JP5591638B2 - Method and apparatus for delivery of targeted video programming - Google Patents

Description

The present invention relates generally to the delivery of targeted video programming to viewers, and more specifically, targeted or customized, apparently targeted to viewer characteristics. Relates to the determination of various viewer characteristics for linear delivery of video programming resulting in a linear television program.

Currently, it is common for individuals to record television programs for later viewing using commercially available video cassette recorders (VCRs). Typically, the VCR may be manually placed into a recording mode or programmed to record the selected program later. To program a VCR, the user enters the date, time and channel of the program he wishes to record, or enters the identification code of the desired program.
Increasingly, viewers of television programs have more choices about which programs to watch. For example, cable television has dramatically increased the number of channels available to viewers compared to channels available via conventional television antennas. Digital satellite systems offer more viewing options. Digital broadcasting of programs via cable television systems is expected to further increase the number of channels available to viewers.
One effect of the increased number of viewing options is the increased difficulty of deciding which program to watch. People, particularly those who are busy with schedules, may not have time to select and watch programs to decide which programs they like or dislike. Otherwise, a program that may be desirable for the viewer will not be viewed if the program is broadcast at a time that is inconvenient for the viewer. The user can select a program to watch and decide whether to like the program. However, in the situation where there are hundreds of program selections each week, this can take a significant amount of time and it is likely that some desired program will be missed.
There are companies that sample a population's television viewing habits by monitoring programs that a very small set of TV viewers watch. These companies also collect other demographic information about the people they monitor for sample generation. These samples provide valuable information about the population's television viewing habits that the samples cover. By regularly analyzing these samples, a general mathematical model can be formed that describes the behavioral patterns of the population during television viewing. Each individual viewer has a very personal preference for television programs, which can be quite different from the model derived from the general population sample behavior. At the same time, mathematical models derived by monitoring the behavior of a single user can be inaccurate due to the limited amount of information gathered by seeing only a single user in a short period of time. . Supporting the calculation of the mathematical model by sending the entire sample of the behavior pattern of the sample population to all viewer devices, the high bandwidth required to transmit this information to each device, and the viewer device Is counterproductive because of the processing power and memory requirements to process this information. Sending all users' personal viewing habits to the server and calculating a mathematical model for each user raises concerns about the viewer's privacy and also requires a return channel from the viewer device to the server .
With a mechanism that automatically determines the viewer's personal preferences accurately, a very personal TV viewing environment can be provided to the viewer. For households with a large number of members, by correctly identifying individual members and their preferences, the device can provide the most satisfying entertainment experience for individual viewers.
Methods have been developed for providing text data to viewers. The closed caption encoding method inserts closed caption text data into the vertical blanking interval of the video signal, thereby transmitting the text data in synchronization with the associated video data. However, closed caption text data must be inserted by the video program creator in the vertical blanking interval of the video signal. As a result, the headend operator cannot use the vertical blanking interval of the video signal to insert other text data such as sports, weather, stock market, news, advertisements and other data.
An electronic program guide (hereinafter “EPG”) gives viewers an on-screen list of upcoming TV programs on cable TV channels. An EPG data service provides EPG. The EPG data is converted into a video signal by the cable head. The EPG data is converted into a video signal by a cable head and transmitted to a viewer's television through a dedicated cable television channel. After tuning to a dedicated cable television channel, the viewer waits for the program to be displayed for a desired period of time. When using EPG data, cable head end operators often have to dedicate a separate cable television channel to EPG data and generate a video signal from EPG data provided by an EPG service provider.
One way to solve this problem is to modulate the EPG data into an FM carrier and transmit an FM carrier with a video signal on one of the cable television channels. A dedicated peripheral device for demodulating EPG data from the FM carrier is provided in the viewer's television tuner. The EPG data is then stored until the viewer requests that the EPG data be presented on the viewer's television. Once selected, the EPG data is then displayed on the viewer's television instead of other video programming.
A data controller for managing the flow of text data and electronic EPG data is disclosed in US Pat. No. 5,579,055.
Furthermore, preference agents for television programs have used the Bayesian method. See, for example, US Pat. No. 5,704,017 (hereinafter “the '017 patent”). However, in the '017 patent, a collaborative filtering system is used to predict a television viewer's desired preferences based on viewer attributes. A system that implements' 017 needs to communicate with a belief network via a two-way communication network, and the personal information of the viewer is disclosed to the network. '017 does not take advantage of the rich EPG information available for television programs that can be used to identify various features that contribute to viewer choice.

It is an object of the present invention to provide a method and apparatus for selecting and delivering targeted video data to a particular viewer, typically a television viewer. Video data may be targeted according to various viewer characteristics, including viewing characteristics, demographic characteristics, shopping characteristics, and others.

Accordingly, it is an object of the present invention to provide a method and apparatus for determining user preferences of television viewers. The present invention analyzes the demographic and other characteristics of the general population as well as sample viewing habits to explain the relationship between the audience viewing pattern of the general population and related characteristics, and for broadcast networks and the like. A method for generating parameters of a mathematical model that can communicate with a viewer device using a directional communication channel is defined. The present invention also defines a method for calculating user preferences of individual viewers using these parameters by the viewer device without leaking personal information about viewing habits to the outside.
The present invention relies on calculated viewer preferences to store and present broadcast content that may be of interest to one or more individual viewers who utilize the same device to view video data. The apparatus is also described in detail. Accordingly, the present invention also discloses a method for calculating the preferences of multiple viewers accessing the same viewing device, such as a multiple viewer household with a single television receiver, for example.
In another aspect, the present invention determines a user's viewing preference by monitoring a program viewed by the user and records a program corresponding to the user's preference. In accordance with the principles of the present invention, an apparatus for recording television programs includes a preference agent for retrieving attribute information corresponding to each television program viewed by a user of the apparatus. The preference agent generates classification information indicating the user's viewing preference as a function of the attribute information. The recording manager causes recording and storage of the television program having attribute information matching the classification information to the storage device.
In a further aspect, the present invention determines a user's viewing preferences by processing attribute information associated with the program that the user views and creating a profile of features that describe the viewer. Features include viewing preferences, demographic information (eg, age, gender, educational background, occupation, income, political and religious trends, marital status, sexual preference, racial background, home and work geographic location ), May include reading preferences, shopping preferences, music preferences, but is not limited thereto.
Embodiments using the principles of the present invention advantageously cause recording of programs that match a viewer's viewing habits. Thus, such an embodiment provides the viewer with a stored program that matches the user's viewing preferences that the viewer can view freely. Thus, the viewer is freed from the burden of deciding which program to watch from among hundreds of possible programs.
According to a further aspect of the invention, the program can be recorded for storage according to the available capacity of the storage device. Further, the program may be deleted in response to the selection by the user or based on the priority order indicated by the viewing preference of the user. Here, the program with the lowest priority is deleted first to make room for the newly recorded program. Program priority can again be a function of time. That is, more recently recorded programs are given higher priority.
According to a further aspect of the invention, determining which programs to record can also be a function of priority. That is, a program designated for recording is given the highest priority, followed by a program having attribute information corresponding to one or more user-specified criteria, followed by an attribute corresponding to recording preference information A program with information follows.
According to a further aspect of the invention, the user specification request has a first type request including information indicating a specific program and a specification indicating one or more programs having attribute information corresponding to the user specification. It may be in the form of a second type of request involving.
According to a further aspect of the invention, the user may record a currently broadcast program that the user is watching by generating a pause input. This advantageously allows the user to interrupt viewing of the currently broadcast program by recording the remainder of the program for subsequent viewing. Program viewing options will be given to the user in the form of menus that provide an easy-to-use interface for program selection and viewing and other options including play, pause, delete, fast forward, rewind, etc.
