding candidate from the above-mentioned plurality of answer candidates.

By the way, in the present embodiment, the questionnaire conducted to acquire the above evaluation data is designed so as to ensure the reliability of the data, and specifically, to ensure the representativeness. Questionnaire respondents are selected, and each question in the questionnaire is also asked in appropriate expressions.

(Metadata storage)
The metadata storage unit 32 stores the program metadata of the broadcasted program for each program, and stores the advertisement metadata of the broadcasted advertisement for each advertisement. Here, to add to the program metadata and the advertisement metadata, as described above, each data includes text information indicating the performer and the content (image) as explanatory information.

In addition, the program metadata further includes a program genre as explanatory information of the broadcasted program. In addition, the advertisement metadata further includes the category of the product handled by the advertisement as the explanatory information of the broadcasted advertisement. The program genre indicated by the program metadata and the product category indicated by the advertisement metadata correspond to a preset classification category for the broadcasted program, and correspond to a major classification when the broadcasted program is classified.
In addition, each broadcasted program may be classified into any one genre, or may be classified into a plurality of genres.

(Numerical processing unit)
The quantification processing unit 33 performs a process of quantifying the explanatory information indicated by the program metadata and the advertisement metadata. Specifically, the digitization processing unit 33 vectorizes the explanatory information (strictly speaking, text information indicating the performers, contents, etc.) indicated by each metadata. More specifically, the digitization processing unit 33 digitizes (vectorizes) the performer information by applying word2vec to the performers indicated by each metadata. Further, the digitization processing unit 33 digitizes (vectorizes) the content information by applying LDA (Latent Dirichlet Allcation) to the content indicated by each metadata.

As described above, in the present embodiment, the explanatory information indicated by each metadata is digitized (vectorized), which makes it easier to handle the explanatory information in the subsequent processing. Specifically, the data using the explanatory information. It becomes easier to perform processing (for example, machine learning described later). However, the present invention is not limited to this, and at the time of machine learning, the explanatory information may be used as it is (that is, in a state where it is not quantified).

The method for digitizing (vectorizing) the explanatory information indicated by the metadata is not limited to the above-mentioned method, and if there is another method that can be preferably used, it can be freely selected and used. can do.

(Specific part)
The specific unit 34 specifies the correspondence between the index and the explanatory information regarding the broadcasted program based on the program evaluation data for each program and each monitor M and the program metadata for each program. Similarly, the specific unit 34 specifies the correspondence between the index and the explanatory information regarding the broadcasted advertisement based on the advertisement evaluation data for each advertisement and each monitor M and the advertisement metadata for each advertisement.

Here, the index will be described. The index is derived from the evaluation indicated by the evaluation data obtained for a certain program / advertisement. Specifically, the index is composed of each of the impressions indicated by the evaluation data obtained for a certain program / advertisement and the numerical value corresponding to each impression, for example, "Pounding 3.0" and "Laughing". It is expressed by a notation such as "1.8". As for the numerical value according to the impression, a known technique for quantifying the impression can be used, and for example, the impression can be quantified by using LDA (Latent Dirichlet Allcation).
In addition, when deriving an index for a certain program / advertisement, the index is derived by adding the explanatory information shown by the metadata about the program / advertisement in addition to the evaluation indicated by the evaluation data obtained for the program / advertisement. You may.

The specific unit 34 of the present embodiment obtains all the indexes of the broadcasted programs broadcasted in the predetermined period (for example, the above-mentioned questionnaire survey period), and specifies the correspondence between the indexes of all the broadcasted programs and the explanatory information. To do. Similarly, the specific unit 34 of the present embodiment obtains all the indexes of the broadcasted advertisements broadcasted in the above-mentioned predetermined period, and specifies the correspondence between the indexes of all the broadcasted advertisements and the explanatory information.

More specifically, the specific unit 34 performs machine learning for each broadcasted program broadcasted in a predetermined period using program evaluation data for each program and each monitor M and program metadata for each program. Build a mathematical model that shows the correspondence between indicators and explanatory information about broadcasted programs. In addition, the specific unit 34 performs machine learning for each broadcasted advertisement broadcasted in a predetermined period by using the advertisement evaluation data for each advertisement and each monitor M and the advertisement metadata for each advertisement, and the broadcasted advertisement. Build a mathematical model that shows the correspondence between the index and the explanatory information.

