JP5417270B2 - Broadcast plan generation apparatus, broadcast plan generation method, and broadcast plan generation program - Google Patents

Description

  The present invention relates to a technique for generating a broadcast plan for content broadcast on digital signage.

  Digital signage is also called electronic signage and electronic posters, etc. "It is possible to use electronic display devices such as displays connected to a communication network in various places such as outdoors, storefronts, public spaces, and transportation facilities. It is defined as “the system to send”. It is characterized in that content can be distributed to various places via a communication network, and the contents can be switched and displayed in a timely manner according to time zones and places. For example, it is possible to switch the contents for each time zone in accordance with the characteristics of a place such as “there are many office workers on the way home at night and many housewives in the daytime”.

  Such digital signage can be said to be an information transmission method that cannot be realized by conventional media such as signs, posters, and televisions.

[Prior art 1]
Non-Patent Document 1 describes a method for editing a broadcast plan of content used in a standard digital signage system.

  As shown in FIG. 11, the standard digital signage system displays a distribution server 201 that manages and distributes contents and broadcast plans, a player 202 that reproduces contents based on the broadcast plans, and the reproduced contents. The display 203 includes an operation terminal 204 that inputs setting values and contents for generating a broadcast plan to the distribution server 201. A plurality of players 202 are connected to one distribution server 201, and one or more displays are connected to one player.

  A basic operation flow of this standard digital signage system will be described with reference to FIGS.

  First, based on the content registration operation by the operator, the operation terminal 204 registers content such as video / still image files and telop character strings in the distribution server 201 (S301).

  Next, based on the playlist generation operation by the operator, a playlist in which the content playback order or display order is set is generated by the operation terminal 204 and the distribution server 201 (S302). For example, playlists A and B as shown in FIG. 13A are generated. Play list A defines that three contents A1, A2, and A3 are played back in a cycle of 60 seconds in the order of A1, A2, and A3. On the other hand, it is defined in the playlist B that the four contents B1, B2, B3, and B4 are played back in a cycle of 90 seconds in the order of B1, B2, B1, B3, B1, and B4. Note that a general digital signage system provides a graphical user interface for generating a playlist.

  Next, on the basis of a schedule generation / registration operation performed by the operator, a schedule in which the playback date / time (playback date / playback start time / playback end time) of the playlist is set is generated by the operation terminal 204 and registered in the distribution server 201. (S303). For example, a schedule as shown in FIG. 13B is generated. In this schedule, playlist A is played from 6 o'clock to 12 o'clock, playlist B is played from 12 o'clock to 16 o'clock, and playlist C is played from 16 o'clock to 23 o'clock. It is defined to play. Specifically, for example, 6:00 is set for the playback start time of playlist A, and 12:00 is set for the playback end time.

  The set of playlists and schedules corresponds to the broadcast plan of the content, and as shown in FIG. 13 (c), the broadcast that lists the playback start time and playback end time of each content from the playlist and schedule A plan is generated. In this example, the playlist A that makes a round in 60 seconds is repeatedly played 360 times in 6 hours (= 360 minutes) from 6:00 to 12:00. Similarly, the playlist B that makes a round in 90 seconds is repeatedly played 160 times in 4 hours from 12:00 to 16:00 (= 240 minutes). In particular, for the content B1, since the number of times of reproduction is set three times during one cycle, the content B1 is reproduced 480 times (= 3 times × 160 times) in the four hours.

  Next, when the time designated by the operator is reached, the distribution server 201 distributes the registered content and the broadcast plan of the content to each player 202 (S304). Broadcasting plans are accumulated.

  Finally, when the time set in the content broadcast plan is reached, the content corresponding to that time is automatically played by each player 202 and displayed on the display 203 (S305).

  This type of content operation method has almost no designation of the time at which the content is broadcast from the content broadcaster (the advertiser if the content is an advertisement), and manipulates the number of broadcasts and the time length of the content (hereinafter referred to as the scale). It is very efficient if one can decide arbitrarily.

  However, there is a demand from broadcasters and advertisers not only to select the content playback / display location but also to set the broadcast time, the number of broadcasts, and the scale in detail in order to enhance the broadcast effect and advertisement effect.

  For example, if the advertiser designates the number of airings of content A1, A2, A3 as 10 times, 20 times, and 50 times according to the advertising budget, simple repeat playback of playlist A cannot be used. A new playlist incorporating A1, A2, and A3 at a ratio of 1: 2: 5 must be generated. In addition, when the scale is different for each content, it becomes difficult to assign the content to the empty time frame without a gap.

  That is, according to the related art 1, there is a problem that the operation cost increases because the work of generating the content broadcast plan becomes more complicated as the degree of freedom of the broadcast application conditions and the advertisement application conditions increases.

[Prior art 2]
Non-Patent Document 2 describes a method for expressing a digital signage advertisement application (order). Specifically, when applying for an advertisement to digital signage, an advertiser or its agent designates a value for an advertisement application item as shown in FIG. The digital signage system of Non-Patent Document 2 accepts a plurality of advertisement applications from an advertiser or its agency, and automatically assigns a broadcast time zone to advertisements in the order of application.

  With reference to FIGS. 15 and 16, the broadcast plan generation method shown in Non-Patent Document 2 will be described. FIG. 15 is a diagram showing four types of advertisement applications (orders). FIG. 16 is a diagram showing broadcast time slots assigned based on the four types of advertisement applications. Note that the horizontal axis of FIG. 16 represents time, the vertical axis represents four types of orders, and the broadcast time zones assigned based on the orders are indicated by dot hatches. The numbers in the dot hatches indicate the order of application (priority) of the order.

