JP2009516952A - Optimal selection of TV programs by time margin - Google Patents

Abstract

A system 800, apparatus 700, and method for selecting the best set of TV programs for reception and recording by at least one tuner are provided. A TV program is broadcast over a set of channels during a given time [b, e] having a start time b and an end time e. Some and all of the TV programs are considered, and for each or all of the TV programs, user preference values are provided for their reception and recording. In addition, pre-roll and post-roll are included to ensure that the partial TV program is received entirely. A minimum cost network algorithm is used that solves the selection and viewing problem but does not necessarily need to be optimal. Additional time is allowed to ensure that the entire part of the TV program has been received before or after the reception of the part of the TV program.

Description

  The present invention uses a time margin to make a (almost) optimal selection of partial television programs to be recorded using a limited number of tuners during a given time interval. Relates to a system, apparatus and method that distinguishes between switching to another TV channel after recording a program and staying tuned to the same channel by forcing a time margin in the previous situation. A part is defined as part or all of a TV program.

  Personal video recorders (PVRS) can be used to record TV programs on either hard disks or DVDs. Programming these PVRs is typically accomplished using an electronic program guide (EPG) by simply clicking on the TV program to be recorded. In addition, for an incoming TV program, a recommender may be available that predicts how many viewers will like the program.

  With the increasing number of available TV programs via terrestrial, satellite or cable connections, the viewer's task of selecting TV programs to watch or record is rapidly increasing and cannot be handled manually. With the advent of digital television, the number of channels and hence the number of options to choose from has increased. Thus, the viewer can no longer overview all available programming content and may miss the TV program that the viewer is interested in.

  Due to the large number of TV programs available each week, printed TV program guides are cumbersome and electronic program guides (EPGs) are being developed as a solution to this problem. EPG provides TV programs that are available for multiple channels on a TV screen. However, only a limited number of TV programs can be displayed on the TV screen at once, i.e. the number of channels and the length of the time interval of the part displayed on the screen are very limited. This results from the poor resolution of the TV screen for the display of this kind of texture information. Also, even if the resolution is high enough, the task of selecting from a large number of available TV content afflicts the average viewer.

  The viewer's problem of which partial TV program should be selected and when and how to arrange the viewing time among various TV channels, ie the selected partial TV program is completely received In order to improve the problem of developing a viewer-specific viewing schedule that guarantees that, EPG offers the option to search (filter) by keyword, but the time between selections will complete the reception of the program Not allowed. In this way, EPG can reduce the number of TV programs to be screened and selected by the viewer to a manageable number, but due to the lack of partial TV program termination resulting from switching channels. It cannot be dealt with. Another way for viewers to thin out TV programs that are not of interest to multiple viewers is through the use of recommender systems based on artificial intelligence (AI) technology, but there is a lack of the same termination problem. These AI recommender systems maintain a viewer's favorite profile that shows what they like and dislike, and to what extent viewers are newly provided using this profile Each newly provided TV program is scored as to whether it likes the TV program. The TV programs scored above a predetermined viewer-specific tolerance are then highlighted in the EPG or the viewer is provided with a list of these TV programs. However, the problem with these solutions is that they do not consider whether the TV programs overlap in time, so the viewers still have a higher-scoring TV program to receive. The viewer's specific schedule must be edited to allow enough time between them for the end of reception of the selected part. In other words, filters and recommenders provide a list of individual TV programs, so such a list is an incomplete solution to the problem of viewers scheduling TV programs to watch, especially if the channel is There is no way to leave enough time between the partial TV programs selected to allow termination when switched between receptions. A similar situation exists for recording TV programs.

  Only a finite number of M ≧ 1 tuners available for receiving / recording a selected partial TV program with a given time interval that allows an end time when the channel needs to be switched There is a need for a solution for efficient selection of partial TV programs for a given time interval that maximizes a predetermined viewer preference when the viewer has.

  The system, apparatus and method of the present invention include a (substantially) optimal of partial TV programs with a limited number of tuners, including a time margin in which channels are switched between successive partial TV programs. It provides an effective algorithm that is guaranteed to make the right choice. That is, the selection substantially maximizes the sum of viewer preference values associated with the selected partial TV program while ensuring complete reception. These preferences are assumed to be known to recommenders who provide initial filtering of available partial TV programs before (substantially) optimal selection is made using the present invention, for example. The

  Given a time interval in which the viewer wishes to select a partial TV program, all partial TV programs included in this interval are first collected, and each collected partial TV program A function of values is obtained. Secondly, the start and end time points of each partial TV program are determined as nodes of the directed graph G, and set S is the start and end time points within a given time interval. For partial TV programs having at least one of Third, a time 0 source and a time ∞ sink node are formed, and the G graph is drawn from the source through the G time-ordered nodes to the sink, ie, the central timeline or edge A chain is formed from the source node through each successive time-ordered node of G, each edge having a cost of “0” and a capacity of M, where M ≧ 1 is a TV program Is the number of tuners available to receive

  Next, a similar timeline for channel c is constructed for the partial TV programs in S broadcast on each channel c, and the TV channels already received and those not received at a given time. A distinction is made between That is, in addition to the nodes already included in the central timeline, each TV program in S that is broadcast on TV channel c is represented by a pair of start and end nodes, and channel c instead of the central timeline. It is further divided into zero or more partial TV programs with edges between them on the timeline. This allows each TV program to start receiving at its start time or the end time of another TV program, and each TV program starts at its end time or the start time of another TV program. Reception can be terminated. Each such resulting partial TV program is added to S and has an associated viewer preference value determined by a function of the viewer specific value. As will be shown below in the detailed description, the loss of optimality does not arise from this constraint (see Theory 1 below).