Preferably, the preference agent organizes the recording preference information in the form of a database organized according to the categorization parameters. Programs may be received in either analog or digital format. Programs stored in digital format are advantageously presented to the user in the form of additional channels. This allows the user to easily switch between programs (recorded or broadcast) simply by switching channels.

1 is a high level block diagram of a program selection device using the principles of the present invention. 1 is a high level block diagram of a system that employs the principles of the present invention. It is a figure which shows two examples of program information. It is a figure which shows the example of a spot color and a preference value. FIG. 6 is a flow chart illustrating data analysis performed on a representative sample. FIG. 6 is a flow chart illustrating data analysis performed on a representative sample. FIG. 5 is a diagram illustrating a process for calculating a user selection prediction error. It is a figure which shows the step in the process of regression analysis. FIG. 4 is a diagram illustrating the relationship between two correlated features. It is a figure which shows the process which determines the spot color characteristic of the spot color in a program. It is a figure which shows the process which determines the spot color characteristic of the spot color in a program. It is a figure which shows the process which determines the spot color characteristic of the spot color in a program. It is a block diagram which shows the example of a preference distribution recording format. FIG. 11 lists several sample values for the different fields described in FIG. It is a block diagram which shows the example of a special color recording format. It is a figure which shows the example for broadcasting special color as a part of EPG data. It is a figure which shows the example of a user selection recording format. It is a block diagram which shows the input and output for the production | generation of a user selection log | history. FIG. 6 is a flowchart illustrating a process of learning preference values performed on a program selection device. It is a block diagram which shows the input and calculation for calculation of the relevance value of selection history recording. It is a figure of the graph showing the relationship regarding age. It is a figure of the graph showing the repetition rate regarding a relationship. FIG. 6 is a block diagram illustrating inputs and outputs for a process of updating past selection history. FIG. 6 is a flowchart illustrating a process of updating a past selection history. FIG. 4 is a flowchart illustrating a process for calculating a preference value performed on a program selection device. It is a block diagram which shows the input and output for preference calculation with respect to a program. It is a figure which shows distribution of income with respect to a different program. It is a figure which shows distribution of sex with respect to a different program. 1 is a diagram of a system architecture for providing targeted advertisements. FIG. FIG. 7 is a graph showing the relationship between belief functions and probabilities for users belonging to a specific demographic feature value. It is a figure of the flowchart of a demographic feature recording format. It is a figure of the flowchart of advertisement target determination recording format. It is a block diagram which shows operation | movement of a certain function which the television recording system of FIG. 1 performs. It is a block diagram which shows operation | movement of a certain function which the television recording system of FIG. 1 performs. FIG. 2 is a diagram illustrating an alternative hardware configuration of a system that implements the principles of the present invention. FIG. 2 is a diagram illustrating an alternative hardware configuration of a system that implements the principles of the present invention. FIG. 2 is a flow chart illustrating a further aspect of the operation of the television recording system of FIG. 1 is a block diagram of one embodiment of a system for providing EPG data and text data to a viewer. FIG. FIG. 3 is a block diagram illustrating one embodiment of a data controller for receiving EPG data and text from a data provider, formatting the data for display, and inserting the data into a cable television channel vertical blanking interval. . FIG. 30 is a diagram illustrating one example of an information field of EPG data read from the EPG database of FIG. 29. FIG. 4 is a diagram illustrating a data format of data read from a database for insertion into an assigned cable television channel. FIG. 30 is a flowchart showing an operation of the EPG transaction formatter of FIG. 29. FIG. 30 is a flow chart illustrating one embodiment of the operation of the text transaction formatter of FIG. 29. FIG. 6 is a diagram illustrating one embodiment of a set top box for use in receiving text data and EPG data. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates various aspects of a process for creating and using multiple viewer profiles in accordance with the present invention. FIG. 6 illustrates a method for delivering targeted electronic content without compromising user privacy. 1 is a diagram showing an embodiment of a system according to the present invention. FIG. 4 is a diagram illustrating custom linear programming according to the present invention.

In FIG. 1, television control system 100 operates in accordance with the principles of the present invention to record a television program in response to user input 102 and television signal 104. The television control system 100 transmits a signal to the television monitor 108 for viewing by the user. Preferably, in the digital embodiment, programs that the system 100 records are presented to the user in the form of additional channels. Therefore, the user can quickly determine the stored programs available for viewing by changing the channel. The user can also change channels between stored programs or between a stored program and a program that is currently being broadcast. If the user changes the channel from a recorded program to another program, playback of the recorded program is preferably paused. Instead, whether the playback of the recorded program is paused or continued is an option that the user can select. As described further herein, a user may identify a program for recording by designation of a particular program or by designation of particular attributes of the program, such as a comedy / drama, actor. When manually specifying a program for recording, the user may specify that the program should be recorded once or repeatedly, such as weekly, when broadcast.
Signal 104 includes a first component 105 that includes information necessary to display the video and audio components of the television program on television monitor 108. The signal 104 preferably also includes a second component 107, referred to herein as “attribute information”. Examples of such attribute information 107 are from DVB-SI and ATSC-SI formats, StarSight Telecast, Inc. (Fremont, Calif.) And TV data (TVData) of Glenfold, NY. This is information that can be used through various unique formats such as Star Site EPG data and TVData.
The attribute information 107 for any particular program varies depending on the program type, but typically includes multiple such as program start time, program duration, program title, and other attributes (categories) of the program As well as associated values corresponding to each of the categories. The preference agent 110 processes the attribute information 107 and generates a “category-value” pair 115. For example, if the program attribute is duration, the category may be duration and the value of the category may be 120 minutes. If the attribute of the program is a title, the category may be a title and the value may be “Star Wars”. The other category-value pairs of a movie are: a description category with a short description of the movie is a value, a leading actor category with the name of the starring star of the movie is a value, a director category with the director's name It may include being a value, a theme category having a theme such as adventure, comedy, etc. being a value and an evaluation category having a rating by a particular critic being a value. A category-value pair of a sports game such as a football game may include the name of the team that is playing, the location of the game, and a specific tournament such as a playoff or super bowl.
The category-value pair 115 (preference information) indicates the viewing preference of the user. The data shown in FIG. 1 as associated with the category-value pair 115 includes not only weighting information for the associated category value, but also other information shown below for illustrative purposes. The preference agent 110 maintains preference information 115 in the form of a preference database 116. Television programs 105 recorded by system 100 are preferably stored separately with associated attribute information 107. In an alternative embodiment, category value pairs 115 (with or without associated values) are stored with television program 105, and raw attribute information 107 is not maintained by system 100.
In response to the user viewing habits, the preference agent 110 generates data for each category stored in the preference database 116 and for each value of each category. The data generated by the preference agent 110 for each category and value is preferably the time that the user views a particular category and / or value relative to the total amount of time that the particular category and / or value is available for viewing. The amount of The relative amount of time a user views a program is a convenient indicator of the user's relative viewing preferences. However, other indicators of user viewing preferences may be used. Program source switch 114 operates in response to user input 102 and selects either a currently broadcast program or a stored program from storage device 106 via television signal 104.
The recording manager 112 operates to cause recording and storage of the television program 105 and the attribute information 107 according to the information generated by the preference agent 110 and stored in the preference database 116. The recording manager 112 also responds to user requests to record specific programs and user requests to record programs with specified category-value pairs.