  Among the four types of orders, the time designation order T displays an advertisement having a length of 3 units of L501 (for example, 45 seconds when 15 seconds is a unit time) from the display start time of T17 to the display end time of T20. To ask you to do. Since the display start time and the display end time are fixedly specified in this time designation order T, a time zone from T17 to T19 is assigned (see reference numeral 700 in FIG. 16).

  On the other hand, the bulk order B is a request for repeatedly displaying an advertisement having a time scale of 3 units called XP from the display start time of T1 to the end of the display time of T5000. Since the cycle time length (reproduction allowable time interval) is 10 unit hours, the advertisement of 3 unit times is broadcast once each in the time zone indicated by reference numerals 701 to 703 in FIG.

  On the other hand, the repeat order R is a request for repeatedly displaying an advertisement having a scale of 2 unit time called “Feat” from the display start time of T1 to the display end time of T23. Since this repetitive order R has a cycle time length of 2 unit hours and a cycle interval of 8 unit hours, time periods of T01 and T02, T11 and T12, and T21 and T22 must be assigned. Note that T03 to T05, T13 to T15, and T23 to T25, which are time zones in which the repetitive order R is not allocated, are allocated to the bulk order B described above.

  On the other hand, the standard schedule order S is a request for repeatedly displaying an advertisement having a scale of 2 unit time called McD from the display start time of T1 to the display end time of T30. Since the cycle time length is 10 unit hours, it is desired to broadcast each advertisement of 2 unit times once in the 10 unit time frame indicated by reference numerals 704 to 706 in FIG. The time zones of T06 and T07 and T26 and T27 can be assigned to the reference numerals 704 and 706, respectively. However, since there is no free time slot for two unit hours in the time frame of reference numeral 705, the broadcast plan according to the advertisement application Cannot be generated. In such a case, the advertiser or its agent is counter-proposed whether there is no problem with one less broadcast time.

  As described above, according to the related art 2, since the broadcast time zone is allocated in the order of the application, there is a problem that it is impossible to always generate an optimal broadcast plan for the advertiser or its agency. That is, as the application application is delayed, it is difficult to assign the broadcast time period of the advertisement to an intended time period, and the application application cannot be accepted and must be refused. Another problem is that a gap (a time zone in which no advertisement is allocated) is created in the generated broadcast plan, as in the time zones T16 and T20 in FIG.

[Prior art 3]
Patent Document 1 lists playback time zones and contents, and by specifying a content playback priority mode such as “in order of decreasing playback count” or “in order of decreasing playback count”, the playback count is averaged, A content repeat playback device that eliminates unfairness is disclosed. This is to automatically generate a broadcast plan based on the popularity of the content.

  However, the content played back with digital signage has a variable content length and is not fixed like 15 seconds or 30 seconds. Therefore, when the number of playbacks is averaged, Advertisers are more advantageous because they occupy longer airtime, and there is a problem that fairness between broadcasters or advertisers cannot be maintained. In addition, when the content is mechanically repeatedly reproduced, there is a possibility that content in the same industry may flow continuously.

JP 2005-165878 A

"Digital Signage Standard System Guidelines", Digital Signage Consortium System Subcommittee, Version 1.0, November 1, 2008 John V. Harrison, 1 other, "A virtual marketplace for advertising narrowcast over digital signage networks", Electronic Commerce Research and Applications 3, 2004, p.163-175 Yaginuma, 1 other, “Meta-strategy for combinatorial optimization problem”, IEICE D-I, J83-D-I, 1, 2000, p.3-25

  As described above, according to the prior art, there is a problem that the generation work of the content broadcast plan becomes more complicated and time-consuming as the broadcast application conditions and the advertisement application conditions can be specified in detail.

  The present invention has been made in view of the above, and an object of the present invention is to generate a broadcast plan that is optimized based on a preset criterion while satisfying predetermined application conditions.

The broadcast plan generation device according to claim 1, wherein each time length and each designated broadcast count of a plurality of contents expected to be broadcast on digital signage, and a plurality of broadcast available time frames capable of broadcasting the contents The storage means for storing, the time length of the content acquired from the storage means, the designated number of broadcasts, and the broadcastable time frame are used to indicate which broadcastable time frame the content is allocated and broadcast in which order Generation means for generating an arbitrary broadcast plan, a first condition that all contents for each designated number of broadcasts are allocated to the broadcast plan, and a second condition that all contents fit within all broadcastable time frames while satisfying, possess and optimization means effect index for evaluating the goodness of predefined broadcast program to optimize the arbitrary broadcast program so as to maximize the said effective ^{_{Indicators, τΣ k = 1 n {N}} k d k + δ (1 + Σ i = 1 Nk-1 (1-e - (Dki-dk) / δ))} ( where, tau is passed through the vicinity of the digital signage units average passers number of time Atari, N _k is designated broadcast frequency of the k-th content, d _k is the time length of the k-th content, [delta] is the average residence time per unit time of passersby remained in the vicinity of the digital signage, e is a function of the base of the exponential function (the number of _Napiers ), and D _ki is a function of a broadcast interval from when the k-th content is broadcast for the i-th time until the next broadcast starts .

  According to the present invention, the storage means stores each time length and each designated number of broadcasts of a plurality of contents expected to be broadcast on digital signage and a plurality of broadcastable time frames in which the contents can be broadcast. Using the time length of the content acquired from the storage means, the specified number of broadcasts, and the broadcastable time frame, an arbitrary broadcast plan that shows in what order the content is allocated and broadcasted in which broadcastable time frame Generated and defined in advance while satisfying the first condition that all contents for each designated number of broadcasts are allocated to the broadcast plan and the second condition that all contents fit within the broadcastable time frame In order to optimize any broadcast plan so that the effect index for evaluating the goodness of the broadcast plan is maximized, the release is optimized according to preset criteria while satisfying the predetermined application conditions. It is possible to produce a plan.