  What is important in the present invention is a graph G including a central timeline from which the entire or partial TV program edge included is executed, and in the timeline of channel c. , For each TV channel c, this edge represents the negative preference values for all and part of these TV programs, all or part of this TV program being included in the set S.

  Finally, the time margin is included in this model, i.e. the graph, as the time difference of the edges that run from the central timeline to the timeline of a channel, and this time margin is parallel to the timeline of a channel Required to remove the edge and properly represent the viewer's preference value in the timeline for that channel.

  Given a graph G with this direction, the least-cost network algorithm is applied to this graph to determine a subset of the partial TV program S ′ for a given time interval, which is almost the largest viewer. Satisfaction, i.e., the overall audience preference value for the partial TV program set S 'is substantially maximized. The selected subset is guaranteed to be approximately optimal. The algorithm for selecting the subset is guaranteed to run in polynomial time.

  The system, apparatus and method of the present invention provides a way to determine the optimal TV experience for the viewer, which is a partially preferred TV program using at least one tuner. Recording as well as watching.

  The system, apparatus and method of the present invention provide mathematical formulas and solutions to obtain an optimal solution to the viewer's TV program scheduling problem. The scope of the system, apparatus and method of the present invention is based on a predetermined number of available tuners based on predetermined viewer preferences for all available TV programs being broadcast at a given time. Which TV program should be selected for viewing or recording.

  Simplified problem setup was first explained and simplified to detail the algorithms developed for the simplified problem, including more factors to solve more complex problem setups Problem settings continue to expand.

In the preferred embodiment of the system, apparatus and method of the present invention, several assumptions are made first.
1. A preference value or score is provided for each of a plurality of candidate TV programs and indicates how many viewers prefer the candidate TV program.
2. Each preference score is given explicitly by the viewer or by the recommender tool.
3. The overall preference score for a number of selected TV programs is given by the linear sum of the provided preference scores for the individual TV programs.
4). The TV program selected for viewing is assumed to be viewed at the moment the TV program is broadcast, and if the TV program is viewed later, it is simply assumed that it is recorded during the broadcast. The

  Each simplification and extension is modeled as a directed graph that transforms the problem into a minimum cost network flow problem. Since the minimum cost network flow problem can be solved in polynomial time, the corresponding TV program selection problem is effectively solved.

  The first model uses a single tuner to select a complete TV program. The problem of selecting a complete TV program is defined as follows:

により与えられるこのサブセットの全体の値は最大にされる。 Definition 1 (Program selection problem (PSP)). A set S of TV programs s is provided, and for each TV program s, a start time bs and an end time es are also defined. Furthermore, for each TV program, the value vs> 0 is given, which reflects the overall value when this TV program is (completely) viewed. The problem is to determine the subset S′⊆S, so that TV programs in the subset S ′ do not compete in time, ie bs> et for all s, tεS ′, s ≠ t. Vbt> es,
(Outside 1)

The overall value of this subset given by is maximized.

  The TV program value vs is either given explicitly by the viewer or estimated using a recommender. Considering the recommender score ps of the TV program s as a preferred specific gravity, the overall value is given by vs = ps (es−bs). Maximizing the overall value across multiple TV programs in a given time interval corresponds to maximizing the average preference specific gravity.

  The TV program selection problem is formulated as a minimum cost flow problem as follows. The set V of nodes of the network G is given by a source node and a sink node and a node for the start or end time of each TV program. Therefore, it is shown below.

ここでノード０はソースを表し、∞はシンクを表す。Ｖにおけるノード（すなわち時間）をｔ1,…,ｔnで示し、ｎ個の個別の時間は、増加する順序で配置される。ここで、全ての時間は、ソース時間ｔ1＝０よりも大きく、シンク時間ｔn＝∞よりも小さいことが想定される。

Here, node 0 represents a source and ∞ represents a sink. The nodes (ie, time) in V are denoted by t1,..., Tn, and the n individual times are arranged in increasing order. Here, it is assumed that all the times are larger than the source time t1 = 0 and smaller than the sync time tn = ∞.

  The edge set E consists of two parts. First, by definition, there is an edge on the source-to-sink timeline that contains the node edge (ti, ti + 1) for all i = 1,..., N−1 with zero cost. Second, for each TV program sεS, an edge (bs, es) is added with a cost -vs. All edges are assumed to have capacity 1.

  For an example graph structure, consider the set of TV programs shown in FIG. 1 that covers TV programs A, B,..., J in the time interval from 18: 00h to 22: 00h. The corresponding set V of time points is given by F = {0, 18:00, 19:00, 19:30, 20:00, 20:30, 21:00, 22:00, ∞} The set is as given in FIG.

  Here, determining the non-overlapping subset with the maximum corresponds to finding one flow in G from source to sink with the lowest cost. In this minimum cost flow, if the edge corresponding to a TV program holds a flow of 1, this TV program is included in S ', otherwise it is not included in S'. Thus, the problem is solved by a least cost flow algorithm. In other words, PSP is solved efficiently.