The signal transmitted to the monitor 108 preferably takes the form of a conventional analog. Alternatively, the signal transmitted to the monitor 108 may be digitally encoded. The exact form of the signal transmitted to the monitor is not critical and may take the form that a particular monitor requires. The television signal 104 received by the television control system 100 may take one of a variety of signal formats including analog encoded signals encoded according to well-known NTSC or PAL standards. Alternatively, the signal 104 may be digitally encoded as transmitted by a commercially available digital satellite system (DSS) or in accordance with the MPEG-2 (Motion Picture Expert Group-2) standard. In any given embodiment of television control system 100, signal 104 may take various forms as described above. For example, the television control system 100 may be coupled to receive input from a digital satellite system, and the input is digitally encoded. Television control system 100 may also be coupled to receive input from a joint antenna television system (CATV) that encodes signals in either analog or digital form. The television control system 100 may also be coupled to receive analog or digital signals from a conventional home antenna.
The attribute information 107 may be DVB-SI (Digital Video Broadcasting Service Information) or Advanced Television System Committee (ATSC) as defined by the European Telecommunications Standards Institute (ETS). It can be transmitted to the television control system 100 simultaneously with the television program 105 in various ways, including industry standards such as the ATSC digital television standard as defined. For illustration purposes, in the DVB-SI protocol, programming for the next 6 hours is transmitted every 8 seconds for each channel. As a further example, program information for the next seven days is available from an interactive on-screen TV program guide available from Starsight Telecast, Inc. Further programming information such as the next seven days may be obtained in other ways. For example, by receiving information in a time division multiplexed manner on a particular channel. Such information can be easily transmitted when the user performs an operation that does not require a moving image on the screen, such as when the user is displaying a control menu on the screen.
Alternatively, the television control system 100 can download the attribute information 107 separately from the television program 105 through a separate communication session via a modem or vertical blanking interval (VBI) included in the television signal. Such separate communication sessions include data download mechanisms supported by MPEG-2, DVB-SI and DSS protocols.
The attribute information 107 can take the form under the DVB-SI protocol as shown below.
event_id (event_id)
start_time, (start time)
duration (duration)
DESCRIPTOR1 (descriptor 1)
DESCRIPTOR2 (descriptor 2)
...
DESCRIPTORn (descriptor n)
The event-id field is a unique alphanumeric code assigned to the program. The descriptor may be a “Short Event Descriptors”, “Extended Event Descriptors” or “Content Descriptors” that includes the following information:
Short Event Descriptor (short event descriptor):
[
event_name-length (event name-length)
event_name (event name)
event_description-length (event description length)
event_description (event description)
]
Extended-even (equally extended)
[
ITEM1 (Item 1)
ITEM2 (Item 2)
...
...
ITEMn (item n).
]
content descriptor:
[
CONTENT1 (content 1)
CONTENT2 (content 2)
...
...
CONTENTn (content n).
]
The ITEM (item) includes the following information.
[
item_description_length (item description length)
item_description (item description)
item_value_length (item value length)
item_value (item value)
]
An example item description may be “director” and the item value may be “Martin Scorsese”. CONTENT (content) includes the following information.
[
DVB-SI defined theme (DVB-SI prescribed theme)
DVB-SI defined sub-theme, programmer defined theme, programmer defined subtheme (DVB-SI defined sub theme, programmer defined theme, programmer defined sub theme)
]
Examples of themes and sub-themes are movies and comedies, respectively. Programmer defined themes and sub-themes are values that an EPG data provider can provide.
The category-value pair 115 is generated from the above type of information. The category-value pair 115 takes the following format: That is, category name-category value, where the category name is "title", "director", "theme", "program type", etc., and the category value is "next sign" (Seinfeld), It can be “Martin Scorsese”, “Comedy”, “Sitcom”, etc. Generation of the category-value pair 115 from the attribute information 107 enables the preference agent 110 to generate a category that does not clearly exist in the attribute information 107. For example, the category-value pad 115 may be title-fortainers, description-football and description search rules-football (and) San Francisco. Therefore, the preference agent 110 can generate the category-value pair 115 from the attribute information 107 even when there is no field in the attribute information corresponding to the created category-value pair.
The preference database 116 is preferably generated by first downloading category-value pairs from a third party source such as Starsite, Telecast, Inc. Advantageously, such a source may provide customized information for a particular geographic area and date. For example, the database may include data that gives sports events that include local teams a higher rating than other sports events. In addition, seasonal or vacation programs may be indicated as being preferred during a particular season or vacation. For example, a program that includes summer activities is shown as having a higher weight during the summer than at other times of the year. The preference database is changed as described in this specification according to the viewing habits of the user. Furthermore, the preference database can be periodically updated from third party sources to reflect the aforementioned seasonal or holiday updates.
Categories in the preference database 116, such as those received from third party sources, are either pre-defined or dynamically created from attribute information 107 received for the program 105. Dynamically created categories and associated values are preferably given a default rating by the preference database 116. Examples of preference information created by the preference agent 110 or downloaded to the preference agent 110 are shown below. In the following example, the numbers in the three columns of the category statistics and value statistics portion are the weighted duration watched (in seconds) (ranging from 0 to 1000) and the programs that match that particular category or value. Indicates the amount of time (in seconds) that was available. Information is preferably stored in the form of database records.
category:
Channel 1000
Title 1001
Title-Substring 1002
Genre information 1003
Description 1004
Description substring 1005
Episode name 1007
Type 1008
Star 1009
Director 1010
Production year 1011
MPAA rating 1012
Critic Evaluation 1013
value:
Titanic 2000
Ami 2001
3rd Rock From the Sun 2002
The Gods Must Be Crazy 2003
Next Seinfeld 2004
Headline News 2005
Bugs and Duffy 2006
News 2007
004 2008
005 2009
063 2010
49ers 2011
Sitcom 2012
Comedy 2013
Movie 2014
News 2015
San Francisco 49ers 2016
Bushman (A Coke bottle raises havoc for a tribe of African bushmen) 2017
John Mayers 2018
Lucy Barnett 2019
Marius Weyers 2020
Sandra Prinsloo 2021
Jeff Bridges 2022
Valerie Perrine 2023
Phil Hartman 2024
Jamie Uys 2025
Lamont Johnson 2026
1981 2027
1973 2028
1996 2029
Three Star 2030
Two Stars 2031
Nude 2032
Violence 2033
Adult thing 2034
Adult language 2035
Category value pair:
1001 2001
1001 2002
1001 2003
1001 2004
1001 2005
1001 2008
1000 2009
1000 2010
1002 2011
1003 2012
1003 2013
1003 2014
Category statistics:
1001 1000 31104 4022280
1002 1000 31104 4022280
1003 1000 31104 2613384
1004 1000 20304 1996596
1005 1000 20304 1996596
1006 1000 5238 1259028
1007 1000 3438 369450
1008 1000 13266 812970
Value statistics
2001 1000 1638 88074
2002 1000 6714 178560
2003 1000 6552 387054
2004 1000 5400 165600
2005 1000 1600 9000
2006 1000 3600 28800
2011 500 1800 10800
In the above example, 14 categories are given (1000-1013), followed by 36 values. Next, the correspondence (category-value pairs) between categories and values is shown. Next, the category and then value data are shown. This data is organized in three columns as described above.
In the above-described embodiment shown in FIG. 1, category and value ratings are dynamically generated by preference agent 110 instead of being stored in preference database 116. In an alternative embodiment, the ratings may be stored in a preference database along with category-value pairs.
Here, referring to FIG. 2, preference determination is used to predict a user's preference when selecting a TV program. This prediction is based on (i) analysis of individual user viewing habits, (ii) analysis of viewing habits of representative samples of users and (iii) television program EPG data available for the duration of the collected samples. Based.
The user's viewing habits are described below in the context of the actual program viewed by the user. However, other information may be just useful for the purposes of the present invention, which means that the user will not be able to watch or record a program that they will only watch before switching to another channel. The program that the user schedules, the program that the user records but does not see or only sees, and that the user requests information from the program guide (such as an on-screen EPG menu) or that the user requests information Contains information about programs that are not or that the user requests information but does not choose to watch.