  The broadcast plan generation device according to claim 2 is the broadcast plan generation device according to claim 1, wherein the optimization unit does not satisfy the number of contents not satisfying the first condition and / or the second condition. Using the difference between the number of broadcastable time frames and the effect index as an evaluation function, searching for a broadcast plan that minimizes the evaluation function while searching for a broadcast plan close to the broadcast plan starting from the arbitrary broadcast plan It is characterized by that.

  According to the present invention, the difference between the number of contents not satisfying the first condition and / or the number of broadcastable time frames not satisfying the second condition and the effect index is used as an evaluation function, and the broadcast plan is started from an arbitrary broadcast plan. While searching for a broadcast plan that is close to, a search is made for a broadcast plan that minimizes its evaluation function, so that even if all the constraint conditions are not satisfied, a broadcast plan that is close to optimal can be generated.

  The broadcast plan generation device according to claim 3 is the broadcast plan generation device according to claim 2, wherein the neighboring broadcast plan is generated by changing a part of elements constituting the broadcast plan. It is characterized by that.

5. The broadcast plan generation method according to claim 4 , wherein the broadcast plan generation method is performed by a broadcast plan generation device that generates a broadcast plan of content to be broadcast by digital signage. The broadcast plan generation device includes storage means, and the digital signage each time length and each designated broadcast times of the plurality of content broadcast is desired in a first step of storing a plurality of broadcast possible time frame that can broadcast the content in the storage means, said storage Using the time length of the content acquired from the means, the specified number of broadcasts, and the broadcastable time frame, an arbitrary broadcast plan that shows in what order the content is allocated to which broadcastable time frame and broadcasts is generated. 2 step, the first condition that all contents for each designated broadcast count are allocated to the broadcast plan, and all the broadcastable time frames A third step of optimizing the arbitrary broadcast plan so that the effect index for evaluating the goodness of the predefined broadcast plan is maximized while satisfying the second condition that the content is contained. Features.

  According to the present invention, the storage means stores each time length and each designated number of broadcasts of a plurality of contents expected to be broadcast on digital signage and a plurality of broadcastable time frames in which the contents can be broadcast. Using the time length of the content acquired from the storage means, the specified number of broadcasts, and the broadcastable time frame, an arbitrary broadcast plan that shows in what order the content is allocated and broadcasted in which broadcastable time frame Generated and defined in advance while satisfying the first condition that all contents for each designated number of broadcasts are allocated to the broadcast plan and the second condition that all contents fit within the broadcastable time frame In order to optimize any broadcast plan so that the effect index for evaluating the goodness of the broadcast plan is maximized, the release is optimized according to preset criteria while satisfying the predetermined application conditions. It is possible to produce a plan.

A broadcast plan generation program according to a fifth aspect causes a computer to function as the broadcast plan generation apparatus according to any one of the first to third aspects .

  According to the present invention, the storage means stores each time length and each designated number of broadcasts of a plurality of contents expected to be broadcast on digital signage and a plurality of broadcastable time frames in which the contents can be broadcast. Using the time length of the content acquired from the storage means, the specified number of broadcasts, and the broadcastable time frame, an arbitrary broadcast plan that shows in what order the content is allocated and broadcasted in which broadcastable time frame Generated and defined in advance while satisfying the first condition that all contents for each designated number of broadcasts are allocated to the broadcast plan and the second condition that all contents fit within the broadcastable time frame In order to optimize any broadcast plan so that the effect index for evaluating the goodness of the broadcast plan is maximized, the release is optimized according to preset criteria while satisfying the prescribed application conditions. It is possible to produce a plan.

  According to the present invention, it is possible to generate a broadcast plan that is optimized based on preset criteria while satisfying predetermined application conditions.

It is a figure which shows the functional block structure of a broadcast plan production | generation apparatus. It is a figure which shows the input information and output information of a broadcast plan production | generation apparatus. It is a figure which shows the specific example of input information and output information. It is a figure which shows the relationship between a broadcast plan and a broadcast space | interval. It is a figure which shows the relationship between a broadcast interval and an evaluation function value. It is a figure which shows the whole operation | movement flow of a broadcast plan production | generation apparatus. It is a figure which shows the production | generation flow of a 1st initial solution. It is a figure which shows the production | generation flow of a 2nd initial solution. It is a figure which shows the production | generation flow of the best solution. It is a figure which shows the production | generation procedure of a neighborhood solution typically. It is a figure which shows the whole system structure of a standard digital signage system. It is a figure which shows the basic operational flow of a standard digital signage system. It is a figure explaining the situation until the broadcast plan of a content is drawn up with a basic operation flow. It is a figure which shows an example of an advertisement application item. It is a figure which shows an example of an advertisement application. It is a figure which shows the airing time slot | zone allocated based on the advertisement application.

  Hereinafter, an embodiment for carrying out the present invention will be described with reference to the drawings. However, the present invention can be implemented in many different modes and should not be construed as being limited to the description of the present embodiment.

[About the configuration of the broadcast plan generator]
Initially, the structure of the broadcast plan production | generation apparatus which concerns on this Embodiment is demonstrated. This broadcast plan generation device generates a broadcast plan for content broadcast on digital signage. Hereinafter, an advertisement will be described as a representative example of content. Note that the advertisement is merely an example, and the type of the advertisement is not limited as long as the content includes video and / or audio.

  FIG. 1 is a diagram illustrating a functional block configuration of a broadcast plan generation apparatus according to an embodiment. This broadcast plan generation apparatus includes an advertisement application condition storage unit 101, a frame information storage unit 102, an advertisement condition input unit 103, a neighborhood search method execution unit 104, and a broadcast plan output unit 105.