  As in the previous equation, all capacities are 1, the least cost flow problem is reduced to the shortest path problem (from source to sink), and the (negative) cost is used as the edge length. The

  Minimum cost flow problem. Suppose that a directed graph G = (V, E) is given. Assume that there is a cost cij and capacity uij = 1 associated with each edge ij at E and a demand (supply) bv associated with each node v at V. The problem is the supply / demand constraints for each node

及びエッジの制約

And edge constraints

を満たす最小コストのフローを発見することである。フローのコストは、以下のように定義される。

It is to find the flow with the minimum cost that satisfies The cost of the flow is defined as follows:

本発明では、定義により、それぞれのエッジのキャパシティは１である。さらに、ｂ1＝−１，ｂn＝１、及びｉ＝２,…,ｎ−１についてｂi＝０である。 (Outside 2)

In the present invention, by definition, the capacity of each edge is 1. Further, bi = 0 for b1 = -1, bn = 1, and i = 2,..., N-1.

  A large number of TV programs can be received in parallel whenever more than one tuner is available and the selected TV program is recorded. The problem is also related to a two-stage TV programming approach, in which first it is determined which TV program is received and then stored in PM (Persistent Memory) and then recorded in PM. TV schedule is determined from all contents.

  Definition 2 (multi-tuner program selection problem (MPSP)). A set S of TV programs is given, and for each TV program s a start time bs, an end time es and a value vs are given. In addition, a predetermined number M of tuners are also provided. The problem is to determine the subset S′⊆S. This selects at most M TV programs at all times, ie for all times x

であり、
（外３）

により与えられるこのサブセットの全体の値が最大にされる。

And
(Outside 3)

The overall value of this subset given by is maximized.

  MPSP is resolved in a manner similar to PSP using the equivalent network flow transformation described above. For MPSP, the edge (ti, ti + 1) capacity changes to M, reflecting the availability of M tuners, and it is desirable that a program be received over multiple purchases. The capacity of the edge of the TV program remains 1 so that there is no.

  Next, the minimum cost flow of M from the source to the sink is determined and each TV program whose corresponding edge in the graph holds the (one) flow in this discovered minimum cost flow solution is S 'include.

  Next, the single tuner solution described above for a complete TV program is extended to include viewing some TV programs only partially. For this reason, the value function vs (l, r) is introduced for each TV program s indicating the value to watch the program during the partially opened interval [l, r), with bs ≦ l ≦ r ≦ es. Is done. For a function of this value, the following is assumed:

定義３（部分的な番組選択問題（ＰＰＳＰ））。ＴＶ番組のセットＳが与えられ、それぞれのＴＶ番組ｓについて、開始時間ｂsと終了時間ｅsが与えられる。さらに、それぞれのＴＶ番組について、値の関数ｖsが与えられる。問題は、それぞれのＴＶ番組ｓについて、
（外４）

として、番組を視聴するインターバルの（おそらく空である）セットＩs＝｛［ｌ1,ｒ1），．．．,［ｌk_s,ｒk_s］｝を決定することであり、これにより、ＴＶ番組の選択されたインターバルは、時間的にコンフリクトしない。このことは、全てのｓ,ｔ∈Ｓ’，ｓ≠ｔ及び全ての［ｌ,ｒ）∈Ｉs，［ｌ’,ｒ’）∈Ｉtについて、ｌ≧ｒ’ ｖ ｌ’≧ｒを有し、すなわち、ｓの任意のインターバルは、ｔの任意のインターバルよりも遅く、逆も然りであることを意味する。なお、半分の開いたインターバルを使用したので、１つのインターバルは、別のインターバルが終了するのと同じ瞬間で正確に開始することが可能である。目的は、
（外５）

により与えられる選択されたインターバルの全体の値が最大にされることである。

Definition 3 (Partial Program Selection Problem (PPSP)). A set S of TV programs is given, and a start time bs and an end time es are given for each TV program s. In addition, a value function vs is given for each TV program. The problem is that for each TV program
(Outside 4)

, A set of intervals (probably empty) for watching the program Is = {[l1, r1),. . . , [Lk _s , rk _s ]} so that the selected interval of the TV program does not conflict in time. This has l ≧ r ′ v l ′ ≧ r for all s, t∈S ′, s ≠ t and for all [l, r) ∈Is, [l ′, r ′) ∈It. That is, any interval of s is slower than any interval of t, and vice versa. Note that because we used half open intervals, one interval can start exactly at the same moment that another interval ends. My goal is,
(Outside 5)

The overall value of the selected interval given by is maximized.

  The value function is mostly of arbitrary form, and in the following description it is constrained to make the value function convex and will be explained in the next section. A more limited form that can be handled more efficiently is a function of linear values with penalties and is described below.

Convex value function.
The convex value function has the following relationship for all α∈ [0,1]

が保持される条件を満たす値の関数ｖである。

Is a function v of a value that satisfies the condition that is held.

  Most actual value functions satisfy this condition. For example, the convex value function is

のような線形の関数であるか、又は

Or a linear function such as

のような二次の関数である。

A quadratic function such as

  One property of the convex value function is that it reduces the number of points that can switch from one TV program to another, as the following theory states.

  Theory 1. For all TV programs s ∈ S, if vs (l, r) is convex, PPSP can be used for all s ∈ S and [l, r) ∈ Is, without loss of optimality.

によるソリューションに制限される。

Limited to solutions by

  In other words, a TV program starts receiving only at its start time or at the end time of another TV program, and ends receiving only at that end time or at the start time of another TV program.