By the above procedure, the correspondence between the index and the explanatory information is specified for each of the broadcasted program and the broadcasted advertisement. In the above-described machine learning, in the present embodiment, metadata showing explanatory information digitized (vectorized) by the digitizing processing unit 33 is used. That is, the specifying unit 34 of the present embodiment specifies the correspondence between the index and the digitized (vectorized) explanatory information. However, as described above, the explanatory information may not be quantified at the stage of machine learning, in other words, the correspondence between the index and the non-quantified explanatory information may be specified.

By the way, regarding machine learning, known methods such as ID3 (Iterative Dichotomiser 3), correlation rule learning, neural network, deep learning, genetic programming, functional logic programming, support vector machine, clustering, principal component analysis, cluster analysis , Bayesian networks, extreme learning machines, etc. are available. In addition, machine learning methods that may be developed in the future are also available.

Further, in the present embodiment, regarding the mathematical model constructed by machine learning, the validity of the correspondence between the index and the explanatory information is appropriately verified by using a part of the above-mentioned evaluation data as teacher data (verification data). To do. As a result, the credibility of the processing result is ensured for the data processing (specifically, grouping described later) performed using the above correspondence.
The verification of the validity of the correspondence is not limited to the verification method using a part of the evaluation data as the teacher data, but the statistical verification method (for example, the holdout method, the intersection confirmation method, the boost trap method, etc.) ) May be done.

(Indexing part)
The indexing unit 35 applies the correspondence (specifically, a mathematical model) specified by the specific unit 34 to the broadcasted programs and the broadcasted advertisements broadcast during the predetermined period (for example, the questionnaire survey period described above). Determine the index. Specifically, the indexing unit 35 targets all the broadcasted programs broadcasted in the predetermined period, and applies the explanatory information indicated by the program metadata of each broadcasted program to the above mathematical model. As a result, the index (that is, the impression and the degree thereof) of each broadcasted program to be targeted is calculated.

Similarly, the indexing unit 35 targets all the broadcasted advertisements broadcast during the above-mentioned predetermined period, and applies the explanatory information indicated by the advertisement metadata of each of the targeted broadcasted advertisements to the above-mentioned mathematical model. As a result, the index (that is, the impression and the degree thereof) of each broadcast advertisement that is the target is calculated.

Further, the indexing unit 35 of the present embodiment can calculate these indexes for the broadcast scheduled advertisement and the broadcast scheduled program to be broadcast in the future. Such a function will be described later.

(Classification department)
The classification unit 36 groups (classifies) the target when the index is calculated by the indexing unit 35, that is, the broadcasted program and the broadcasted advertisement broadcast in a predetermined period. Specifically, the classification unit 36 classifies each of all the broadcasted programs broadcasted in the predetermined period into one of a plurality of groups based on the index calculated for each broadcasted program. At least one broadcast program belongs to each group. In other words, each broadcasted program is classified into one or more groups.

More specifically, the classification unit 36 refers to the program metadata of each broadcasted program broadcasted in the predetermined period, and each broadcasted program broadcasted in the predetermined period according to the program genre indicated by the program metadata. Is divided into genres. After that, the classification unit 36 classifies each of the plurality of broadcast programs belonging to each genre into one of the plurality of groups based on the index calculated by the indexing unit 35 for each genre.

The classification (grouping) based on the index may be performed for all program genres, or the top n places (n is a natural number of 2 or more) when the number of broadcasted programs belonging to each program genre is calculated. ) May be implemented only for the program genres up to. Further, the type of the genre to which each broadcasted program belongs may be one type or a plurality of types.

Similarly, the classification unit 36 refers to the advertisement metadata of each broadcasted advertisement broadcasted in the predetermined period, and classifies each broadcasted advertisement broadcasted in the predetermined period according to the product category indicated by the advertisement metadata. Divide. After that, the classification unit 36 classifies each of the plurality of broadcast advertisements belonging to each product category into one of the plurality of groups based on the index calculated by the indexing unit 35 for each product category. Each group has at least one broadcast advertisement. In other words, each broadcast advertisement is divided into one or more groups.