One method for creating preference decisions is through the use of Bayesian networks used to predict user behavior, and hence the use of preference agents, which is a knowledge-based strategy. In this method, a knowledge engineer interviews an expert about his / her expertise. Experts and knowledge engineers determine world distinctions that are important for decisions in the field of experts. Knowledge engineers and experts then define dependencies between variables and probability distributions that quantify the strength of the dependencies.
The second method is a Bayesian network, which is a data-based policy. In this second method, knowledge engineers and experts first determine domain variables. Data is then accumulated for those variables and an algorithm is applied to create a belief network and predict user behavior from this data. This accumulated data comes from real-world examples of domains. This second strategy exists for domains that contain only discrete variables.
The Bayesian method can be performed using an algorithm known as the “EM” algorithm, which, as will be appreciated by those skilled in the art, calculates the maximum posterior probability value (MAP value) of a model parameter to determine user behavior. Predict. The EM algorithm is based on Dempster's “Maximum Likelihood From Incomplete Data Through the EM Algorithm” (“Maximum Likelihood From Complete Data The EM Alloy 39 )) And incorporated herein by reference.
After computing the probabilities for each variable, each variable in the belief network is then scored for goodness to adapt to predict user preferences by giving a subscore. As will be appreciated by those skilled in the art, the subscore is generated by using the marginal likelihood of the expected data at the maximum posterior probability value.
Those of ordinary skill in the art will recognize, “Bayesian Updating in Recursive Graphical Cosb” by Jensen, Lauritzen and Olesen, “Bayesian Updating in Recursive Graphical Co.” , Technical Report R-89-15, Institute of Electronic Systems, Aalburg University, Denmark), this step can be performed using any standard model that predicts user behavior estimation algorithms. Will.
The television viewing habits of the population are sampled to generate a representative sample of viewer behavior 117. This task is typically performed by companies such as Nielsen Media Research, which collects user behavior samples and conducts market research. The sampling result is stored in the viewer behavior database. The format of the viewer behavior database is owned by the sampling company. The viewer behavior database includes demographic information about people in the household participating in the sampling. This information includes, but is not limited to, race, age, annual income and gender. The viewer behavior database also includes information about all television programs that each viewer watched during the sampling period.
Program information and television program schedules are available 104 in the EPG database. Program information for a television program includes information about various attributes of the program, including but not limited to title, program type and program category of the television program and actors appearing on the television program.
FIG. 3 lists two examples of program information. Examples 1 and 124 describe program information for an audiovisual television program, and Examples 2 and 125 describe program information for a television program based on graphic text.
The viewer behavior database is analyzed 118 to identify TV program features and how such identified features affect viewing habits in the user's representative sample. These features and their impact on the viewing population are then used in supporting the process of predicting individual user TV program selection. Figures 4 and 126 list some examples of such features. Some of these features are derived from program information 104. Some of these traits look for user preferences for traits associations or other traits associations with other viewing parameters including but not limited to time, day of week, holidays, work days and weather, season Is derived by Some of these features are derived by data analysis of the viewer behavior database. Any given program exhibits a different degree of the identified feature. This is expressed as the trait-ness of the spot color in the program. For example, the comedy trait in Seinfeld can be 1.2, and the comedy trait in Mad About You. 79. The spot color of the spot color in the program is calculated 118 as part of the data analysis.
The present invention also provides a method for determining viewer preferences. In one embodiment, an individual viewer selects one show to view from the set of available programs by evaluating the probability preference function for each program and selecting the program with the highest score. . The preference function is modeled as a collection of preferences for a particular spot color and the degree to which the program exhibits that spot color. The preference function can be calculated as follows:
l (p) = Σλ (tn) ^* tn
Where l (p) represents the preference for program p,
λ (tn) represents the preference for a particular spot color tn
tn represents the degree of the spot color tn indicated by p.
When a viewer watches television, it is assumed that the viewer selects the program with the highest value l (p). For any given viewer, as the value of λ (tn) better represents its actual preference, the accuracy of determining that preference for a TV program becomes higher. Thus, for any given user, if all relevant features are known along with the exact preference of each of those features, the preference function will always accurately predict the preference of any given program. Will.
However, identifying all relevant features and determining the exact preference of each viewer is a non-trivial task and is repeatedly reached by the process of regression analysis.
Although most of the traits associated with a particular user can be derived directly from the EPG information itself, some traits are found by analyzing the user's viewing habits over time. Features that are suitable for discovery by the process of regression analysis are typically hidden or related features.
Hidden features affect the user's viewing habits, but cannot be derived from EPG information. For example, sitcoms that emphasize a particular racial background resonate more with viewers of that racial background. Another example of the influence of hidden features would be the strong preference of Sitcom “Frasier” among a set of users who have a strong preference for Sitcom “Cheers”. Assuming that the names of actors appearing in Sitcom are not available in the EPG data, this familiarity to both Sitcoms by the same set of people is due to the presence of several features that both Sitcoms share in common: It can be explained by the presence of one of the central characters in sitcom. Such hidden preference criteria need to be captured by adding hidden features in the preference function calculation.
A related feature is a feature that, when combined with other features, has a different impact on the user's viewing habits. For example, a user will have a preference for any given Seinfeld episode and a preference for any first sitcom broadcast for the first time. However, its preference for the first episode of Seinfeld will be large enough to require an additional feature, namely "New Seinfeld", and will fully explain its preference for the first episode of Seinfeld.
The preference of each trait for a given user must be refined as well from the initial approximation by regression analysis. Examples of sample user features and related preferences include, but are not limited to, those listed in FIG. The spot color “NBC <=> News” is an example of a related spot whose associated spot colors are “Channel NBC” and “Program Type News”. The user's preference for news programs may be 1, and the preference for “NBC channel” may be 2, while that preference for NBC news would be 13. That is, the user does not always watch a news program or generally a program on the NBC channel, but the person has a strong preference for the news program on the NBC channel. Considering only its preference for “Channel NBC” and “Program Type News”, its preference for “NBC News” cannot be explained.
This procedure of identifying a spot color is first performed using a determination of the favorite distribution of each spot color within the representative sample along with the viewing habits of the representative sample.
Some of the outputs of the representative sample viewing habit analysis 102 are a set of preference determination parameters. That is, (i) a spot color indicated by a repeated program and a degree indicating such a spot color, and (ii) a favorite distribution list of each of the spot colors in the viewing population.
One method for deriving these parameters is shown schematically in FIGS. 5 (a) and 5 (b). The different stages of the flowchart are further described in the following sections. An initial set of preference values is calculated 129 for each user. This initial preference value may be reached in various ways. One aspect would be to base it on the amount of time that saw a given feature versus the amount of time that was available during the user's viewing period.
Using these initial preference values for the traits, a preference function l (p) is calculated for all programs that were available for the time the user viewed a given program. Next, these programs are arranged in descending order 140 of preference values for the period 141 during which the viewer views the program. The actual program that the user watched at a given time is highlighted 143. If all highlighted programs in FIG. 6 had the highest score, the preference function correctly identified the user's preference. If the program viewed by the user ranks below any other competing program available at that time, the preference value used does not correctly reflect the user preference. For purposes of illustration and not limitation, if there are N programs ranked above the actually viewed program, the preference function error for this program will be referred to as N 142. Such a value of N is determined for each program viewed by the user by comparing it to the preference functions of other competing programs. In the example illustrated in FIG. 6, the value of N is the time when the user watches the TV during the time of 10: 00-10: 30, 10: 30-11: 0, 11: 00-11: 30. Is calculated against. Using regression analysis, each preference value for a spot color is adjusted incrementally to reduce the average value of N. The set of preference values that yields the lowest average value of N is considered the best set of preference values for that user.