  The advertisement application condition storage unit 101 has a function of storing application conditions for advertisements that are expected to be broadcast on digital signage. Each designated number of broadcasts of each advertisement given as an input and the type of each advertisement (hereinafter referred to as genre) are stored in association with an advertisement identifier or the like.

  The frame information storage unit 102 has a function of storing information of a plurality of broadcastable time frames in which an advertisement can be broadcast with the digital signage, and specifically stores a broadcastable time zone including date information. .

  The advertisement condition input unit 103 has a function of inputting the advertisement application condition acquired from the advertisement application condition storage unit 101 and the frame information acquired from the frame information storage unit 102 to the subsequent neighborhood search method execution unit 104. .

  As shown in FIG. 1, the neighborhood search method execution unit 104 includes an initial solution generation unit 104a, an effect index calculation unit 104b, a broadcast plan update unit 104c, and a broadcast plan storage unit 104d.

  The initial solution generation unit 104a uses the advertisement application conditions and the frame information input by the advertisement condition input unit 103, and an arbitrary broadcast plan that indicates in which order the content is allocated and broadcast in which broadcastable time frame (Initial solution) is generated.

  The effect index calculation unit 104b has a function of calculating an effect index value of the generated broadcast plan (initial solution or intermediate solution) using an effect index that evaluates the goodness of a predefined broadcast plan. Note that the intermediate solution refers to a broadcast plan at an intermediate stage until the optimal solution updated by the subsequent broadcast plan update unit 104c.

  The broadcast plan update unit 104c has a function of updating the broadcast plan (initial solution or intermediate solution) so as to maximize the effect index while satisfying the application condition (constraint condition). The application condition (restriction condition) is that the first restriction condition that all advertisements for each designated number of broadcasts are allocated to the broadcast plan, and that all advertisements fit within the broadcastable time frame. The second constraint condition and the third constraint condition that advertisements of the same genre are not adjacent to each other are referred to.

  That is, the neighborhood search method execution unit 104 repeatedly executes the process by the effect index calculation unit 104b and the process by the broadcast plan update unit 104c until the effect index is maximized, so that the optimized broadcast plan (approximate (Optimum solution) is generated and stored in the broadcast plan storage unit 104d.

  The broadcast plan output unit 105 has a function of acquiring a broadcast plan from the broadcast plan storage unit 104d and outputting it to a display device or another device (not shown).

[Input information and output information of broadcast plan generator]
Next, input information input to the broadcast plan generation device and output information output from the broadcast plan generation device will be described.

  First, input information will be described. The input information includes an advertisement that has received a broadcast application, advertisement application conditions that are temporarily stored in the advertisement application condition storage unit 101, and frame information that is temporarily stored in the frame information storage unit 102. Since the broadcast plan generation device generates a broadcast plan for an advertisement, the advertisement itself is unnecessary when generating the broadcast plan, but is described here as an input for convenience of explanation.

Line conditions, the ad has received the broadcast of the application _{A k (k = 1,2, ...} , n) specified for, and specify the number of times aired _{N k} of advertising _{A k,} length (time length) of advertising _{A k} d and _k, have the item of the genre genre (k) of the advertising _{a k.} The genre is a type of advertisement, and examples thereof include beverages and clothing.

2 (a) it is, n ad type _A 1, ..., and the line conditions of the _{A n} represents schematically. For example, the advertisement A ₁ has a scale of d ₁ and the designated number of broadcasts N ₁ is four. Here, the number of broadcasts is arbitrarily designated at the time of application, but as is generally done, the designated number of broadcasts _Nk may be set in proportion to the advertising budget. The scale d _k is specified by a multiple of unit time (for example, 15 seconds).

The frame information is a broadcastable time frame Z _i (i = 1, 2,..., N _z ) in which an advertisement can be broadcast. Since digital signage displays are generally shared for multiple purposes, it is necessary to allocate advertisements to free time except for the time when content other than advertisements (for example, facility information and news) is aired. There is. Broadcastable time frame Z _i is a time zone having broadcast start time b _i and frame time length s _i as attributes, and there is a gap between different broadcastable time frames as shown in FIG. Are specified so as not to overlap each other.

Next, output information will be described. The output information is a broadcast plan optimized by the neighborhood search method execution unit 104. As shown in FIG. 2 (c), this broadcast plan is a list of advertisement numbers q _i = q _ij (q _ij ∈ [1, 2,..., N]) that are broadcast during the broadcast time frame Z _i . , J = 1, 2,..., N _i ).

For example, as shown in FIG. 3A, when advertisements A ₁ , A ₂ and A ₃ and designated broadcast times N ₁ = 4, N ₂ = 2 and N ₃ = 1 are given as advertisement application conditions. Although there are a plurality of executable broadcast plans, for example, a broadcast plan 1 shown in FIG. 3B and a broadcast plan 2 shown in FIG. 3C are generated. Broadcast plan 1 is expressed as q ₁ = (1,2,1); q ₂ = (1,2); q ₃ = (3, 1).

Note that if all advertisements A _k fit within the available broadcast time frame Z _i without any gap, ΣN _k d _k = Σs _i holds. However, even if this equation is satisfied, there is not always a broadcast plan that fits in the broadcastable time frame Z _i .

[About optimization of the broadcast plan]
Broadcast plan optimization is the optimization of any broadcast plan generated using the application information and frame information that is input information so that the effect index is maximized while satisfying the application conditions (constraints). It is to be. That is, an optimized broadcast plan is generated by solving a problem of approximately maximizing a pre-designated effect index among combinations of broadcast plans satisfying the advertisement application condition (constraint condition). Here, an effect index used at the time of optimization will be described.

