JP5116811B2 - Program recommendation device, method and program - Google Patents

Description

  The present invention relates to a program recommendation apparatus, method, and program, and more particularly, to a program recommendation apparatus, method, and program for viewing a television program of a user provided by a video distribution service using an Internet line.

  A number of information recommendation methods have been proposed. For example, in order to recommend a news article, the user is given a five-step evaluation for each news article, the past evaluation history is viewed, the degree of preference similarity between users is measured, and the news article preferred by a user with a similar preference Has been proposed (see, for example, Non-Patent Document 1).

  Similarly to the above-described method, the user is evaluated for each piece of information, and attention is paid to text included in the information evaluated by the user in the past evaluation history. There has been proposed a method for recommending information in which words similar to the text appear, considering that the text reflects the user's interest (for example, see Non-Patent Document 2).

  Also, let the user specify the text of the part of interest, estimate the user's interest based on the appearance frequency of the words contained therein, search the information in the community based on the word representing the user's interest, There is a method of presenting only information of a portion that is of interest to the user (for example, see Patent Document 1).

JP 11-338869 A

Resnick, P. et al., "GroupLens: An open architecture for collaborative filtering of netnews". CSCW '94: Proceedings of the 1994 ACM conference on Computer supported cooperative work, ACM Press New York, 1994. Mooney R.J. et al, "Content-Based Book Recommending Using Learning for Text Categorization", Proceedings of the 5th ACM Conference on Digital Libraries, ACM Press New York, 2000

  Multi-channel video distribution services using internet lines are progressing. Under these circumstances, it is difficult for the user to find a program that is really interesting. It is also difficult to grasp and record all the programs of interest.

  When a program is recommended using the technique disclosed in Patent Literature 1, a program related to an interesting word designated by the user is recommended. In this case, there is a problem that a word must be specified. Such a system that requires user interaction is inconvenient for the user.

  When recommending a program by the methods described in Non-Patent Documents 1 and 2, if the user who uses the user ID is truly single, the user's interest is estimated using the information of the program viewed by the user, and the program Can be recommended. However, in the program recommendation, the user ID is not always used by only one user. For example, consider a case where one user ID is used at home. If my mother likes dramas and her son likes anime, there is a problem that it is not possible to recommend drama or anime for such a family. Thus, an appropriate program cannot be recommended.

  The present invention has been made in view of the above points. In a video distribution service using an Internet line, a program is appropriately recommended for a plurality of users who use the same user ID according to viewing time. It is an object of the present invention to provide a program recommendation apparatus, method, and program that can be performed.

In order to solve the above problems, the present invention (Claim 1) is a program recommendation device that uses viewing time information in a video distribution service using an Internet line,
Program information acquisition means for acquiring program information in the video distribution service using the Internet line and storing the program information in the program information storage means;
Viewing information acquisition means for acquiring a viewing program and a viewing time of each user ID in the video distribution service using the Internet line, and storing the viewing information in the viewing information storage means;
A plurality of true users who use each user ID using the information on whether or not each user's program is viewed in the viewing information stored in the viewing information storage means and information on the viewing start time, and the plurality of users Parameter estimation means for calculating the true user's interest and the probability of viewing each program based on that interest,
Estimating the probability of viewing each program of each user ID by using the calculated true user, the interest of the true user, and the probability of viewing each program based on that interest, and recommending Recommended score calculation means for storing in the score storage means;
When a user identifier is given, a recommendation score stored in the recommendation score storage unit is acquired for each program currently being broadcast stored in the program information storage unit of the user, and the recommendation score is acquired. Program recommendation means for selecting identifiers of the top N programs as recommended programs.

Further, the present invention (Claim 2) provides the parameter estimation means,
When calculating the probability of viewing the program, an EM algorithm is used that estimates by maximizing the expected value of likelihood.