Here, the method adopted by the classification unit 36 when performing classification (grouping) will be described. A known classification method can be used, and the K-means method known as a simple clustering method is particularly preferable. .. This K-means method is a clustering method by unsupervised learning, and according to this method, an index calculated by the indexing unit 35 for a broadcasted program / broadcasted advertisement is displayed in a multidimensional coordinate space as shown in FIG. Plots are plotted and divided into k clusters (k is a natural number of 2 or more) according to the proximity of the plots. The k clusters correspond to the "plurality of groups" in the present invention.
By the way, FIG. 5 is an image diagram of grouping, and specifically, is an explanatory diagram of clustering by the K-means method. For convenience of explanation, FIG. 5 shows the multidimensional coordinate space on which the plot is drawn as a two-dimensional coordinate space, but the present invention is not limited to this, and is a higher-dimensional coordinate space. May be good.

(Input information acquisition section)
The input information acquisition unit 37 acquires the input information input as the explanatory information for the broadcast schedule advertisement and the broadcast schedule program to be broadcast in the future. Specifically, the input information for the scheduled broadcast advertisement is the product category, performers, advertisement content (image), etc. of the scheduled broadcast advertisement, and is transmitted from, for example, the advertising company C that handles the scheduled broadcast advertisement, and is investigated. Entered on the company R side. Strictly speaking, an employee of the research company R or the like inputs the above input information through the input device 11e of the server computer 11. As a result, the input information acquisition unit 37 acquires the input information input as the explanatory information of the broadcast scheduled advertisement.

The acquisition route of the input information of the scheduled broadcast advertisement is not limited to the above contents. For example, the explanatory information about the scheduled broadcast advertisement is input on the advertising company C side, and the data indicating the input information is the network. It may be transmitted to the server computer 11 of the research company R via the route.

The input information about the scheduled broadcast program is the broadcast date and time, broadcast day, program title, program genre, performer, etc. of the scheduled broadcast program, and is distributed as program guide information from, for example, the TV station T that broadcasts the scheduled broadcast program. To. The input information acquisition unit 37 acquires the input information input as the explanatory information of each broadcast scheduled program by sequentially receiving the program guide information distributed from each television station T.

The acquisition route of the input information of the scheduled broadcast program is not limited to the above contents. For example, the explanatory information about the scheduled broadcast program is transmitted from each TV station T by e-mail or fax, and the research company R. An employee or the like may input the above information through the input device 11e of the server computer 11.

(Judgment unit)
The determination unit 38 determines (predicts) the group to which the broadcast scheduled advertisement and the broadcast scheduled program belong among the plurality of groups when the classification unit 36 groups (clusters) the broadcasted program and the broadcasted advertisement. Explaining the group determination (group prediction) by the determination unit 38, before the group determination is performed, the indexing unit 35 first uses the input information acquired by the input information acquisition unit 37 to broadcast the scheduled program / broadcast schedule. Determining advertising indicators. Specifically, by applying the above input information to the correspondence relationship (specifically, a mathematical model showing the correspondence relationship) specified by the specific unit 34, an index of the broadcast schedule program / broadcast schedule advertisement related to the input information. Is calculated.

After that, the determination unit 38 determines the group to which the broadcast scheduled program / broadcast schedule advertisement belongs based on the index calculated by the index unit 35. Specifically, the determination unit 38 plots the index calculated for the broadcast schedule program / broadcast schedule advertisement in the coordinate space shown in FIG. 5, and the plot exists in which cluster (group). By specifying whether to do so, the group to which the broadcast scheduled program / broadcast scheduled advertisement belongs is determined.

As described above, in the present embodiment, the group to which the broadcast scheduled program / broadcast scheduled advertisement belongs is determined by selecting one from a plurality of candidate groups, but the present invention is not limited to this. Absent. For example, for each of the above-mentioned plurality of groups, the probability that the broadcast scheduled program / broadcast schedule advertisement belongs is obtained, and if the above probability is higher than the reference value for two or more groups, all of those groups are selected. It may be treated as a group to which the broadcast scheduled program / broadcast scheduled advertisement belongs.

In the present embodiment, the indexing unit 35 calculates an index for each of the scheduled broadcast program and the scheduled broadcast advertisement, and the determination unit 38 calculates an index for each of the scheduled broadcast program and the scheduled broadcast advertisement, based on the index. Determine (predict) the group to which the scheduled broadcast program belongs and the group to which the scheduled broadcast advertisement belongs.

(Similarity calculation unit)
The similarity degree calculation unit 39 calculates the degree of similarity between the group to which the scheduled broadcast advertisement determined by the determination unit 38 belongs (advertisement group) and the group to which the scheduled broadcast program belongs (program group). Explaining the procedure for calculating the degree of similarity, first, the degree of similarity calculation unit 39 associates the program group with the explanatory information of the broadcasted program belonging to the program group.