  Proof. Consider an interval [l, r) εIs of a TV program sεS that does not satisfy equation (1). Just before time l, the maximum number of tuners needs to be used, otherwise 1 can be lowered to improve the solution. When reception of TV program s occurs forward from time l, there must be a TV program tεS that stops being received at time l, ie, It is the interval [l ′, r ′). Since l does not satisfy Expression (1), r ′ <e. Consider two alternative solutions, one with l replaced by l1 = 1−ε and r ′ replaced by r1 ′ = r′−ε = l1 and one replaced by l2 = 1 + ε. with r 'replaced by l and r2' = r '+ ε = l2. Compared to the original solution, the cost of the first solution is the amount and value of the original solution

だけ異なり、第二のソリューションのコストは、量

The only difference is the cost of the second solution, the amount

だけ異なる。

Only different.

  Therefore, follow the following.

この場合、後段の不等号は、ｖs及びｖtの凸型の特性に基づいている。したがって、少なくとも１つの差Δ1及びΔ2は、負ではなく、すなわち、２つの代替的なソリューションのうちの少なくとも１つは、少なくともオリジナルのソリューションと同様に良好である。したがって、最適さの損失なしに、ｌ及びｒ’の何れかをそれらがｅt又はｒになるまで増加することができるか（後者のケースでは、インターバル［ｌ,ｒ］はＩsから除かれる）、又は、ｌ及びｒ’をそれらがｂs又はｌ’になるまで減少することができる（後者のケースでは、［ｌ’,ｒ’）はＩtから除かれる）。
同様なやり方で、式（２）が最適さの損失なしに想定されることを証明できる。

In this case, the latter inequality sign is based on the convex characteristics of vs and vt. Thus, the at least one difference Δ1 and Δ2 is not negative, ie at least one of the two alternative solutions is at least as good as the original solution. Therefore, can either l and r ′ be increased until they are et or r without loss of optimality (in the latter case, the interval [l, r] is removed from Is), Alternatively, l and r ′ can be reduced until they are bs or l ′ (in the latter case, [l ′, r ′) is removed from It).
In a similar manner, it can be proved that equation (2) is assumed without loss of optimality.

  As a result of Theory 1, assuming that the value function is convex, it is necessary to consider a limited number of time points for starting and stopping partial viewing of a TV program. Based on this, the least cost network of FIG. 2 is expanded as shown in FIG. The cost of the TV program edge is determined by the value for the corresponding part of the TV program. The optimal flow in the expanded graph corresponds to the PPSP optimal solution. In such a flow, multiple edges are used per program, corresponding to receiving TV programs during multiple intervals. For example, the optimal flow is performed through edges I1 and I7 in the expanded graph, meaning that TV program I is selected from 19: 0 to 19:30 and 20: 0 to 21:00. . Since there is only one flow, the same time interval is never covered twice. Furthermore, selecting two consecutive portions is at most as good as selecting the combined portions (eg, selecting I5 is at least selecting I1 and I2). As well).

The number of edges in the expanded graph is much higher than in the original graph, but is bounded by a polynomial. More precisely, the number of edges per TV program is quadratic in the number of different time points. Furthermore, since there is only one tuner, the number of edges is limited to the best number between each pair of time points, thereby limiting the total number of edges to O (n ² ).

A function of linear values with penalties.
In this alternative, the value function equation is further limited to a linear function with a penalty to end the TV program early or to start the TV program later. More formally, the formula

の値の関数を考え、

Given a function of the value of

は、ＴＶ番組ｓが早めに終了するか、送れて開始する場合にペナルティを与える関数である。なお、この値の関数は、式（３）の右手側における全ての３つの項が凸型である時（すなわちペナルティ関数が凹型であるとき）、想定に従い、凸型である。式（３）における値の関数が与えられると、図４に示される別の拡張された最小のコストネットワークを定義することができる。より正式には、このグラフは、以下のように構築される。

Is a function that gives a penalty when the TV program s ends early or starts being sent. The function of this value is convex according to the assumption when all three terms on the right hand side of Equation (3) are convex (that is, when the penalty function is concave). Given the function of values in equation (3), another extended minimal cost network shown in FIG. 4 can be defined. More formally, this graph is constructed as follows:

The set of nodes V is firstly associated with each time point t1,. . . , Tn. Then, for each TV program with time points bs = ti and es = tj, where j−i> 1, the extra node t _k ^s is the respective intermediate time point, ie k = i + 1,. It can be constructed for j-1. In FIG. 4, the node labeled “19: 00A” is an example of a node relating to an intermediate time point.

The set of edges that all have capacity 1 is defined as follows: First, there is an edge (ti, ti + 1) for all i = 1,..., N−1 as before having zero cost. Second, consider each TV program sεS with time points bs = ti and es = tj, and distinguish between the two cases as follows.
If j−i = 1, add edge (bs, es) with cost −vs.
If j−i> 1, add many edges.

ここで、ペナルティをもつ線形の値の関数によるＰＰＳＰは、このグラフにおけるソースからシンクへの１つの最小のコストフローを発見することに対応する。さらに、ＴＶ番組当たりの多数の部分は、図３のグラフによるように選択される場合がある。

Here, PPSP with a function of a linear value with a penalty corresponds to finding one minimum cost flow from source to sink in this graph. Furthermore, a number of parts per TV program may be selected as per the graph of FIG.