Advertisement _{A k (k = 1,2, ...} , n) and, and the specified number of times aired _{N k} of advertising _{A k,} and scale _{d k} of advertising _{A k,} genre genre of advertising _{A k} and (k), the time length s _i broadcast possible time frame _{Z i (i = 1,2, ...} , N z) as input and information, broadcast program _{q i (i = 1,2, ...} , N z) as the output information, wherein (1 The objective function of the effect index R (q _i ) shown in FIG.

ここで、τはデジタルサイネージの近傍を通過した単位時間辺りの平均通行人数、δはデジタルサイネージの近傍で留まった通行人の単位時間辺りの平均滞留時間、ｅは指数関数の底（ネイピア数）、Ｄ_ｋｉはｋ番目の広告がｉ回目に放映されてから次に放映開始されるまでの放映間隔である。なお、放映間隔Ｄ_ｋｉは、図４に示すように、放映計画ｑ_ｉと尺ｄ_ｋとから算出される。

Here, τ is the average number of passers per unit time that passed near the digital signage, δ is the average stay time per unit time of passersby who stayed near the digital signage, and e is the bottom of the exponential function (number of Napiers) , D _ki is a broadcast interval from when the k-th advertisement is broadcast for the i-th time until the next broadcast starts. The broadcast interval D _ki is calculated from the broadcast plan q _i and the scale d _k as shown in FIG.

Moreover, the broadcast plan to be optimized shall satisfy all the following constraints. The first constraint condition is that all advertisements A _{k for} each designated number of broadcasts are allocated to the broadcast plan q _i . That is, equation (2) needs to be satisfied.

ここで、１（・）は、括弧内の等式が成立する時に１を返す関数である。

Here, 1 (•) is a function that returns 1 when an equation in parentheses is established.

The second constraint condition is that all advertisements A _k can be accommodated in all broadcastable time frames Z _i , in other words, the total of advertisement scales d _k assigned to the broadcastable time frames Z _i can be broadcasted. It is smaller than the time length of the frame Z _i . That is, it is guaranteed that the broadcast will not be interrupted in the middle of the advertisement, and the formula (3) needs to be satisfied.

ここで、ｄ（ｑ_ｉｊ）は、放映可能時間枠Ｚ_ｉのｊ番目の広告の尺を表している。

Here, d (q _ij ) represents the scale of the j-th advertisement in the broadcastable time frame Z _i .

  The third constraint condition is that advertisements of the same genre are not adjacent, in other words, competing genres are not continuous. That is, it is a condition for prohibiting the advertisement of the same industry from being arranged, and it is necessary to satisfy Expression (4).

ここで、ｇｅｎｒｅ（ｑ_ｉｊ）は、放映可能時間枠Ｚ_ｉのｊ番目の広告の業種や商品等のジャンルを表している。

Here, genre (q _ij ) represents the genre such as the business type or product of the j-th advertisement in the broadcastable time frame Z _i .

The effect index R (q _i ) of the formula (1) used in the present embodiment is called the number of advertisement views, and the expected value of the person who viewed the advertisement is calculated for each advertisement. This is the expected number of people. Even if the same person sees the same advertisement twice or more, only one person is counted, but when the same person sees another advertisement, the number of advertisements seen is counted.

A qualitative description of the number of advertisement views is as follows. The first term of equation (1) is the total exposure time _N k _{d k} ad _{A k,} scale _{d k} and designated broadcast times _{N k} and has the more exposure time the larger _N k _{d k} ad _{A k} is It becomes long and represents that many people contact the advertisement _Ak .

On the other hand, the second term of Equation (1) represents the effect of the broadcast interval _{D ki} ad _{A k,} the broadcast interval _{D ki} ad _{A k} becomes shorter, one person more than once, the same ad Since the probability of touching increases, the number of advertisement views decreases.

Therefore, any function may be used as long as the effect index value increases as the broadcast interval D _ki increases.

Further, in order to maximize the number of advertisement views, the broadcast interval D _ki may be increased uniformly while ensuring the exposure time N _k d _k . Since τ is the average number of people per unit time, even if the value of τ changes, it is obtained as a solution and does not affect the broadcast plan. On the other hand, since δ is the average residence time, the longer the residence time, the more people who see a plurality of advertisements, so the number of advertisement views increases.

Here, the second term of the effect index is defined as an exponential function. However, a monotonically increasing and downward convex function (for example, tanh) may be used for the broadcast interval D _ki that is a variable. . Then, a broadcast plan is generated in which the broadcast interval D _ki is evenly increased.

Note that the broadcast plan (solution) q _i that satisfies all of the first constraint condition to the third constraint condition is called a feasible solution. Since all ads A _k by the first constraint and the second constraint is assigned to broadcast possible time frame Z _i, whether the third constraint conditions are satisfied or are selective.

[About the search method of the broadcast plan]
Next, a broadcast plan search method executed when the broadcast plan update unit 104c updates the broadcast plan will be described. As described above, generating an optimized broadcast plan is to obtain an approximate solution to an optimization problem of a combination of broadcast plans that satisfy the application condition (constraint condition). In this embodiment, A technique called a multi-start local search (MLS method) is used.

  The local search method (LS method) starts from an appropriate initial solution (starting point) and sequentially searches for neighboring points, that is, evaluates neighboring points using an objective function, and evaluates the evaluation value. This is a method of finding a local optimum solution by repeating the procedure for updating the solution if becomes larger than the evaluation value so far.

  The multi-start local search method (MLS method) is a method of repeatedly executing a local search method while randomly generating an initial solution and storing the best local optimal solution obtained as a result.