Explaining the above procedure concretely with reference to FIG. 6, the similarity calculation unit 39 describes the performers, contents, etc. of each of the broadcasted programs belonging to a certain program group (group X in the case of FIG. 6). After specifying the explanation information of, all the specified explanation information is aggregated (merged) as the explanation information of a certain program group. Further, as shown in FIG. 6, each of the aggregated explanatory information is digitized (vectorized) by the digitization processing unit 33. By the way, FIG. 6 is an explanatory diagram of processing for explanatory information of programs / advertisements belonging to the group.

Next, the similarity degree calculation unit 39 associates the explanatory information of the broadcasted advertisement belonging to the advertisement group with the advertisement group by a procedure substantially the same as described above. Specifically, the similarity calculation unit 39 specifies the description of the performers, contents, etc. for each of the broadcasted advertisements belonging to a certain advertisement group by the procedure shown in FIG. 6, and provides all the specified explanatory information. , Aggregate (merge) as explanatory information of a certain ad group. In addition, each of the aggregated explanatory information is digitized (vectorized) by the digitization processing unit 33.

After that, the similarity degree calculation unit 39 calculates the degree of similarity between the program group and the advertisement group by using the explanatory information aggregated for the program group and the explanatory information aggregated for the advertisement group. More specifically, the similarity calculation unit 39 is based on the explanatory information of the program group quantified by the quantification processing unit 33 and the explanatory variables of the advertisement group also quantified by the quantification processing unit 33. Calculate the degree of similarity between groups. Here, the calculation method of the degree of similarity is not particularly limited, and a suitable calculation method can be selected and used from known calculation methods.

Further, the calculation of the degree of similarity is repeated a plurality of times for one advertisement group by changing the program group, and more specifically, the calculation is repeated as many times as the number of program groups.

As described above, the program group having the highest degree of similarity with the certain advertisement group can be specified by repeatedly calculating the program group for a certain advertisement group. Here, the program group having the highest degree of similarity with a certain advertising group is the program group that best matches the certain advertising group, and will be hereinafter referred to as "best match group".

The best match group is a program group in which the viewer's impression when viewing a scheduled broadcast program belonging to the same group is expected to be similar to the impression when viewing a broadcast scheduled advertisement belonging to a certain advertising group. That is, the advertising space (spot CM frame) set during the broadcasting time of the broadcast scheduled program belonging to the best match group becomes the "recommended advertising space" for the broadcast scheduled advertisement belonging to a certain advertising group.

<< About data processing flow >>
Next, as an operation example of the data processing device 10, the flow of data processing (hereinafter, data processing flow) executed by the data processing device 10 will be described.

In the data processing flow described below, the data processing method of the present invention is adopted. That is, the following description includes a description of the data processing method of the present invention, and also describes the data processing flow described below (specifically, the grouping process, the group determination process, and the degree of similarity described later). Each step of the calculation process) corresponds to a step constituting the data processing method of the present invention.

The data processing flow according to the present embodiment is roughly divided into three processes, and specifically, it is composed of a grouping process, a group determination process, and a similarity degree calculation process. The grouping process is a process of classifying each of the plurality of broadcast advertisements / broadcasted programs broadcasted in a predetermined period into one of the plurality of groups, and proceeds according to the flow shown in FIG. 7. FIG. 7 is a diagram showing a flow of grouping processing.

The group determination process is a process for determining each of the group to which the scheduled broadcast advertisement belongs (advertisement group) and the group to which the scheduled broadcast program belongs (program group), and proceeds according to the flow shown in FIG. FIG. 8 is a diagram showing a flow of group determination processing.

The similarity degree calculation process is a process of calculating the similarity degree between the advertisement group and the program group, and proceeds according to the flow shown in FIG. FIG. 9 is a diagram showing a flow of the similarity degree calculation process.
The processing procedure will be described in detail below.

(Grouping process)
When carrying out the grouping process, first, a questionnaire survey was conducted on the impressions of TV programs (broadcast programs) and TV commercials (broadcast advertisements) that were broadcast during the survey period, and Monitor M actually watched them during the survey period. Answer the impressions of the TV program / TV commercial that you did. After that, the research company R acquires data showing the results of a questionnaire regarding the impressions answered by each monitor M, that is, program evaluation data and advertisement program data for each monitor M and for each program / advertisement.