Consider further the example in FIG. 4 for the correspondence between the flow in the graph and the partial TV program. TV program A is completely selected when both edge (18:00, 19: 00A) and edge (19: 00A, 19:30) form one flow. In this situation, the edges (19:00, 19: 00A) and (19: 00A, 19:00) do not form a flow. The cost of the flow through these two previous edges accurately sums -v _A corresponding to the TV program A selection value v _A. Selecting only the first part of A corresponds to one flow through edges (18:00, 19: 00A) and (19: 00A, 19:00), resulting in a cost of -2v _A / 3 + c _A ^fe , ie the value 2 v _A / 3 for this first part and the penalty c _A ^fe that terminates A early. Selecting only the second part of A corresponds to one flow through edges (19:00, 19: 00A) and (19: 00A, 19:30), resulting in a cost −v _A / 3 + c _a ^sl, that is, the penalty c _a ^sl starting this second part after the value v _a / 3 and a.

  The total number of nodes is O (| S | n) and the number of edges is

であるので、拡張されたグラフにおけるノード及びエッジの数は、この例のサイズにおいて多項式で境界付けされている。

Thus, the number of nodes and edges in the expanded graph is bounded by a polynomial in the size of this example.

  The next extension of PSP is to allow multiple tuners as well as partial TV programs, as given by the following definition.

Definition 4 (Multi tuner partial program selection problem (MPPSP)). A set S of TV programs is given, and a start time bs and an end time es are given for each program s. Furthermore, for each program s, a value function vs is given and the number of tuners m is given. Issue, for each TV program s∈S, bs ≦ l1 <r1 < l2 <r2 <··· < by lk _s <rk _s ≦ es, the interval for viewing or recording a program (perhaps empty) set Is = {[l1, r1), ..., [lk s, is to determine the rk _s)], Thus, TV programs are received on the selected interval of m vertical tuner, i.e. all the time For x

であり、
（外６）

により与えられる選択されたインターバルの全体の値が最大にされる。

And
(Outside 6)

The overall value of the selected interval given by is maximized.

  Unfortunately, directly utilizing an expanded graph of a function of convex values with adjusted capacity as described above does not work. This is because the expanded graph may select several parts of the same TV program that are received in parallel. For example, for the example of FIG. 3 with three tuners, the resulting solution is to receive TV program I and partial TV programs I1 and I5, and for 19:00 to 19:30 TV programs Means that the part is received three times.

  For the function of linear values with penalties, the graph shown in FIG. 4 is applied. The change is that the capacity of the edge (ti, ti + 1) from 1 to m needs to be increased and m minimum cost flows from source to sink need to be found. The reason that this graph can be reused is that for each TV program and each interval, there is at most one edge covering it and therefore no part is received more than once.

  Referring to FIG. 7, store a given set of N ≧ 1 TV programs s, each having a start time bs and end time es, and a viewer-preferred function νs provided by the viewer or recommender. An apparatus 700 for selecting a set S (701.1) from a memory module 701 is shown. For a tuner, the value function is selected from the group consisting of a convex convex function and a linear function. The apparatus further comprises a processor module, which is a minimum cost network for selecting an optimal set S′⊆S that includes a partial TV program that is received and possibly recorded by a tuner with M ≧ 1. Runs the algorithm module or even has itself.

  Preferred embodiments include pre-rolls and post-rolls. There may be some inaccuracies in the start and end times of the program. Thus, if a program is selected for reception, in order to receive the program completely, the tuner needs to be tuned to the corresponding channel from time εc (bs) ≦ bs to time λc (es) ≧ es. is there. Here, εc (t) is a function that indicates how early to tune to channel c to receive content scheduled forward from time t, and λc (t) is scheduled until time t. Indicates how late to tune to channel c to receive content. Both Εc and λc are assumed to increase strictly.

  The quantity bs-εc (bs) is called the pre-roll of the program s and the quantity λc (es) = es is called the post-roll of the program s. The main impact of program pre-roll and post-roll on the program selection problem is when two consecutive programs are received on the same tuner and the programs need to be completely received if they are on different channels There is a time between the scheduled end time es of the first program and the scheduled start time bs of the second program. Such a problem does not exist for two consecutive programs of the same channel.

  In order to reflect the fact that a long interval is required to obtain the full value of program s in channel c, the function of the value given in equation (3) is redefined as:

この場合、ペナルティ関数は、以下のように再び定義される。

In this case, the penalty function is defined again as follows:

代替的に、ペナルティ関数は、以下のように再び定義される。

Alternatively, the penalty function is redefined as follows:

これら後者のペナルティ関数は、ペナルティが小さくなると、より多くの時間が終了ポイントを超えて含まれることを反映する。これらペナルティ関数は、凸型であり、式（４）の再び定義された値の関数はなお凸型である。したがって、理論１が適用され、番組の受信を開始又は停止する開始ポイントεc（ｂs）及び終了ポイントλc（ｅs）が考慮される。

These latter penalty functions reflect that as the penalty is reduced, more time is included beyond the end point. These penalty functions are convex, and the function of the redefined value in equation (4) is still convex. Therefore, Theory 1 is applied, and the start point εc (bs) and the end point λc (es) at which the program reception is started or stopped are considered.

  Definition 5 (Pre-roll and post-roll program selection problem (PSPR)). Given a set of channels C, each channel c has an early function εc and a slow function λc. Furthermore, given a set S of TV programs, each program s has a start time bs and an end time es, and a function vs (value redefined) given by equation (4). Finally, the number of tuners M ≧ 1 is given.