  The multi-start local search method is disclosed in Non-Patent Document 3.

[Operation flow of the broadcast plan generator]
Next, the operation flow of the broadcast plan generation apparatus will be described. In the operation flow described below, instead of maximizing the above-described effect index R (q _i ), the evaluation function f (q _i ) of Expression (5) in which the number of violations of each constraint condition is added as a penalty is used. Therefore, the evaluation function is minimized.

この評価関数ｆ（ｑ_ｉ）は、第１の制約条件〜第３の制約条件に違反する違反回数と効果指標Ｒ（ｑ_ｉ）との差からなる関数である。λは、ペナルティの重みパラメータである。違反回数とは、各制約条件を満たさなかった際にカウントされる数であり、第１の制約条件の場合には、放映計画に割り付けられなかった広告の数であり、第２の制約条件の場合には、コンテンツを収められなかった放映可能時間枠の数であり、第３の制約条件の場合には、隣り合った広告の数である。

This evaluation function f (q _i ) is a function composed of a difference between the number of violations violating the first to third constraint conditions and the effect index R (q _i ). λ is a penalty weight parameter. The number of violations is a number counted when each constraint condition is not satisfied. In the case of the first constraint condition, the number of violations is the number of advertisements that are not allocated to the broadcast plan. In the case of the third restriction condition, it is the number of adjoining advertisements.

As shown in FIG. 5, the evaluation function f (q _i ) is monotonously decreased with respect to the broadcast interval D _ki . When the application condition (constraint condition) is severe, there may be no feasible solution that satisfies all the constraint conditions. However, by adopting the evaluation function f (q _i ) with a penalty term, Even when the constraint conditions are not satisfied, it is possible to generate a broadcast plan that is close to the optimal solution.

  Next, the operation flow of the broadcast plan generation apparatus will be described. FIG. 6 is a diagram illustrating an overall operation flow of the broadcast plan generation apparatus.

First, the neighborhood search method execution unit 104 initializes the best broadcast plan (hereinafter referred to as the best solution) q _min at the present time as empty (φ), and initializes the evaluation value f _min of the best solution as ∞. (S101).

Next, an arbitrary broadcast plan (hereinafter referred to as an initial solution) q _i is generated by the initial solution generation unit 104a using the advertisement application conditions and the frame information (S102). A method for generating the initial solution q _i will be described later.

Next, by executing the neighborhood search method based on the multi-start local search method by the effect index planning unit 104b and the broadcast plan update unit 104c, while searching for an intermediate solution that is close to the initial solution starting from the initial solution q _i An improved solution q _i ′ that approximately minimizes the evaluation function f and an evaluation value f (q _i ′) of the improved solution q _i ′ are obtained (S103). The improved solution q _i ′ and the evaluation value f (q _i ′) of the improved solution q _i ′ are obtained while satisfying the first constraint condition to the third constraint condition. The calculation of the evaluation value f (q _i ′) requires the number of violations of each constraint condition. The number of violations is appropriately counted when searching for a nearby intermediate solution in the broadcast plan update unit 104c. The neighborhood search method will be described later.

Next, the broadcast plan update unit 104c determines whether or not the evaluation value f (q _i ') of the improved solution q _i ' obtained in S103 is smaller than the evaluation value f _min of the best solution so far (S104). ).

Next, as a result of the determination in S104, when f (q _i ′) <f _min , the best solution is obtained, and therefore the best solution q _min and the evaluation value f _{min of the} best solution are respectively improved solutions q _i ′ and the evaluation value f (q _i ′) of the improved solution q _i ′ are replaced (S105). On the other hand, if the result of determination in S104 is other than f (q _i ′) <f _min , the solution so far is the best, so S105 is not executed and the process proceeds to S106 in the subsequent stage.

  Next, the broadcast plan update unit 104c determines whether a predetermined end condition is satisfied (S106). If the end condition is not satisfied, the process returns to S102 and the processing up to S105 is repeated. If the end condition is satisfied, the operation flow is ended. An example of the termination condition is that the number of iterations of the loop of S102 → S103 → S104 → (S105) → S102 exceeds a predetermined number.

[Initial solution generation method]
Here, two methods for generating an initial solution in S102 of the above-described operation flow will be described.

  First, a first initial solution generation method obtained by modifying an FFD (First Fit Decreating) method, which is a basic heuristic of the bin packing method, will be described. In this generation method, if an advertisement with a large scale remains until the end, it is likely that the second constraint will be violated without being within the broadcastable time frame. Is the method.

FIG. 7 is a diagram showing a flow of generating the first initial solution. First, the broadcast plan (solution) q _i (i = 1, 2,..., N _z ) is initialized with the sky (φ) (S1001).

Next, the advertisements A _k are sorted in descending order of the scale d _k (S1002). That is, d _k ≧ d _{k + 1} is established for all k = 1, 2,..., N−1.

Next, the following processes of S1004 to S1007 are repeated for all advertisements A _k (k = 1, 2,..., N) (S1003).

Next, the following processes of S1005 to S1007 are repeated for the designated broadcast count _Nk (S1004).

Next, the broadcastable time frame Z _i (i = 1, 2,..., N _z ) is sequentially scanned (S1005), and the advertisement _Ak is inserted into the first broadcastable time frame Z _i (q _i = q _i +). {K}) (S1006, S1007).

In the first initial solution generation method described above, the broadcastable time frame is scanned in the order of Z ₁ , Z ₂ ,..., _ZNz , but the broadcastable time is used to make the initial solution random. The scanning start position of the frame can also be determined by a random number. That is, the scan start position i ₀ of the broadcastable time frame is a random number that takes an integer of 1 to N _z , and the broadcastable time frames are inspected in the order of i = i ₀ , i ₀ +1,. Then, the head i = 1 is returned to perform scanning.