Then, at the start of the grouping process, the storage device 12 constituting the data processing device 10 stores the acquired program evaluation data and the advertisement program data for each monitor M and for each program / advertisement (S001).

In addition, the research company R acquires program metadata showing explanatory information of TV programs broadcast during the survey period from TV station T, and advertises advertising metadata showing explanatory information of TV commercials broadcast during the survey period. Obtained from company C. Then, the storage device 12 stores the acquired program metadata and advertisement metadata for each program / advertisement at the start of the grouping process (S002). Note that step S002 may be performed before step S001 described above, or may be performed in parallel with step S001.

Next, the server computer 11 constituting the data processing device 10 accesses the storage device 12 to read the program evaluation data for each program and each monitor M and the program metadata for each program, and uses these data. Conduct machine learning. As a result, the server computer 11 constructs a mathematical model showing the correspondence between the index derived from the evaluation indicated by the program evaluation data and the explanatory information, and specifies the correspondence (S003).

Further, in step S003, the server computer 11 reads the advertisement evaluation data for each advertisement and each monitor M and the advertisement metadata for each advertisement from the storage device 12, and performs machine learning using these data. As a result, the server computer 11 constructs a mathematical model showing the correspondence between the index derived from the evaluation indicated by the advertisement evaluation data and the explanatory information, and specifies the correspondence.

In the present embodiment, the server computer 11 digitizes (vectorizes) the explanatory information indicated by the program metadata / advertisement metadata read from the storage device 12, and uses the digitized explanatory information for the above machine learning. .. As a result, in step S003, the correspondence between the index and the digitized (vectorized) explanatory information is specified.

Next, the server computer 11 targets each of the plurality of TV programs / the plurality of TV commercials broadcast during the questionnaire survey period, and the mathematical model constructed in step S003 for the explanatory information indicated by the metadata of the target. Applies to. As a result, the server computer 11 determines the index of each target TV program (broadcasted program) / each TV commercial (broadcasted advertisement) (S004).

Next, the server computer 11 classifies each of the plurality of TV programs / the plurality of TV commercials broadcast during the questionnaire survey period into one of the plurality of groups based on the index determined in step S004 of the previous stage. (S005). Specifically, for example, each TV program / each TV commercial is classified (group) by dividing the index of each TV program / each TV CM determined in step S004 into k clusters by the K-means method. Divide).

In the present embodiment, the server computer 11 divides each TV program according to the program genre indicated by the program metadata of each TV program, and then describes the TV programs belonging to each program genre for each program genre. Classify (group) according to the procedure of. Further, similarly, the server computer 11 divides each TV commercial according to the product category indicated by the advertisement metadata of each TV commercial, and then divides each TV commercial into the TV commercials belonging to each product category as described above. Classify (group) according to the procedure. At least one TV program / TV commercial belongs to each group.

When all the steps up to the above are completed, the grouping process ends. The grouping process is repeatedly carried out at an appropriate timing. For example, when the above-mentioned questionnaire survey is newly carried out, it is preferable that the grouping process is carried out within a certain period after the completion of the questionnaire survey.

(Group judgment processing)
The group determination process is performed after the grouping process is performed. In the present embodiment, the group determination process is started by receiving an implementation request from the advertising company C or the like. For example, the advertising company C plans to place a new advertisement (hereinafter referred to as a new advertisement A), and the group is set in the planning. When the execution of the determination process is requested, it is started with this as an opportunity. In the following, the flow of the group determination process will be described by taking a case where the new advertisement A is a scheduled advertisement to be broadcast as a specific example.

At the time of the group determination process, first, the server computer 11 constituting the data processing device 10 acquires the input information regarding the explanatory information of the new advertisement A (S011). As described above, the explanatory information of the new advertisement A can be obtained from, for example, the advertising company C or the like.

Further, the server computer 11 acquires input information regarding explanatory information of a television program (scheduled broadcast program) scheduled to be broadcast in the future from each television station T or the like (S012). The input information acquired in this step S012 is, for example, explanatory information of a television program to be broadcast several months later, specifically, the broadcast date and time, the broadcast day, and the program of the television program. It is a title, performers, etc., and is distributed from each TV station T as program guide information.
Incidentally, step S012 may be carried out before step S011 described above, or may be carried out in parallel with step S011.