Issue, for each program S∈S, as _{εc s (bs) ≦ l1 <} r1 <l2 <r2 <... <lk s <rk s ≦ λc s (es), the interval for viewing or recording a program (perhaps Empty set) Is = {[l1, r1),..., [Lk _s , rk _s ]]. This eliminates the need to tune more than M channels simultaneously, ie for all times x

であり、且つ、
（外７）

により与えられる選択されたインターバルの全体の値が最大にされる。

And
(Outside 7)

The overall value of the selected interval given by is maximized.

  For PSPR, it is assumed that TV programs on a particular channel do not overlap in time, ie, for all s, tεS, s ≠ t, cs = ct, [bs, es] ∩ [bt, et] = Ο. It should be noted that the selected intervals of two such TV programs may overlap while being received by the same tuner. For example, for the example of FIG. 1, assuming a 5 minute pre-roll and post-roll, programs A and B are the same channel, so the interval [17:55, 19:35) is selected for program A and the interval [19:25, 20:05) is selected for program B.

  Considering this as a minimum cost flow problem, for pre-roll and post-roll it is trivial to build a corresponding graph where the cost is correctly modeled and the program is selected several times in the same time interval. Not a thing. Thus, the graph shown in FIG. 5 is constructed, which has the problem that not all possible solutions for PSPR are obtained, and similarly that the solution is only guaranteed to be nearly optimal ( This is further explained below). FIG. 5 shows a graph of the example of FIG. 1 for a 5 minute pre-roll and post-roll case showing only the display of the first channel and the central timeline. In the central timeline, the labels “b” and “e” indicate whether the time point represents the start time or end time of the TV program, where a further subscript indicates the respective TV program. Next, the following is defined.

全ての時間ポイントのセットは、

The set of all time points is

により与えられる。

Given by.

  In addition, t1 = 0, tn = ∞ and all i = 1,..., N−1 represent individual time points at T by t1,. The graph is a center consisting of a node for each time point ti, i = 1,..., N and an edge (ti, ti + 1), i = 1,. Including the timeline. The capacity of these edges is set to m and their cost is zero.

により与えられる。 Next, nodes and edges associated with the TV channel cεC are defined. As theory 1 shows, tuning to channel c begins at the time point when the program on TV channel c starts or at the point where the program on another TV channel ends. The time point of the former set is given by T _c ^b and the latter set is given by T _c ^e . Shakingly, tuning to channel c is stopped at the point in time when the program on channel c ends or the program on another channel starts. The former set is given by T _c ^e and the latter set is
(Outside 8)

Given by.

で開始されるとき、受信は、時間ε_c ^-1（ｔ）から前方にスケジュールされるＴＶチャネルｃのコンテンツについて確かであり、同調が時間
（外１０）

で停止されるとき、これは、受信は、時間λ_c ^-1（ｔ）までスケジュールされるＴＶチャネルｃのコンテンツについて確かであることを意味する。したがって、セット Tuning is time (outside 9)

Reception is certain for the content of TV channel c scheduled forward from time ε _c ^-1 (t), and tuning is time (outside 10)

This means that the reception is certain for the content of TV channel c scheduled until time λ _c ⁻¹ (t). Therefore, set

が導入され、この時間ポイントから及びこの時間ポイントまでコンテンツが受信されるのが確かであるブロードキャストスケジュールにおける時間ポイントを示す。ここで、増加する順序でｔc1,…,ｔcn_cによりＴcにおける個別の時間ポイントをナンバリングして、それぞれの時間ポイントｔci，ｉ＝１,…,ｎ_cについてノードがグラフに導入され、Ｔcにおける連続する時間ポイントのそれぞれのペアについて、キャパシティ１のエッジ（ｔci,ｔc,i+1）,ｉ＝１,…,ｎ_c−１が導入される。

Is introduced and indicates the time point in the broadcast schedule where it is certain that content will be received from and to this time point. Here, tc1 in increasing order, ..., and numbering the discrete time points in Tc by tcn _c, each time point tci, i = 1, ..., the n _c node is introduced in the graph, continuous in Tc For each pair of time points to be introduced, an edge (tci, tc, i + 1), i = 1,..., N _c −1 of capacity 1 is introduced.

からチャネルｃのタイムラインの時間ポイントε_c ^-1（ｔ）にエッジが描かれる。さらに、それぞれの時間ポイント
（外１２）

について、チャネルｃのタイムラインの時間ポイントλ_c ^-1（ｔ）から中央のタイムラインの時間ポイントｔにエッジが描かれる。 Next, an edge having all the capacity 1 is added between the central timeline and the timeline of the channel c. Therefore, each time point (outside 11) in the central timeline

To an edge at time point ε _c ^-1 (t) of the timeline of channel c. Furthermore, each time point (outside 12)

Is drawn from the time point λ _c ⁻¹ (t) of the timeline of channel c to the time point t of the central timeline.

Finally, costs are defined for edges in the timeline of channel c, and edges to and from the central timeline. For this reason, consider each TV program s on channel c, ie cs = c. Let tci and tcj be the start and end times of program s, ie tci = bs and tcj = es (note that these time points are indeed at Tc). The two cases can be distinguished as follows.
When j−i = 1, the cost −vs is given to the edge (tci, tcj).
When j-1> 1, it is as follows.

上記以外のエッジは、０のコストを得る。

Edges other than the above get a cost of zero.