  Next, a second initial solution generation method obtained by modifying the round robin method used for process management of the operating system will be described. This generation method is a heuristic inspired by the round robin method, and is a method in which unallocated advertisements are sequentially assigned to broadcastable time frames so that advertisement broadcast intervals are as uniform as possible.

FIG. 8 is a diagram showing a flow of generating the second initial solution. First, broadcast schedule _{(solutions) q i (i = 1,2,} ..., N z) is initialized with empty (phi), unallocated advertising number _{r k} ad _{A k} is initialized with _{N k} (S1101).

Next, for k = 1, 2,..., N, _k that maximizes the ratio r _k / N _{k of the} number of unallocated advertisements is obtained (S1102).

Next, the broadcastable time frame Z _i (i = 1, 2,..., N _z ) is sequentially scanned (S1103), and the advertisement _Ak is inserted into the first broadcastable time frame Z _i (q _i = q _i +). {K}) (S1104, S1105).

Then, unallocated advertising number _{r k} is decremented by 1 (S1106), if the finished allocation of all advertising, i.e. _{r k} is equal 0 (zero), the process ends (S1107).

  The above two are examples of initial solution generation, and other generation methods can be used. However, since it takes time to converge the solution if the initial solution is generated at random, it is desirable to start from an initial solution that is close to the optimal solution. In addition, an initial solution that satisfies the first to third constraint conditions may be generated.

[How to generate the best solution]
Here, the best solution generation method in S103 to S106 of the above-described operation flow based on the neighborhood search method will be described.

FIG. 9 is a diagram showing a flow for generating the best solution. First, the broadcast plan (initial solution) q _i (i = 1, 2,..., N _z ) generated by the initial solution generation unit 104a is input, and the evaluation value f of the best solution is obtained using the initial solution q _{i. min} is initialized with f (q _i ) (S1201).

Next, a neighborhood solution (corresponding to the above-described improved solution and conceptually a nearby point) q _i ′ of the initial solution (or intermediate solution) q _i is generated (S1202). Here, a procedure for generating a neighborhood solution will be described below. FIG. 10 is a diagram schematically showing a neighborhood solution generation procedure.

First, an advertisement number list Q ← q ₁ + q ₂ +... + Q _Nz connecting the initial solutions (or intermediate solutions) q _i is generated (S1202a).

Next, if the initial solution (or intermediate solution) q _i is an infeasible solution that does not satisfy the first constraint condition and there is an unallocated advertisement, the advertisement number of the unallocated advertisement is added to the end of Q (S1202b).

Next, two elements Q _l and Q _m are randomly selected from the advertisement number list Q and are exchanged to generate a new advertisement number list Q ′ (S1202c).

Next, the element Q _j is extracted in order from the top of the new advertisement number list Q ′, and if it fits in the broadcastable time frame Z _i (i = 1, 2,..., N _z ), it is packed (q _i ′). ← q _i '+ {Q _j }), and when Z _i is full, it is repeated to fill the next frame (S1202d).

A neighborhood solution q _i ′ is generated by the processing of S1202a to S1202d. In order to speed up the search convergence, it is also effective to preferentially select elements that do not satisfy the first to third constraints in S1202c.

Next, the evaluation value f (q _i ′) of the generated neighborhood solution q _i ′ is obtained using the evaluation function f (q _i ) (S1203).

Next, it is determined whether or not the obtained evaluation value f (q _i ′) of the neighborhood solution q _i ′ is smaller than the evaluation value f _min of the best solution so far (S1204).

Next, as a result of the determination in S1204, if f (q _i ′) <f _min , the best solution has been obtained, so the evaluation value f _min of the best solution is updated with the evaluation value f (q _i ′). Then, the initial solution (or intermediate solution) q _i of S1202 is replaced with the improved solution q _i ′ (S1205), the process returns to S1202, and a search is further made for a better solution near the intermediate solution q _i .

On the other hand, if the result of determination in S1204 is other than f (q _i ′) <f _min , the previous solution is the best, and therefore the initial solution (or intermediate solution) is not executed without executing S1205. It is checked whether or not all the neighborhoods of q _i have been searched (S1206). When all the neighborhood searches are completed or the search for neighboring points is completed a predetermined number of times, the processing is terminated.

  In this embodiment, the multi-start local search method is used as an approximate solution for combination optimization, but an annealing method, an iterative local search method, or the like can also be used. In addition, although there is a problem that calculation time is required, it is possible to use an exact solution such as a branch and bound method instead of an approximate solution.

In the present embodiment, the first to third constraint conditions are set corresponding to the advertisement application conditions, but various other constraint conditions can also be set. For example, it is possible to expand to handle content that has not been determined, such as a still image. In addition, when the minimum value and the maximum value of the scale of an advertisement are given by d (q _ij ) (“−” below d) and d (q _ij ) (“−” above d), respectively, It is also possible to transform the second constraint condition expressed in (3) into Expression (6).

  It is also possible to prepare filler (filled grass) content in advance and modify the filler content to be inserted when the second constraint condition is not satisfied with an equal sign, that is, when a gap is formed. .

  According to the present embodiment, the length of each advertisement and the number of designated broadcasts that are expected to be broadcast on digital signage, and the plurality of broadcastable time frames in which the advertisement can be broadcast are stored in the application condition. Which is stored in the unit 101 and the frame information storage unit 102, and the neighborhood search method execution unit 1604 uses the time length of the content, the designated number of broadcasts, and the broadcastable time frame respectively acquired from the storage units. The first constraint condition that all broadcast advertisements for each designated number of broadcasts are allocated to the broadcast plan, and an arbitrary broadcast plan that shows in what order the advertisements are allocated and broadcast in the broadcastable time frame, and all Good pre-defined broadcast plan while satisfying the second constraint that all advertisements fit within the broadcastable time frame of and the third constraint that ads of the same genre are not next to each other Since any broadcast plan is optimized so that the effect index to be evaluated is maximized, the advertiser can specify the application conditions in detail, and it has been optimized based on preset criteria while satisfying the application conditions (constraints) Broadcast plans can be generated automatically.