Further, the present invention (Claim 3) is a program recommendation method using viewing time information in a video distribution service using an Internet line,
In an apparatus having program information storage means, viewing information storage means, recommendation score storage means, program information acquisition means, viewing information acquisition means, parameter estimation means, recommendation score calculation means, program recommendation means,
The program information acquisition unit acquires program information in a video distribution service using the Internet line, and stores the program information in a program information storage unit;
The viewing information acquisition means acquires the viewing program and the viewing time of each user ID in the video distribution service using the Internet line, and stores the viewing information in the viewing information storage means;
The parameter estimation means uses the information on whether or not each user has watched each program in the viewing information stored in the viewing information storage means and the information on the viewing start time. A parameter estimation step for calculating a user, the interest of the plurality of true users, and a probability of viewing each program based on the interest;
The recommended score calculation means uses each user ID calculated in the parameter estimation step using each true user, each true user's interest, and the probability of viewing each program based on that interest. A recommended score calculating step of estimating the probability of viewing each program of the program and storing the program in the recommended score storage means;
When the program recommending unit is given a user identifier, a recommendation score stored in the recommended score storing unit is obtained for each program currently being broadcast stored in the program information storing unit of the user. And obtaining a program recommendation step of selecting identifiers of the top N programs with the recommended scores as recommended programs.

Further, according to the present invention (Claim 4), in the parameter estimation step,
When calculating the probability of viewing the program, an EM algorithm is used that estimates by maximizing the expected value of likelihood.

  The present invention (Claim 5) is a program recommendation program for causing a system to function as each means constituting the program recommendation apparatus according to Claim 1 or 2.

  As described above, the present invention specifies a plurality of users who use one user ID by using past viewing time information, and thus conforms to a true user, which is impossible with conventional technology. More accurate program recommendation is possible. This leads to an increase in viewing in video distribution services using the Internet line.

It is a block diagram of the program recommendation apparatus in one embodiment of this invention. It is an example of the program information table in one embodiment of this invention. It is an example of the viewing-and-listening information table in one embodiment of this invention. It is an example of the recommendation score table in one embodiment of this invention. It is a flowchart of the process at the time of the program information update in one embodiment of this invention. It is a flowchart of the process of the program dictation in one embodiment of this invention. It is a flowchart of the process at the time of the recommendation score update in one embodiment of this invention. It is a flowchart of the detailed process of step 340 of FIG. 7 in one embodiment of this invention. It is a flowchart of the process at the time of the program recommendation in one embodiment of this invention.

  An object of the present invention is to recommend a more appropriate program to each user even in a situation where a plurality of users use one user ID in a video distribution service using an Internet line. . Specifically, a method for estimating a true user from a user ID using information related to viewing time and recommending a program more suitable for the interest of the true user is realized. For example, when recommending a program to a certain family, it is considered that my mother is watching it during the daytime on weekdays, so my son is watching a drama program during the daytime on a weekday evening. So recommend an anime program.

  That is, the probability that “user u views program i at time t” is modeled as follows.

v はモデルによって得られる仮想ユーザで、平日の昼にドラマを見る「母」や、平日の夕方にアニメを見る「息子」を表す。z は番組のトピックで、「ドラマ」や、「アニメ」を表す。式（１）のようにモデル化する場合、v、zの数は予め設定しておくものとし、それらの値を V、Z とする。τ_uvは仮想ユーザの視聴時刻の平均であり、σ_uvは仮想ユーザの視聴時刻の標準偏差である。

v is a virtual user obtained by the model, and represents a “mother” watching a drama on weekday noon and a “son” watching animation on a weekday evening. z is the topic of the program and represents “drama” or “animation”. When modeling as shown in Equation (1), the numbers of v and z are set in advance, and these values are V and Z. τ _uv is an average of viewing times of virtual users, and σ _uv is a standard deviation of viewing times of virtual users.

If the probability that the user u views the program i at the time t is accurately known, it is possible to know which program the user views at the current time. For this purpose, the parameters P (v | u), σ _uv , τ _uv , P (i | z), and P (z | v, u) appearing in the equation (1) must be obtained correctly.

  Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of a program recommendation device according to an embodiment of the present invention.
The program recommendation apparatus 1 shown in FIG. 1 includes a program information processing unit 10, a viewing information processing unit 20, a recommendation score calculation unit 30, a program recommendation unit 40, a storage unit 50, a communication unit 60, and an input / output unit 70. The unit 60 is connected to the network 3, and the input / output unit 70 is connected to the external device 4 such as an input device or a display device.

  The recommendation score calculation unit 30 includes a parameter estimation unit 31 that estimates the parameter of the above formula (1).

  The storage unit 50 includes a program information table 51, a viewing information table 52, and a recommendation score table 53. Each table will be described below.

<Program information table 51>
As shown in FIG. 2, the program information table 51 includes a program ID field, a title field, a broadcast start time field, and a broadcast end time field.

  The program ID field is an identifier that identifies the program, and is set by the program information processing unit 10.

  In the title field, the title of the program is set as text data by the program information processing unit 10.

  In the broadcast start time field, the program information processing unit 10 sets the broadcast start time of the program.

  In the broadcast end time field, the program information processing unit 10 sets the broadcast end time of the program.

<Viewing Information Table 52>
As shown in FIG. 3, the viewing information table 52 includes a program ID field, a user ID field, and a viewing time field.

  The program ID field corresponds to an identifier of a program that appears in the program information table 51 and is set by the viewing information processing unit 20.

  In the user ID field, an identifier for identifying a user who has watched the program is set by the viewing information processing unit 20.

  In the viewing time field, the viewing information processing unit 20 sets the time when the user started viewing the program.

<Recommended score table 53>
As shown in FIG. 4, the recommendation score table 53 includes a user ID field, a program ID field, and a recommendation score field.

  In the recommendation score field, the recommendation score score (q, u) of the user u for the program q is set by the recommendation score calculation unit 30.

  The operation in the above configuration will be described below.

[Process when updating program information]
First, processing at the time of program information update in an embodiment of the present invention will be described in detail.

  FIG. 5 is a flowchart of the process when updating the program information according to the embodiment of the present invention. Assume that information about program p has been updated.

  Step 110) The program information processing unit 10 composes and inserts a line describing the information of the program p in the video distribution service acquired via the Internet line.

[Processing during program viewing]
FIG. 6 is a flowchart of processing when viewing a program according to an embodiment of the present invention.
Step 210) The viewing information processing unit 20 acquires the program viewed by the user, the viewed user, and the viewing start time from the external terminal 2 via the network 3, and the program ID field of the viewing information table 52, the user A row w in which values of information acquired in the ID field and the viewing time field are set is generated and inserted into the viewing information table 52.

[Process when updating recommended score]
FIG. 7 is a flowchart of a process at the time of updating a recommendation score according to an embodiment of the present invention.

  For example, the recommended score may be updated at the same time as the user views the program.

  Step 310) The recommendation score calculation unit 30 gives the number of virtual users V and the number of topics Z, which are hidden parameters for recommendation. This hidden parameter setting method is a constant parameter stored in advance in storage means such as a memory.

Step 320) The recommended score calculation unit 30 refers to the viewing information table 52, all the user sets U appearing in the viewing information table 52, the viewing program i _um of each user u∈U and the viewing time t _um (m = 1, ..., M). However, M represents the number of programs viewed by the user u and is different for each user.

  Step 330) The recommended score calculation unit 30 refers to the program information table 51 based on the program ID in the program ID field acquired in Step 320, and acquires all program sets I appearing in the program information table 51.

Step 340) The parameter estimation unit 31 calculates an EM algorithm (“EM algorithm” described in the literature (Bishop, MC, “Pattern recognition and machine learning”, 2006)) estimated by maximizing the expected value of likelihood. The parameters P (v | u), σ _uv , τ _uv , P (i | z), and P (z | v, u) are calculated for each user u∈U and each program i∈I.

Step 350) The recommended score calculation unit 30 refers to the program information table 51 and obtains a program set P that is currently being broadcast, that is, the current time falls between the value in the broadcast start time field and the broadcast end time field. To do.
Step 360) The parameters P (v | u), σ _uv , τ _uv , P (i | z), P (z | v, u) calculated by the recommended score calculation unit 30 in Step 340 and the equation (1) Is used to calculate the recommendation score score (p, u) = P (p, t _now | u) for each program p∈P of each user u∈U. However, t _now represents the current time.