One method of the regression process is illustrated in FIG. At the beginning of the process, the preference values for the special colors a, b, c are λa1, λb1, λc1,... (144, 145). λa1 represents the preference for the spot color “a”, λb1 represents the preference for the spot color “b”, and so on. The preference function for all related programs is calculated using this initial set of preference values. The preference value λa1 of the first spot color “a” is changed incrementally to determine the average value of N. The values λa2, 147 that yield the lowest average value of N are taken as new preference values for the spot color “a”. This new preference value λa2, 148 is used here to replace the old value and determine the remaining λ value. The above set of computed preference values will predict user preferences with high accuracy if the set of traits considered in the preference function includes all the traits that are relevant to the user. An average value of N that does not converge (to 0 or some other acceptable value) indicates that not all features associated with the user have been considered in the calculation of the preference function. Improve user preference determination with the introduction of related or hidden additional features.
The determination of relevant features can be accomplished in a variety of different ways.
One method is that the related values of a number of features are not reflected in the program information obtained from the EPG, but by application of a heuristic heuristic about human viewing habits. For example, a user with a preference for Seinfeld is very likely to have a much higher preference for the first or new episode of Seinfeld. Such heuristic heuristics regarding the associated values of multiple features can be communicated to each set top box via the headend. Another way to determine related features is to use an algorithmic search that seeks common features in the program and introduces new related features to improve the user preference function.
The determination of hidden features in different programs can be accomplished in a number of ways. One of these methods is illustrated in FIG. The possibility of the presence of a hidden feature is always present when there is a strong correlation between any two features that exist in different programs. The y-axis represents increasing preference values for higher values of y. Every point on the x-axis of FIG. 8 represents a user arranged such that the preference value for the spot color “tn” increases with higher values of x. If the preference value of the strongly correlated spot color “tm” is also plotted for the same value on the x-axis, it will show the relationship of the spot color “tn” to the preference graph. The correlation between any two features can be explained by the presence of a common hidden feature. Thus, the traits “tm” and “tn” can be expressed mathematically as follows:
tm = tx + tm ′
tn = Ctx + tn ′
Where C is a constant indicating the amount of correlation between the spot colors “tm” and “tn”.
Hidden features can also be identified by applying heuristics or some other suitable aspect.
While all programs of a particular genre show some common features, the degree to which these features show varies from program to program. This degree of trait for a repetitive program can be quantified, which best explains the viewing choice of viewing or not viewing the program in a representative sample. For example, consider a user who has a preference for comedy. The user's decision to watch a comedy program is influenced by whether the amount of comedy likeness that the program exhibits crosses a comedy-like threshold set by its preference for comedy.
One method for determining the spot color of spot color T in a given program N is illustrated in FIGS. 9 (a)-(c). The highlighted program represents the program 156 that the user has actually seen. All available programs are arranged in descending order of rank 154 as computed by the preference function. In the case of user 1 155, the program N 156 that the person has seen appears in the third rank, but the preference function should actually rank program N at the top. In this case, the error margin 157 in the spot color characteristic of the spot color T is 3. In the case of the user 4, the program N that the user watched was ranked highest. The error margin in the spot color characteristic of the spot color T is zero. For user 2 who did not watch program N, it ranked highest. The error margin in this case can be considered to be a constant K. The proposed value of K may then be the number of programs available to the user. Similarly, for user 3, the error margin can be considered as K-1. Such an error margin is calculated for all users who watched program N, and an average error margin is calculated.
Using the regression analysis, the spot color value of spot color T is adjusted incrementally to reduce the average value of the error margin. The spot color value that produces the lowest average error margin is considered to be the best spot color value of spot color T shown by program N.
The feature value may be assigned by a rule of thumb or some other suitable manner.
One output of a representative sample viewing habit analysis is a list of preference distributions among the viewing populations of each feature. This information is shown schematically in FIG. 10, where every individual user set-top box is made available with the broadcast TV program and program information. “Number of spot colors” 164 represents the total number of spot colors identified after analyzing the viewing habits of a representative specimen. The distribution of preference for each spot color is given in the “spot color record” 165. Information included in the spot color record includes a spot color name 167, a spot color type 168 (hidden or related, etc.), a spot color preference distribution 170, and a spot color distribution parameter 172. Examples of various possible values during spot color recording are given in FIG.
FIG. 12 shows a schematic representation of one of the traits present in a repeating program. This information is conveyed along with the preference distribution of the spot color. “Number of programs” 173 represents the number of programs to which the spot color information is sent. Related features and the degree to which they are shown are included in “Program Record” 174. The “program identification” 177 information is used to identify the TV program to which this recording belongs. An example of a program recording for the Seinfeld episode is given in FIG. As given in the example, certain traits such as “program name-Seinfeld”, “channel-NBC”, “program type-sitcom” can be derived directly from the EPG, which is called a static trait. Additional features such as “comedy” that store the degree of comedy likeness associated with the Seinfeld program may also be conveyed as part of the EPG data.
Referring again to FIG. 2, program information and program content (in the form of an EPG) are transmitted from the broadcast headend along with preference determination parameters 119. Examples of preference determination parameters include (i) a spot color indicated by each program and a degree indicating such a spot color, and (ii) a distribution list of preference values among the viewing population of each spot color. It is not something that can be done.
This information is received in each household by the program selection device 100 including the preference determination module, the personal preference database 116, the storage device, and the display device 108. The personal preference database 116 is used to store the results of analyzing the viewing habits of individual users. The storage device stores the program according to the program recommended by the preference determination module or according to the user's specific request.
The program selection device 100 monitors the selection of TV programs to be watched and the viewing behavior of each user. This is stored in the storage device or memory in the form of selection history data 189 (see FIG. 15). A schematic display of the selection history data is given in FIG. “Number of selection records” 180 represents the number of user selection choices stored in the selection history data. Each selection record 181 includes information 186 on the competing program available at that time, along with information 185 on the actual program viewed. Since the EPG may not be able to provide information about past programs, it is necessary to store program information. Information about these programs may be obtained directly from the EPG data when information is obtained while the program is still current. The time 183 and duration 182 of watching the program also form part of the selection record. As shown in FIG. 15, the user selection history 189 is derived from each selection 187 made by the user together with program information from the EPG 104.
One process for learning a particular individual's user preferences is shown in FIG. The preference function uses the preference distribution in the representative sample for the spot color identified by the data analysis 118 to minimize the impact of errors introduced due to insufficient sampling in the calculation of the individual spot color preference preference. Turn into. An individual user's response to the configuration question 190 may also be considered as a factor in determining an initial preference value for that individual 192.
User selection history 189 is maintained for a fixed number of hours. The number of times that the user selection history is maintained can preferably be varied to have an increased learning rate during the first few days of a new user using the device. The average error is calculated as N 194 for the users in the representative sample in the same manner as previously described in FIG. If the average error is greater than the acceptable limit 195, a new preference value is calculated 198. The entry in the user selection history is moved to the past selection history. If the average error stays below an acceptable limit, the preference value is calculated 196 only after a pre-defined number of hours.
Another embodiment of the selection record stores the program ID instead of the entire program information. In this embodiment, actual program information is broadcast at a predetermined time from a head end that identifies each program with the same program ID stored in the selection history. In this embodiment, the preference value calculation 198 for the spot color is performed only at this predetermined time.