For program A on channel c in FIG. 5, in this case bA = tc1 and eA = tc4, examples of five types of edges are given below.
(I) Edge of TV channel c (18:00, 18:50)
(Ii) Edge of TV channel c (19:10, 19:30)
(Iii) Edge of TV channel c (18:50, 19:10)
(Iv) Edge from the central timeline to the timeline of TV channel c (19:05, 19:10)
(V) Edge from the timeline of the TV channel c to the central timeline (18:50, 18:55).

  Here, the approach to resolving PSPR is to find the least cost flow of m from source to sink in the previously constructed graph. The time point at which the flow leaves the central timeline ti, i = 1, ..., n indicates when the tuning to the TV channel begins, and the time point at which the flow returns to the central timeline is the tuning to the TV channel. Indicates when will stop. Further, the edge of the timeline of the TV channel c according to one flow indicates a part of the received program.

  Solving the previous minimum cost flow problem provides a solution for PSPR, but there is one solution and not all solutions for PSPR are achieved. For example, if there are 3 minutes pre-roll and 5 minutes post-roll, the interval selected to tune to the corresponding TV channel is at least 8 minutes long. In other words, since a pre-roll and a post-roll are always obtained, a very small tuning interval cannot be achieved. In fact, this is a severe limitation. The algorithm is therefore optimal in the actual case. Furthermore, a given graph is refined with bits, which relaxes this constraint.

  Finally, if the partial program is not received and only the complete program is received, the constructed graph is significantly reduced. The set Tc then includes only the start time point bs and end time point es of the TV program on that TV channel, and only when j−i = 1 applies to the TV program. FIG. 6 shows the resulting graph for this example problem, where an edge with program label s has a cost -vs.

  Referring to FIG. 7, there is shown an apparatus 700 for selecting a set S (701.1) from a memory module that stores a given set of n ≧ 1 TV programs, each TV program having a start time. bs and end time es, and a function νs of values redefined so that a long interval is needed to obtain the entire TV program (part), the latter function being either by the viewer or recommender Provided. For one tuner, the value function is selected from the group consisting of a convex function with a penalty and a linear function. For tuners above 1, the value function is a linear function with a penalty. The apparatus executes a minimum cost network algorithm module that selects an optimal set S′⊆S that includes a partial TV program received by a tuner with M ≧ 1 and possibly having a recorded time margin, or Is further provided with a processor module.

  Referring now to FIG. 8, the system 800 uses the apparatus of FIG. 7 to select an optimal set of TV programs, including partial TV programs, for a given time interval and a given viewer preference. In the TV set, the incoming broadcast signal 704 is scanned for the TV program selected by the apparatus of FIG. 7 (700), and the selected partial TV program is recorded by the tuner (703) (801) and all are controlled / executed by the processor (702).

  Although the preferred embodiment of the present invention has been illustrated and described, it is intended to maximize viewer satisfaction expressed as a function of viewer preference values as described in this embodiment. The system, apparatus and method for selecting the best set of partial TV programs based on pre-roll and post-roll are exemplary and various modifications and variations can be made without departing from the true scope of the invention. It will be appreciated by those skilled in the art that changes are made and equivalent concepts are replaced for that element. In addition, many modifications may be made to adapt the teachings of the present invention to a particular situation without departing from the true scope of the invention, for example, various optimization algorithms. Accordingly, the invention is not limited to the specific embodiments disclosed as the best mode contemplated for carrying out the invention, and the invention includes all embodiments that fall within the scope of the claims. Is intended.

It is a figure which illustrates many TV programs provided through time t. It is a directed graph of the TV program of FIG. 2 is an expanded graph of the example of FIG. 1 including a partial TV program. 2 is an expanded graph of the example of FIG. 1 including a partial TV program having a linear value function with associated edge penalties and costs for TV program A. FIG. FIG. 2 is a graph of the example of FIG. 1 for a case with a 5-minute pre-roll and post-roll showing a display consisting of a first channel and a central timeline. 2 is a graph of the example of FIG. 1 for a case with pre-roll and post-roll when only a complete TV program is received. 1 is a diagram illustrating an apparatus for executing a method for optimal selection of a partial TV program according to the present invention. FIG. FIG. 1 illustrates a system for receiving, recording and displaying a partial TV program set selected in accordance with the present invention.

Claims (20)