Step 370) The recommended score calculation unit 30 sets the recommended score of the row where the value of the user ID field is u and the value of the program ID field is p in the recommended score table 53 for each user u∈U and each program p∈P. p, u) = P (p, t _now | u)

Next, the process of step 340 in FIG. 7 will be described in detail.
FIG. 8 is a flowchart of detailed processing of step 340 in FIG. 7 according to the embodiment of the present invention.

Step 341) The parameter estimation unit 31 is suitable for P (v | u) σ _uv , τ _uv , P (i | z), P (z | v, u) for each user u∈U and each program i∈I. Substitute an initial value. For example, the initial value is set as follows:
<Initial value of P (v | u)>
1. For each user u and each virtual user v

２．各ユーザ u 毎に制約条件

2. Restrictions for each user u

を満たすように

To meet

＜σ_uv（仮想ユーザの視聴時刻の標準偏差）の初期値＞
１．各ユーザ u∈U について、視聴時刻の平均を

<Initial value of σ _uv (standard deviation of viewing time of virtual user)>
1. For each user u∈U, the average viewing time is

とする。

And

  2. For each user u and each virtual user v

＜τ_uv（仮想ユーザの視聴時刻の平均）の初期値＞
各ユーザ u、各仮想ユーザ v について、

<Initial value of τ _uv (average of viewing times of virtual users)>
For each user u and each virtual user v

＜P(i|z) の初期値＞
１．各トピック z、各番組 i について

<Initial value of P (i | z)>
1. About each topic z and each program i

２．各トピック z 毎に制約条件

2. Restrictions for each topic z

を満たすように、

To meet

＜P(z|v,u) の初期値＞
１．各仮想ユーザ v、各ユーザ u、各トピック z について

<Initial value of P (z | v, u)>
1. For each virtual user v, each user u, each topic z

２．各仮想ユーザ v、各ユーザ u ごとに制約条件

2. Restrictions for each virtual user v and each user u

を満たすように

To meet

ただし、上記の初期値設定におけるrand(x,y) は x 以上 y 以下の乱数を表す。また、l の値は 0 で初期化する。

However, rand (x, y) in the above initial value setting represents a random number between x and y. The value of l is initialized with 0.

  The parameter estimation unit 31 repeats the following steps 342, 343, and 344 until the log likelihood l converges.

Step 342) The parameter estimation unit 31 determines the viewing program i _um of each user u∈U and the viewing time t _um (m = 1,..., M), each topic z, and each virtual user v.

を以下の式（２）を用いて計算する（Ｅステップ）。

Is calculated using the following equation (2) (E step).

ステップ３４３）パラメータ推定部３１が、各ユーザ u∈U の視聴番組 i_um とその視聴時刻 t_um (m=1,…,M) と、各トピック z、各仮想ユーザ v について、

Step 343) The parameter estimation unit 31 calculates the viewing program i _um of each user u∈U and the viewing time t _um (m = 1,..., M), each topic z, and each virtual user v.

を用いてパラメータ P(v|u)、σ_uv、τ_uv、P(i|z)、P(z|v,u) を以下の式（３）〜式（７）を用いて更新する（Ｍステップ）。

_Are used to update the parameters P (v | u), σ _uv , τ _uv , P (i | z), and P (z | v, u) using the following equations (3) to (7) ( M step).

ステップ３４４）パラメータ推定部３１が、対数尤度を以下の式（８）を用いて更新する。

Step 344) The parameter estimation unit 31 updates the log likelihood using the following equation (8).

［番組推薦時の処理］
図１２は、本発明の一実施の形態における番組推薦時の処理のフローチャートである。

[Program recommendation process]
FIG. 12 is a flowchart of processing at the time of program recommendation in one embodiment of the present invention.

  The program recommendation process is performed, for example, when the user turns on the TV or changes the channel.

  Step 410) The program information processing unit 10 refers to the program information table 71 based on the current time, and acquires the program set P currently being broadcast.