After that, the server computer 11 applies the acquired input information to the correspondence relationship (strictly speaking, a mathematical model showing the correspondence relationship) specified in step S013 of the grouping process. As a result, an index is calculated for each of the new advertisement A and the TV program scheduled to be broadcast in the future (S013).

Next, the server computer 11 determines the group to which the new advertisement A belongs (advertisement group) based on the index determined in step S013 in the previous stage (S014), and the group to which the TV program scheduled to be broadcast in the future belongs (program). Group) is determined (S015). Here, since there are usually a plurality of TV programs scheduled to be broadcast in the future, in step S015, a program group is determined for each of the plurality of TV programs scheduled to be broadcast in the future.

When the series of steps up to the above is completed, the group determination process ends. Then, in the present embodiment, when the group determination process is completed, the similarity degree calculation process is continuously executed.

(Similarity calculation process)
As described above, the similarity calculation process is continuously performed with, for example, the group determination process. In the following, as in the description of the group determination process, the flow of the similarity degree calculation process will be described by taking a case where the new advertisement A is a scheduled broadcast advertisement as a specific example. In order to make the explanation easier to understand, in the following, only three TV programs (hereinafter, programs P1, P2, P3) are scheduled to be broadcast in the future. Further, in the following, the program group to which each TV program belongs will be referred to as G1, G2, G3, and the advertisement group to which the new advertisement A belongs will be referred to as Ga.

In the similarity calculation step, the server computer 11 calculates the similarity between the advertisement group Ga to which the new advertisement A belongs and the program groups G1, G2, and G3 to which the TV program scheduled to be broadcast in the future belongs.

Specifically, in the similarity degree calculation step, the server computer 11 first aggregates the explanatory information of all the broadcasted advertisements belonging to the advertisement group Ga (S021). Similarly, the server computer 11 aggregates the explanatory information of all the broadcasted programs belonging to each program group for each program group G1, G2, and G3 (S022). Then, the explanatory information aggregated in each group is digitized (vectorized) as described above.

Next, the server computer 11 aggregates and quantifies (vectorizes) the explanatory information of the broadcasted advertisement for the advertisement group Ga and aggregates and quantifies (vectorizes) each program group G1, G2, and G3. Based on the explanatory information of the broadcasted program, the degree of similarity between the program groups G1, G2, and G3 and the advertising group Ga is calculated (S023). This step S023 is repeatedly carried out by changing the program group.

Then, after calculating the degree of similarity with the advertisement group Ga for all the program groups G1, G2, and G3 (Yes in S024), the server computer 11 has the program group with the highest degree of similarity, that is, the best match group. Is specified (S025).

Further, the server computer 11 identifies a TV program scheduled to be broadcast in the future belonging to the best match group, and the advertising company C plans to place a new advertisement A in the ad space set during the broadcast time of the TV program. (For example, mail transmission process or print process on paper or the like) is executed (S026). Then, the advertising space notified in this step S026 becomes a "recommended advertising space" for the new advertisement A.

As an example of the notification method of the recommended advertising space, for example, there is a method of transmitting data for displaying the notification screen shown in FIG. 10 to the advertising company C or the like. As shown in the figure, on the notification screen, along with the explanatory information (specifically, product category, performers, contents, etc.) of the new advertisement A input in this data processing flow, the broadcast of the recommended advertisement space The time zone and the degree of matching with the recommended ad space (specifically, the number of performers matching, the degree of performer similarity, and the degree of content similarity) are displayed. Through this notification screen, it is possible to confirm the broadcast frame from which a useful advertising effect can be obtained for the new advertisement A.

In the present embodiment, the server computer 11 is supposed to carry out steps S025 and S026, but these steps S025 and S026 may be carried out manually.

When the series of steps up to the above are completed, the similarity calculation process ends. Then, when a new implementation request is received from the advertising company C or the like, the above-mentioned group determination process and the subsequent similarity degree calculation process are performed.

<< About the effectiveness of this embodiment >>
In the present embodiment, as described above, an index related to the TV program / TV commercial is derived from the evaluation indicated by the program evaluation data / advertisement evaluation data. Further, by specifying the correspondence between the index and the explanatory information and applying the explanatory information of a certain TV program / a certain TV commercial to the above correspondence, the index of a certain TV program / a certain TV CM is calculated. Further, based on the index, each of the plurality of television programs / plurality of televisions is classified into one of the plurality of groups.