Here, referring to FIG. 17A, after each calculation of the preference value, the entry in the user selection history is moved to the past selection history. The selection history is divided into a user selection history 189 and a past selection history 216 and optimally stores the section history record. The selection history contains information about all selections made by the user between the two calculations of preference values. The past selection history maintains the most relevant information about the most relevant choices the user has made in the past. The number of records in the past selection history is advantageously configurable to suit the memory available for storing such records. Relevance in this context is the importance of the record in calculating the preference value, which includes the age of the record 200 and the repetition rate 203 of the features included in the program 204 being selected and recorded. Depends on many parameters, not limited to. The relationship between the relevance of the record and the age 205 is shown in FIG. The most relevant record is the most recent record. The relationship between the repetition rate of the features included in the program being recorded in section and the relevance is shown in FIG. As the repetition rate decreases, relevance increases to a certain limit 206, and then relevance decreases 207. A good example of the reasoning behind this relationship is the relevance of keeping a record on user choice when the weekly program Seinfeld is available and the Olympic program every four years. At some point, it would be relevant to keep a record on user selection. While it is important to keep a record of Seinfeld and calculate the preferences of the various traits included in Seinfeld, it is pointless to keep a record of viewer preferences for the previous Olympics. Consider a user who has no preference for movies except Friday. In order to accurately determine the user's preference for the “Friday Night” feature of a movie, a selection record belonging to that viewing behavior on Friday must be maintained for at least several weeks. However, if the nature of the iterations is too far apart in time, the relevance is greatly reduced. The “spot color repetition rate versus relevance” graph in FIG. 17B measures increasing relevance values for higher values on the y-axis. The x-axis measures the decreasing repeatability rate of the feature in the program. As shown in FIG. 17 (a), when a certain threshold of repetition rate is exceeded, the relevance greatly decreases.
The process of updating the past history is shown in FIGS. 18 (a) and 18 (b). The process is performed for all records in the user selection history. If there is enough memory in the past selection history 218 to store the new record, the record is removed from the user selection history and added to the past selection history 222. If there is not enough space in the past selection history 218, the records in the past selection history are sorted based on relevance 220, the least relevant records are deleted 221, and space for adding new entries Make 222.
In another embodiment of the present invention, the number of available programs 186 stored in the past selection history can be limited to an optimal number to make maximum use of available memory (see FIG. 18). ). In this embodiment, only programs with higher preference values that represent traits with meaningful preference values are stored in the past selection history record.
Calculation of preference values for different spot colors performed by the program selection device 100 is shown in FIG. This process is performed 198 during the learning process, as illustrated in FIG. By using the current preference value as a starting point, regression analysis is performed so as to minimize errors in prediction of viewer behavior stored in the user selection history and the past selection history. The regression analysis process 223 is the same as that performed in the representative sample 131. If there is any selection record that has an error greater than a predefined threshold 224, the selection record is examined for the presence of a possible associated feature 225. Rules that define the existence of possible related features may be heuristics. A possible association of a spot color can also be found by looking up a list of related spot colors found from a representative specimen. The preference for possible relevant features is calculated by regression analysis 226. If the calculated preference is a meaningful value 231, the preference for the spot color is entered 230 into the preference database 116. Other parameters such as repetition rate for the spot color are also entered into the database 230. The repetition rate is calculated by looking at the repetition rate of the feature in the user history and past history. If there is no past selection record with this feature, the repetition rate is assumed to be a predefined value. When a new selection record is created, the repetition rate is updated. At the end of the process of calculating the preference value (illustrated in FIG. 9), a weighted average of the current preference value and the old preference value is calculated. This forms a new set of preference values. The learning speed can be increased or decreased by changing the weight of the current preference value.
A flow chart for calculating a score for a program for future prediction is illustrated in FIG. As described above, the preference 236 for the program 232 is the function l (p) = Σλ (tn). ^* Calculated by evaluating tn. λ (tn) is calculated as a weighted average of the preference for spot color tn for user 234 and the preference for spot color tn for general population 235.
In another embodiment of the present invention adapted to function on a program selection device operating in a household with a large number of users, the viewer enters the number of viewers in the household during the initialization process. Be asked to. Each viewer can answer a set of set questions that capture their preferences for representative features. An initial set of possible preference values is created for each viewer. After correctly identifying each viewer's session, the learning process updates each viewer's preference database separately. Viewer identification is done either by clear input from the viewer, identifying the viewer, or by automatic viewer identification by monitoring the viewer's television viewing behavior. In order to automatically identify the viewer, an error value between predicted and actual is calculated for all viewers. The viewer with the preference value that yields the smallest error value is selected as a possible viewer. The certainty of this decision is expressed as a probability calculated as a function of the difference between error values for different viewers in the household.
An object of the present invention is to determine a user's demographic characteristics by analyzing the user's viewing habits and, in parallel, the viewing habits of a representative sample of a user whose demographic characteristics are known. These demographic features of the user collectively constitute a demographic profile. Once a user's accurate demographic profile is successfully created, the present invention can receive a set of possible advertisements and show individual users only those targeted to a matching profile.
Some examples of demographic characteristics include, but are not limited to, the user's gender, race, age and income. The output of the representative sample viewing habit analysis provides a basis for determining the demographic characteristics of individual users.
The TV program viewing preference information selected in the viewing habit analysis of representative samples plays an important role in determining demographic characteristics. Different TV programs may be required to determine demographic characteristics. Typically, TV programs where the majority of viewers share common demographic features at a different rate than those observed in the general population are most suitable for the discovery of demographic features. For example, to determine the “income” demographic feature, consider graph (i) of FIG. The x-axis represents annual income in the order of increasing x. For any point on the x-axis, the y-axis represents the probability of finding the annual income of x in a representative sample (which is the same in the general population). Consider graph (ii) here, where the dotted curve plots the same income distribution as represented in graph (i), and the first solid curve (leftmost curve) is representative. The income probability distribution of the viewer of the program P1 in the sample is shown, and the second solid line curve represents the income probability distribution of the viewer of the program P2 in the representative sample. An analysis of the viewing habits of individual users of the programs P1 and P2 can be used to determine the most probable annual income. Using Bayesian inference, this probability can be expressed as:
P (Ixy | Wp1) = P (Wp1 | Ixy) P (Ixy) / P (Wp1)
In the formula, Ixy represents the annual income range of xy.
Wp1 represents watching program p1
P (Ixy | Wp1) represents the probability of income of Ixy when the program p1 is viewed.
P (Wp1 | Ixy) represents the probability of watching the program p1 when the annual income is Ixy.
P (Ixy) represents the probability of income for Ixy in a representative sample.
P (Wp1) represents the probability of watching program p1 in a representative sample.
Therefore, the probability that his income is within a certain range can be determined for the user who watches the program P1.
Programs whose demographic features are shown at the same rate that their viewers are observed in a representative sample (same as in the general population) do not contribute significantly to determining demographic features. Consider the graph of FIG. Graph (i) shows a bar graph representing the probability that a person in a representative sample is male or female. Graph (ii) shows the probability that a person is male or female among viewers of program P3 that is very similar to that observed in the general population. If Bayesian inference is used to determine the gender of a person who has watched program P3, the probability that the person is male or female is the probability that the person is male or female in the general population (this is an already known value). It will be very close to. By analyzing an individual's viewing habits for the program P3, no significant thing can be obtained. On the other hand, analyzing viewing habits for program P4 based on graph (iii) creates more certainty in determining a person's gender than what is already known about the gender distribution in the general population. Will.
The determination of which programs generate high probability values to determine a user's demographic characteristics is how different that viewer's demographic characteristics are from what is observed in the general population by some other means This can be done by applying heuristics with an algorithm that selects programs based on.
Depending on the program content and demographic characteristics of its viewers, one or more features may be determined using the same television program. A completely different set of programs may be needed to study different demographic characteristics.
Furthermore, the present invention contemplates analyzing other preferences of the user in addition to or instead of the user's television viewing preferences. Such information is, but is not limited to, music preferences, reading preferences, shopping preferences (especially preferences for clothes, furniture, jewelry, ornaments, appliances, electrical equipment) Category), political and religious tendencies, and the like. Such information may be obtained from any source, but in certain preferred embodiments it is directed to music clubs, book clubs, other specific sites of interest (eg, specific sports or hobbies) Sites), news sites, and other online sources that users are members of or visit frequently (typically accessible as world wide websites). Other information representing user preferences includes user subscriptions such as newspapers, magazines, newsletters, shopping catalogs, and other publications of particular interest.