  Step 420) The program recommendation unit 40 refers to the recommendation score table 53 for each program p included in the program set P, obtains a recommendation score score (p, u) for the user p's program p, The program to be recommended to the user u is determined. At that time, an arbitrary number N (N = 1,2,3,…) preset in order from the largest recommended score score (p, u) may be set as a program to be recommended, or the recommended score score A certain threshold may be set for (p, u), and a program whose recommendation score exceeds that value may be recommended.

  It should be noted that the processing of each component of the program recommendation device can be constructed as a program and installed in a computer used as the program recommendation device for execution, or distributed via a network.

  In addition, the constructed program can be stored in a portable storage medium such as a hard disk, a flexible disk, or a CD-ROM, and can be installed or distributed in a computer.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made within the scope of the claims.

DESCRIPTION OF SYMBOLS 1 Program recommendation apparatus 2 External terminal 3 Network 4 External apparatus 10 Program information processing part 20 Viewing information processing part 30 Recommended score calculation part 31 Parameter estimation part 40 Program recommendation part 50 Storage part 51 Program information table 52 Viewing information table 53 Recommended score table 60 Communication unit 70 Input / output unit

Claims (5)

In a video distribution service using an Internet line, a program recommendation device that uses information on viewing time,
Program information acquisition means for acquiring program information in the video distribution service using the Internet line and storing the program information in the program information storage means;
Viewing information acquisition means for acquiring a viewing program and a viewing time of each user ID in the video distribution service using the Internet line, and storing the viewing information in the viewing information storage means;
A plurality of true users who use each user ID using the information on whether or not each user's program is viewed in the viewing information stored in the viewing information storage means and information on the viewing start time, and the plurality of users Parameter estimation means for calculating the true user's interest and the probability of viewing each program based on that interest,
Estimating the probability of viewing each program of each user ID by using the calculated true user, the interest of the true user, and the probability of viewing each program based on that interest, and recommending Recommended score calculation means for storing in the score storage means;
When a user identifier is given, a recommendation score stored in the recommendation score storage unit is acquired for each program currently being broadcast stored in the program information storage unit of the user, and the recommendation score is acquired. Program recommendation means for selecting identifiers of the top N programs as recommended programs;
A program recommendation device characterized by comprising: The parameter estimation means includes
The program recommendation device according to claim 1, wherein an EM algorithm that estimates by maximizing an expected value of likelihood is used when calculating a probability of viewing the program. In a video distribution service using an Internet line, a program recommendation method that uses information on viewing time,
In an apparatus having program information storage means, viewing information storage means, recommendation score storage means, program information acquisition means, viewing information acquisition means, parameter estimation means, recommendation score calculation means, program recommendation means,
The program information acquisition unit acquires program information in a video distribution service using the Internet line, and stores the program information in a program information storage unit;
The viewing information acquisition means acquires the viewing program and the viewing time of each user ID in the video distribution service using the Internet line, and stores the viewing information in the viewing information storage means;
The parameter estimation means uses the information on whether or not each user has watched each program in the viewing information stored in the viewing information storage means and the information on the viewing start time. A parameter estimation step for calculating a user, the interest of the plurality of true users, and a probability of viewing each program based on the interest;
The recommended score calculation means uses each user ID calculated in the parameter estimation step using each true user, each true user's interest, and the probability of viewing each program based on that interest. A recommended score calculating step of estimating the probability of viewing each program of the program and storing the program in the recommended score storage means;
When the program recommending unit is given a user identifier, a recommendation score stored in the recommended score storing unit is obtained for each program currently being broadcast stored in the program information storing unit of the user. A program recommendation step of acquiring and selecting an identifier of the top N programs of the recommendation score as a recommended program;
A program recommendation method characterized by: In the parameter estimation step,
4. The program recommendation method according to claim 3, wherein an EM algorithm is used for estimating the likelihood of viewing the program by maximizing an expected likelihood value.   The program recommendation program for functioning a system as each means which comprises the program recommendation apparatus of Claim 1 or 2.

Images