Then, according to the above configuration, it is possible to appropriately classify the programs and advertisements that have been broadcast in the past based on the evaluation of the viewer. Specifically, as explained in the section "Problems to be solved by the invention", if programs and advertisements can be classified based on the evaluation of viewers, the utility value of the classification results will increase. On the other hand, even for the same program or the same advertisement, the evaluation may change if the viewer changes. Therefore, when classifying programs and advertisements based on the viewer's evaluation, the diversity of such evaluation is used. Need to consider.

On the other hand, in the present embodiment, as described above, the programs and advertisements are classified (grouped) by applying the correspondence between the index and the explanatory information. Here, the correspondence relationship is specified by subjecting the program evaluation data / advertisement evaluation data indicating the evaluation of each viewer (strictly, each monitor M) to machine learning, in other words, each viewer. It reflects the evaluation (impression) of. In the present embodiment, since the program and the advertisement are classified by applying such a correspondence, the program and the advertisement are appropriately based on the evaluation of the viewer while considering that the evaluation of the viewer can change in various ways. It becomes possible to classify.

Further, in the present embodiment, since the group to which the program and the advertisement scheduled to be broadcast in the future belong can be determined (predicted), it is possible to grasp in advance which group the program and the advertisement scheduled to be broadcast in the future belong to. It is possible.

Further, in the present embodiment, the degree of similarity between groups can be calculated. Specifically, a group to which an advertisement scheduled to be broadcast in the future belongs (advertisement group) and a group to which a program scheduled to be broadcast in the future belongs (program group). It is possible to calculate the degree of similarity between and. This makes it possible to identify the program group most similar to the advertisement group, that is, the best match group. Then, if a certain advertisement (for example, the above-mentioned new advertisement A) is broadcast using the advertisement space (spot CM frame) set during the broadcast time of a future TV program belonging to the best match group, the above advertisement Can be effectively hit, and in an easy-to-understand manner, it can be hit toward a viewer who tends to react to the advertisement.

In the present embodiment, the degree of similarity between the advertisement group and the program group is calculated to specify the best match group, but the present invention is not limited to this. That is, the degree of similarity between the advertising groups may be calculated to specify the best match group, or the degree of similarity between the program groups may be calculated to specify the best match group.

<< Other Embodiments >>
Although the data processing apparatus and the data processing method of the present invention have been described above with specific examples, the above-described embodiment is merely an example, and other embodiments can be considered. For example, in the above embodiment, a plurality of TV programs (broadcasted programs) broadcasted in the past are classified, and a plurality of TV commercials (broadcasted advertisements) broadcasted in the past are classified. It is not limited to this. For example, it may be in the form of classifying only one of the broadcasted program and the broadcasted advertisement.

Further, the data processing device according to the above embodiment (that is, the data processing device 10 of the present embodiment) has a function of classifying broadcasted programs and broadcasted advertisements, as well as a group to which future broadcasted programs and advertisements belong. It has a function to judge and a function to calculate the degree of similarity between groups. However, the present invention is not limited to this, and a data processing device that has only a function of classifying broadcasted programs and broadcasted advertisements but does not have the remaining two functions may be used.

Further, in the above embodiment, the degree of similarity between the group to which the program to be broadcast in the future belongs (program group) and the group to which the advertisement to be broadcast in the future belongs (advertisement group) is calculated. Although different from such a form, as one reference example, there may be a case where the degree of similarity between a program to be broadcast in the future and an advertisement to be broadcast in the future is calculated. In such a case, the degree of similarity between the two will be calculated based on the explanatory information about the program to be broadcast in the future and the explanatory information about the advertisement to be broadcast in the future.
On the other hand, when calculating the degree of similarity between groups, the explanatory information of all the programs or advertisements belonging to each group is aggregated (merged) and used, so that the number of explanatory information which is an explanatory factor becomes larger. Therefore, it is more advantageous to calculate the degree of similarity between the groups because a more reasonable calculation result can be obtained.

Further, in the above embodiment, the program evaluation data and the advertisement evaluation data are data showing the results of a questionnaire regarding the impressions of the viewers. However, the evaluation data is not particularly limited as long as it is data that can identify the evaluation (impression) of the viewer, and for example, the brain wave, pulse, body temperature, etc. when viewing the program / advertisement of the viewer. It may be data showing biometric information.