  According to the present embodiment, the number of contents that do not satisfy the first constraint, the number of broadcastable time frames that do not satisfy the second constraint, the number of advertisements that do not satisfy the third constraint, and the effect index, The difference between is an evaluation function, starting from an arbitrary broadcast plan, searching for a broadcast plan that is close to the broadcast plan, and searching for a broadcast plan that minimizes the evaluation function, so if all constraints are not met Even so, it is possible to generate a broadcast plan that is close to optimal.

  Finally, the broadcast plan generation apparatus described in the present embodiment is configured by a computer. That is, the advertisement application condition storage unit 101 and the frame information storage unit 102 are realized by a storage unit such as a memory. Further, the advertisement condition input unit 103, the neighborhood search method execution unit 104, and the broadcast plan output unit 105 are realized by a calculation unit such as a CPU and executed by a program.

  In addition, the broadcast plan generating device described in the present embodiment is recorded as a program in a recording medium such as an optical storage device or a magnetic storage device so as to be readable, and this recording medium is incorporated in a computer or recorded on a recording medium. By downloading the program to a computer via an arbitrary communication line or installing it from a recording medium and operating the computer with the program, each processing operation described above can function as a broadcast plan generation device. Of course.

101 ... Advertising application condition storage section (storage means)
102... Frame information storage unit (storage means)
103 ... Advertising condition input unit 104 ... Neighborhood search method execution unit (generation unit / optimization unit)
104a ... Initial solution generation unit (generation means)
104b ... Effect index calculation unit (optimization means)
104c ... Broadcast plan update unit (optimization means)
104d ... Broadcast plan storage (optimization means)
105 ... Broadcasting plan output unit 201 ... Distribution server 202 ... Player 203 ... Display 204 ... Operation terminal S ... Step

Claims (5)

In a broadcast plan generation device that generates a broadcast plan for content broadcast on digital signage,
Storage means for storing each length of time of each of a plurality of contents expected to be broadcast on digital signage and each designated number of broadcasts, and a plurality of broadcastable time frames capable of broadcasting the contents;
Using the time length of the content acquired from the storage means, the specified number of broadcasts, and the broadcastable time frame, generate an arbitrary broadcast plan that indicates in what order the content is allocated and broadcast in which broadcastable time frame Generating means for
While satisfying the first condition that all content for each designated number of broadcasts is allocated to the broadcast plan and the second condition that all content fits within all broadcastable time frames, possess and optimization means effects index for evaluating the merit to optimize the arbitrary broadcast program so as to maximize the,
The effect index is
^{_{τΣ k = 1 n {N k}} d k + δ (1 + Σ i = 1 Nk-1 (1-e - (Dki-dk) / δ))} ( where, tau digital signage unit time thereabouts passed through the vicinity of the Average number of passengers, N _k is the designated number of broadcasts of the k-th content, d _k is the time length of the k-th content, δ is the average residence time per unit time of passers-by staying near the digital signage, and e is an index A broadcast plan generation apparatus , wherein D _ki is a function of the bottom of the function (the number of Napiers), and D _ki is a broadcast interval from when the k-th content is broadcast for the i-th time until the next broadcast starts . The optimization means includes
The difference between the number of contents that do not satisfy the first condition and / or the number of broadcastable time frames that do not satisfy the second condition and the effect index is used as an evaluation function, and starts from the arbitrary broadcast plan and is close to the broadcast plan The broadcast plan generation apparatus according to claim 1, wherein the broadcast plan that minimizes the evaluation function is searched while searching for a special broadcast plan. The nearby broadcast plan is
The broadcast plan generation device according to claim 2, wherein the broadcast plan generation device is generated by changing a part of elements constituting the broadcast plan. In the broadcast plan generation method performed by the broadcast plan generation device that generates the broadcast plan of the content broadcasted by digital signage,
The broadcast plan generation device includes storage means,
A first step of storing, in the storage means, each time length and each designated number of broadcasts of a plurality of contents expected to be broadcast on digital signage, and a plurality of broadcastable time frames in which the contents can be broadcast;
Using the time length of the content acquired from the storage means, the specified number of broadcasts, and the broadcastable time frame, generate an arbitrary broadcast plan that indicates in what order the content is allocated and broadcast in which broadcastable time frame A second step of:
While satisfying the first condition that all content for each designated number of broadcasts is allocated to the broadcast plan and the second condition that all content fits within all broadcastable time frames, A third step of optimizing the arbitrary broadcast plan so that an effect index for evaluating goodness is maximized,
The effect index is
τΣ _{k = 1} ⁿ {N _k d _k + Δ (1 + Σ _{i = 1} ^Nk-1 (1-e ^{− (Dki−dk) / δ} ))} (Where τ is the average number of people per unit time passing near the digital signage, N _k Is the designated broadcast count of the kth content, d _k Is the time length of the k-th content, δ is the average residence time per unit time of a passerby who stayed near the digital signage, e is the bottom of the exponential function (number of Napiers), D _ki Is a function of a broadcast interval from the time when the k-th content is broadcast for the i-th time until the start of the next broadcast). A broadcast plan generation program for causing a computer to function as the broadcast plan generation device according to any one of claims 1 to 3.

Images