From Bayesian inferences, it is possible to attribute a preference for a particular program based on user demographic characteristics, but that is not the only factor that explains the user's choice of a particular program . For example, consider the graph (ii) of FIG. It may be possible for user U1 who has a strong preference for program P2 for unknown reasons and who has a revenue less than x1. Thus, using Bayesian reasoning and just reaching U1's most probable annual income based on his viewing habits of only a single program P2 will not yield the most meaningful results. In order to reinforce Bayesian reasoning beliefs, user viewing habits must be analyzed for a set of programs in which the viewers of each program from that set exhibit similar demographic characteristics. The programs in the set are selected so that the degree of correlation of the spot colors indicated by the programs in the set is minimal. Each program that a user sees from this set of programs enhances the user's belief in Bayesian probability with its demographic characteristics.
To illustrate the need for minimal correlation, consider the previous example. In the case of another program P3 that is very similar in content to P2 and has the same demographic characteristics of that viewer as P2 is available to user U1, he Will have similar preferences as P3. In this scenario, strengthening the Bayesian belief by user U1 watching program P3 is not a decisive contribution. However, if the program P3 is very different in the content and features it shows, the probability of U1 choosing P3 is greatly reduced, preventing the Bayesian probability belief from being raised by mistake. In this situation, the probability of a user who actually has demographic features watching both programs P2 and P3 is quite high, and strengthening belief will be a decisive contribution.
Thus, the Bayesian probability belief function for demographic feature dc1 can be calculated in many ways, including but not limited to:
BF (dc1) = MAX (bp (w1), bp (w2) ..., bp (wm)) ^* cr1 + MAX (bp (wk), bp (wk + 1) ..., bp (wk + m)) ^* cr2 +. . . + MAX (bp (wn), bp (wn + 1) ..., bp (wn + m)) ^* cm
However,
-Bp (wn) is the Bayesian probability of the user having that demographic characteristic assuming the fact that he watched the program wn;
-(W1, w2, ... wm), (wk, wk + 1, ..., wk + m), (wn, wn + 1, ..., wn + m) is a set of programs, and any element of the set is different from another element of the same set Have a high degree of correlation, but have a very low degree of correlation with another set of elements,
-MAX (bp (wn), bp (wn + 1) ..., bp (wn + m)) represents the maximum Bayesian probability value of all possible values in the set;
-Cr1 represents the correlation coefficient of the set (w1, w2,... Wm), cr2 represents the correlation coefficient of the set (wk, wk + 1,..., Wk + m), and so on. Elements of the same set have a correlation coefficient.
The belief function of any demographic feature is represented by the S curve in FIG. The x-axis represents the increasing value of the belief function of the demographic feature for increasing values of x. The y-axis represents the increasing value of the user's probability of exhibiting its demographic characteristics for increasing values of y. As shown in the figure, the probability does not increase further as the value of the belief function becomes higher.
Some of the output of a representative specimen viewing habit analysis is schematically represented in FIG. 23b.
The advertisement is displayed by the program selection device according to the present invention when the user-calculated artificial statistical profile matches the advertisement target profile. Some of the elements included in the advertisement are metadata that includes information about the advertising content (which may consist of video clips, audio clips and / or graphic or text information) and the demographic profile to which this advertisement is targeted. It is.
The metadata contained in the advertisement is schematically represented in FIG. 23c. Here, the “number of target records” means the number of “target records” included in the “target advertisement metadata”. Each “target record” refers to a demographic feature to which this advertisement is targeted. A “target relationship rule” is used to determine a relationship rule. For example, advertisements can be targeted towards users who meet any one of the following criteria:
-Income level is between Ix and Iy
-Ethnographic background is eb1.
A “target relationship rule” will be used to specify the relationship.
“Demographic feature data” and “target advertising metadata” determined by analyzing the viewing habits of representative samples with known demographic characteristics are broadcast from the broadcast headend to the program selection device along with the program content and EPG data. Is done. This is shown in FIG. This is received by each user at each program selection device to determine a user demographic profile. Broadcasting of “demographic feature data” and “target advertising metadata” along with the advertising content may be done on a periodic basis, or always made available by the broadcast network or some other communication mechanism as required. Also good.
In the program selection device, “demographic feature data” and “target advertisement metadata” are collected and stored in a storage device such as a hard disk, a flash ROM, or a main memory, although not limited thereto. This collection of data may be done at a fixed time period or whenever appropriate depending on the embodiment chosen. For each spot color record in the “demographic spot color data”, the program selection device passes the past selection history. Each program on the “spot color value record” that can be used in the past selection history is used to calculate a belief function. The belief function distribution graph in the "spot value record" is then used to determine the probability of the user having that demographic feature. If previous probabilities for this demographic feature are available, a weighted average of the old and newly determined probabilities is taken and stored. Each targeted advertisement for which the “probability satisfaction criteria” of the “target record” is met by the user's demographic profile is selected and displayed at the appropriate time.
As discussed elsewhere in this disclosure, in another preferred embodiment, the present invention allows other targeted content to be provided to the user in addition to the advertisement.
Referring again to FIG. 1, the television control system 100 is preferably implemented by a general purpose digital computer and associated hardware that executes a stored program to implement the functions shown in block 100. The The exact hardware and software platform on which the television control system 100 is implemented is not critical and can take a variety of forms. For example, the television control system 100 may be implemented in a set-top box that can typically be used by an individual at home to receive CATV signals. Another implementation of the television control system 100 is in the form of a personal computer comprised of the necessary hardware and software for receiving and displaying television signals. An example of a set-top box that can be programmed according to the principles described herein is described by IBM Microelectronics in the following document: “Set-Top Box Solutions”, Product # G522-0300-00 (1997). November 19, 19); “Set-Top Box Reference Design Kit”, GK10-3098-00 (April 15, 1998); “Set-Top Box Peripheral Chip”, GK10-3098-00, (1998 4) May 15); “Set-Top Box Solutions: Helping Customers Meet the Challenges of Convergence”, 522-0300-00 (November 19, 1997); and "The challenges of convergence for Set-Top Box manufacturers", G599-0302-00 (11 May 19, 1997). An example of an application programming interface (API) available for a set-top box that can serve as a platform for the embodiments described herein is “Latest Set-Top Box” published by OpenCable. "API Requirements for the Advanced Set-Top Box" (October 21, 1997). An example of an operating system that incorporates functionality to support the embodiments described herein is available from OpenTV, Inc. and is described in the following technical white paper publication by OpenTV: “OpenTVTM Operating Environment” and “Application Development for OpenTVTM”. An advantage of such an operating system is the support provided in the form of function calls for obtaining attribute information 107 from the signal 104. Alternatively, a general purpose operating system, such as Microsoft Corporation's Windows NT operating system, extracts the necessary information from the attribute information 107, and receives the received signal 104 and stored information 105. It may be used in conjunction with additional software that provides the functionality required to perform other operations.
Storage device 106 may include a variety of different types of storage devices. For example, the preference database 116 may be stored in a nonvolatile random access semiconductor memory. Television program 105 and attribute information 107 may be stored on a storage device having a larger capacity, such as a conventional magnetic hard disk drive. In general, the storage device 106 is understood to include various storage devices. As long as the storage device has the capacity and speed to store the necessary information, the exact shape of the storage device 106 is not critical. Storage device 106 may include a conventional video cassette recorder (VCR) that operates to store television program 105 and attribute information 107 on a conventional magnetic tape under the control of system 100.