A method for selecting a set of partial TV programs preferred by a viewer in a given period [b, e] starting at time b and ending at time e, comprising:
To change the start time and end time of the partial TV program received through the channel to include at least one of pre-roll and post-roll, the associated fast function εc and slow function λc are respectively Providing a set C of channels c possessed by:
Defining a set S of N ≧ 1 partial TV programs s, each TV program having a function νs of values received through the provided channel c, a disclosure time bs and an end time es;
Providing M ≧ 1 tuner to receive and record a partial TV program sεS;
Constructing a directed graph comprising a set V of nodes corresponding to the start time and end time and a set of edges with associated costs and capacities for each TV program s received through channel c;
Applying a least cost network flow algorithm to the directed graph, the function νs (l, r) of the value is maximized through the subset S ′ without conflict during the period [b, e], and the subset S Determining the subset S′⊆S such that a partial TV program at 'is received with M ≧ 1;
The directed graph is
A central timeline of start and end nodes respectively modified by the associated early and late functions to include pre-roll and post-roll;
For each channel c, a timeline of channel c of the start node and the end node for each TV program s received through the channel;
Each node of the central timeline having a corresponding node of at least one timeline of channel c according to a predetermined rule;
A method characterized by connecting. Each of the partial TV programs of set S received on a particular channel cεC does not overlap in time with other TV programs of the same channel c,
The method of claim 1. The applying step further comprises the step of calculating a linear sum of the value function νs through the subset S ′.
The method of claim 1. The defining step further comprises the step of a viewer or recommender defining the set S;
The method of claim 1. The value function is a linear function with a predetermined penalty for ending display early and starting display late.
The method of claim 1. The building step includes
Arrange the nodes and edges of the central timeline in order of increasing from the source node at time 0 to the sink node at time ∞ through the modified start time bs and end time es of the TV program sεS, respectively. And steps to
Arranging, for each channel c, the nodes and edges of the timeline of channel c in increasing order through the start time and end time of each TV program s received through channel c;
The method of claim 1 further comprising: Each edge of the central timeline has capacity M ≧ 1 and cost 0,
Each edge of the timeline of each channel has a cost determined according to capacity 1 and a first preset function, and the second node pair of the time point is the TV program sεS from that time. Is the time when the content of the TV program sεS is received through the channel c until that time,
Each edge connecting the central timeline and the timeline of at least one channel has a capacity 1, and the corresponding time points bs and es of the at least one channel timeline are a function of the speed. Connecting the node bs of the central timeline, which is respectively changed by the function of the slowness, and the node es, and the edge has a cost determined according to a second preset function;
The method of claim 6. The applying step further comprises the step of calculating a linear sum of νs through the subset S ′ of the function of values.
The method of claim 7. The defining step further comprises the viewer or recommender defining the set S;
The method of claim 8. The value function is a convex function having a predetermined penalty for ending display early and starting display late,
The method of claim 9. A TV program receiving and sending device preferred by the viewer during a given period [b, e] starting at time b and ending at time e,
A memory module comprising a set S of N ≧ 1 TV programs preferred by viewers numbered s = 1,..., Each TV program having a start time bs, an end time es and a function νs of values. When,
A tuner of M ≧ 1 that receives (and records) the set S of the TV program s;
The partial program model of the set S is formulated and stored in the memory, and a partial program model formulation module for generating a directed graph G of the partial TV program having pre-roll and post-roll is executed, and the memory A processor module that executes a minimum cost network algorithm module that selects a subset of the generated graph G of the partial TV program to be received and recorded
The selected subset of the partial TV program is received by multiple tuners without conflict in the period [b, e], and the value function is maximized through the subset.
A device characterized by that. A system for scheduling a TV viewing session of a partial TV program preferred by a viewer,
A TV set for watching TV programs;
Select and receive a TV program preferred by the viewer and display on the TV at least a portion of the TV program received during a predetermined time interval [b, e] having a start time b and an end time e. And the apparatus of claim 11,
The system characterized by having. A TV program receiving and sending device preferred by the viewer during a given period [b, e] starting at time b and ending at time e,
A memory module comprising data describing a set S of N ≧ 1 TV programs s preferred by the viewer, each TV program having a start time bs, an end time es and a function νs of values;
A set of tuners of M ≧ 1 for receiving and recording the set S of the TV programs s;
2. A processor module for performing the method of claim 1 using data describing a set S included in the memory module and the set of tuners;
A device characterized by comprising: A device for receiving and sending a partial TV program preferred by the viewer during a given period [b, e] starting at time b and ending at time e,
A memory module comprising data describing a set S of N ≧ 1 TV programs s preferred by the viewer, each TV program having a start time bs, an end time es and a function νs of values;
A set of tuners with M ≧ 1 that receives (and records) the set S of the partial TV programs s;
5. A processor module for performing the method of claim 4 using data describing the set S included in the memory module and the set of tuners;
A device characterized by comprising: A system for scheduling a TV viewing session of a partial program preferred by a viewer,
A TV set for watching TV programs;
Select and receive a partial TV program preferred by the viewer, and at least a portion of the TV program received during a predetermined time interval [b, e] having a start time b and an end time e 12. The device of claim 11 for display on a TV;
The system characterized by having. A system for scheduling a TV viewing session of a partial program preferred by a viewer,
A TV set for watching TV programs;
Select and receive a partial TV program preferred by the viewer, and at least a portion of the TV program received during a predetermined time interval [b, e] having a start time b and an end time e 14. The device of claim 13 for display on a TV;
The system characterized by having. A computer program stored in memory,
An executable module for performing the method of claim 1 and a data module comprising as input to the method of claim 2 a set C of channels and a set S of TV programs, the set S comprising: Including n ≧ 1 preferred partial TV programs s, each TV program having a start time bs, an end time es and a function νs of values,
A computer program characterized by the above. A computer program stored in memory,
An executable module for performing the method of claim 3 and a data module comprising as input to the method of claim 3 a set C of channels and a set S of TV programs, the set S comprising: Including n ≧ 1 preferred partial TV programs s, each TV program having a start time bs, an end time es and a function νs of values,
A computer program characterized by the above. A computer program stored in memory,
An executable module for performing the method of claim 9 and a data module comprising as input to the method of claim 4 a set C of channels and a set S of TV programs, the set S comprising: Including n ≧ 1 preferred partial TV programs s, each TV program having a start time bs, an end time es and a function νs of values,
A computer program characterized by the above. A computer program stored in memory,
An executable module for performing the method according to claim 9 and a data module comprising as input to the method according to claim 5 a set C of channels and a set S of TV programs, Including n ≧ 1 preferred partial TV programs s, each TV program having a start time bs, an end time es and a function νs of values,
A computer program characterized by the above.

Images