10 Data processing device 11 Server computer 11a CPU
11b Memory 11c Communication interface 11d Hard disk drive 11e Input device 11f Output device 12 Storage device 21 Data distribution server 22 Monitor terminal 23 Data distribution server 31 Evaluation data storage unit 32 Metadata storage unit 33 Digitization processing unit 34 Specific unit 35 Indexing unit 36 Classification section 37 Input information acquisition section 38 Judgment section 39 Similarity calculation section C Advertising company M monitor (contact person)
R Research Company T TV Station

Claims (13)

An evaluation data storage unit that stores evaluation data indicating the evaluation of the delivered content by a contact person who has come into contact with the delivered content for each of the delivered contents and each of the contacts.
A metadata storage unit that stores metadata indicating explanatory information for explaining the delivered content for each delivered content, and
A specific unit that specifies the correspondence between the index derived from the evaluation and the explanatory information based on the evaluation data for each of the delivered contents and each of the contacts and the metadata for each of the delivered contents. When,
An indexing unit that calculates the index of the target by applying the explanatory information indicated by the metadata of the target to each of the plurality of delivered contents.
A data processing apparatus comprising: a classification unit that classifies each of a plurality of the delivered contents into any of a plurality of groups based on the index calculated by the indexing unit. The specific unit performs machine learning using the evaluation data for each of the delivered contents and the contact, and the metadata for each of the delivered contents, and obtains a mathematical model showing the correspondence relationship. The data processing device according to claim 1, wherein the correspondence relationship is specified by constructing the data processing device. The data processing device according to claim 1 or 2, wherein the delivered content is at least one of a broadcasted program and a broadcasted advertisement that has been broadcast in the past. Classification categories are preset for the delivered contents, and
The classification unit classifies each of the plurality of delivered contents belonging to the classification category into one of the plurality of groups based on the index calculated by the indexing unit according to the classification category. The data processing apparatus according to any one of 3. The data processing device according to any one of claims 1 to 4, wherein the evaluation data is data showing a questionnaire result regarding the impression of the contact person with respect to the delivered content. The data processing apparatus according to claim 5, wherein a part of the evaluation data showing the result of the questionnaire is used as verification data for the validity of the correspondence. An input information acquisition unit that acquires input information input as the above-mentioned explanatory information of the distribution schedule contents to be distributed in the future,
It has a determination unit for determining the group to which the distribution schedule content belongs among the plurality of groups.
By applying the input information to the correspondence, the indexing unit calculates the index of the distribution schedule content.
The data processing device according to any one of claims 1 to 6, wherein the determination unit determines a group to which the distribution schedule content belongs based on the index of the distribution schedule content calculated. The input information acquisition unit acquires the input information for each of the first distribution schedule content and the second distribution schedule content that are different from each other.
The indexing unit calculates the index by applying the input information to the corresponding relationship for each of the first delivery scheduled content and the second delivery scheduled content.
In the determination unit, the group to which the first delivery scheduled content belongs and the second delivery scheduled content are based on the index calculated for each of the first delivery scheduled content and the second delivery scheduled content. Determine the group to which it belongs
The data processing device according to claim 7, further comprising a similarity degree calculation unit for calculating the degree of similarity between the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs. The first delivery schedule is an advertisement to be broadcast in the future.
The data processing device according to claim 8, wherein the second distribution schedule content is a program to be broadcast in the future. The data processing apparatus according to claim 8 or 9, which has a quantification processing unit that performs a process of quantifying the explanatory information indicated by the metadata. The similarity calculation unit includes the explanatory information of the first distribution scheduled content quantified by the quantification processing unit and the explanatory information of the second distribution scheduled content quantified by the quantification processing unit. The data processing device according to claim 10, wherein the degree of similarity between the group to which the first scheduled delivery content belongs and the group to which the second scheduled delivery content belongs is calculated based on the above. The data processing device according to claim 10 or 11, wherein the specific unit specifies the correspondence between the index and the digitized explanatory information. The storage device stores evaluation data indicating the evaluation of the delivered content by the contact person who has come into contact with the delivered content for each of the delivered contents and for each of the contacts.
The storage device stores metadata indicating explanatory information for explaining the delivered content for each delivered content.
The computer identifies the correspondence between the index derived from the evaluation and the explanatory information based on the evaluation data for each of the delivered contents and each of the contacts and the metadata for each of the delivered contents. And
A computer targets each of the plurality of delivered contents, and applies the explanatory information indicated by the metadata of the target to the correspondence to determine the index of the target.
A data processing method, wherein a computer classifies each of the plurality of delivered contents into one of a plurality of groups based on the index.