For purposes of this description, the television control system 100 is integrated with or coupled to a system that includes a tuner and other functions necessary to receive a television signal and extract attribute information 107 from the television signal. And is assumed to perform other functions typically associated with receiving and viewing television signals. In certain embodiments, the television control system 100 may operate in conjunction with a database agent that facilitates interaction with the preference database 116 by allowing information to be stored and retrieved from the database in an optimal manner. The preference database 116 is implemented by a commercially available database product, such as an Oracle Light database product available from Oracle Corporation, which also incorporates the functions for implementing the database agent described above. May be.
The recording manager 112 records the program 105 by periodically starting the sequence of steps shown in FIG. At 201, the recording manager 112 sends a request to the preference agent 110 for the evaluation of all programs for a particular time (X), or alternatively for the evaluation of all programs within a particular time period (X). As an example, the steps shown in FIG. 2 may be performed every 6 hours. In some embodiments, the frequency with which the steps of FIG. 24 are performed may be changeable by the user. The preference agent 110 responds at step 202 by giving an evaluation from the preference database 116 for each program received from the recording manager 112. Next, the recording manager 112 records the program at the time X or within the time period X according to the evaluation received from the preference agent 110. Specifically, the program with the highest rating is given the highest preference for recording, and the program with the lowest rating is given the lowest preference for recording. Recording is limited by storage capacity. For example, if the highest rated program is one hour long, only 30 minutes of recording space is available on the storage device 106, and the one hour program is skipped and rated 30 minutes. Is recorded.
The highest priority for recording a program is specifically given to the program requested by the user. For example, if a user identifies a particular program for recording, such as by specifying a date, time, and channel, or by specifying an identification code for the program, the user is more likely than a program evaluated by a preference agent. Priority is given to the recording of the program. The next highest priority is given to programs that match a particular category value pair specified by the user. For example, if a user does not identify a particular program but specifies that a one hour documentary about travel should be recorded, it matches such a category value pair than a program evaluated by preference agent 110. Priority is given to program recording. In an alternative embodiment, the relative priority among user-specified programs, user-specified category value pairs, and programs evaluated by preference agent 110 can be changed by the user.
The recording manager 112 manages the storage capacity on the storage device 106 by deleting the television program 105 according to the evaluation of such a program generated by the preference agent 110. This is done in a manner similar to that described above for determining which programs are to be recorded. FIG. 25 showing the steps taken by the recording manager 112 to determine which programs should be deleted is similar to FIG. In step 301, the recording manager 301 requests an evaluation from the preference agent 110 for all programs stored on the storage device 106. At step 302, the preference agent 110 responds by providing the requested delete rating. In step 303, the recording manager 112 responds by deleting the program as necessary according to the deletion evaluation received from the preference agent 110. Specifically, when additional space on the storage device 106 is needed to record one or more additional programs, the recording manager 112 first deletes or overwrites the program with the lowest rating. Thus, in conjunction with other stored television programs, the stored television program determined by the preference agent 110 as least preferred is deleted or replaced first, and the one determined to be most preferred is deleted or replaced last. Is done. Deletion of the program occurs only when necessary. Advantageously, as a result of this, the storage device 106 is typically filled to its maximum capacity, thus providing the user with the widest possible variety of programs. The user can specify a program to remain in the storage device 106. Such a program is not deleted by the recording system 100 in the step shown in FIG. Further, the user can specify the program to be deleted, and thus can ignore the steps shown in FIG.
In one embodiment, the preference database is used by system 100 to change the manner in which information about currently aired programs is presented to the user. For example, in such an embodiment, the preference database is rearranged in the order in which currently aired programs are presented so that the program having attribute information 107 rated highest by the preference database 116 is first presented. used. Alternatively, the preference database 116 can be used to organize information about currently aired programs according to various spot preferences stored in the preference database.
Figures 26a and 26b illustrate an alternative hardware configuration for a system employing the principles of the present invention. FIG. 26a shows a hardware configuration that supports the storage and retrieval of digitally encoded audio and video. Interface 902 is a standard digital cable or digital satellite input interface. The interface 906 is a hardware interface to the storage device 106 but preferably takes the form of an IDE or SCSI interface or the proposed IEEE-1394 interface. Interface 906 is an NTSC or PAL encoded video interface. If the television signal 104 takes the form of an analog signal and a CATV signal, as in the case of current television broadcast signals, the signal 104 is stored prior to storage on a digital storage medium such as that shown in FIG. It must be digitized and generally compressed (eg according to the MPEG-II standard).
FIG. 26b shows an embodiment using an analog storage device 106 such as a conventional VCR. If the television signal 104 is analog, the interface 910 takes the form of a conventional NTSC or PAL interface. If the television signal 104 is digital, the interface 910 takes the form of the interface 902 shown in FIG. 26a, and the digital-to-analog converter converts the received signal to analog before storage in the storage device 106. Required to convert to form.
FIG. 27 illustrates the operation of the automatic recording pause feature of the preferred embodiment. If the user is currently watching a broadcast program and wants to stop or temporarily stop watching the program, the recording system 100 advantageously allows the program to be recorded so that the user can subsequently watch the program. You can resume doing. As shown in FIG. 27, at 1002, the user is currently watching a broadcast program. By generating a pause input at 1004 by the user, such as by pressing an appropriate button on a remote control coupled to the recording system 100, the system 100 records at 1006 the program being viewed by the user. The user is free to watch another program or stop watching the television 108 at all. At a later time, if a viewing resume input is received, such as by pressing an appropriate button on the remote control described above, the recording system 100 records the program recorded in step 1006 in step 1010. Is taken out from the start point and can be viewed on the television 108. If the program is still being broadcast when step 1010 is started, recording of the program continues by system 100. Therefore, the user can easily interrupt the viewing of the current broadcast program and resume the subsequent viewing.
Preferably, the recording system 100 supports various functions such as fast forward, rewind and visual scan of recorded programs, as well as other functions supported by the storage medium 106. For example, if the storage medium 106 takes the form of a VCR, program viewing and manipulation functions are limited to standard VCR functions such as fast forward, rewind, forward or reverse visual scanning. If the storage device 106 takes the form of a digital storage medium, higher program search and retrieval functions can be supported.
It will be appreciated that the functions performed by the preference agent 110 and the recording manager 112 are illustrative of particular embodiments. However, the division of tasks between the two modules 110 and 112 may be changed. Further, the data formats 115, 116, 105 and 107 may take a variety of forms.
FIG. 28 illustrates an EPG and text information service according to an embodiment of the present invention. As shown, the billing vendor 10 of the local cable television company communicates via the billing link to the RS-232 port of the system manager 112 located at the cable head end. Billing vendor 10 includes a subscriber database and creates monthly bills for subscribers of the system based on the level of service and any pay-per-view purchases. The system manager 12 may be a personal computer or other processing device that receives a viewer authorization transaction from the billing vendor 10 and generates a transaction for delivery to a distribution device or subscriber. Such a transaction involves the subscriber's set-top box identifying which group of channels it is entitled to receive, which frequency to tune to a particular text data channel, and the audio sound of that text channel. Contains a text channel definition transaction that indicates whether to delete, pagination delay between pages, etc.
The system manager 12 also communicates via the headend link to the RS-232 port of the headend controller (HEC) 14 which controls transmission of television programming to the subscriber. As described in more detail with respect to FIG. 29, the HEC 14 communicates over the control link to the RS-232 port of the information service processor (or data controller) 16 that manages the flow of EPG and text data in accordance with the present invention. As indicated by the dashed lines in FIG. 28, the information service processor (ISP) 16 is preferably located at the cable head end along with the system manager 12, HEC 14, and signal scrambler. However, those skilled in the art will appreciate that not all of the headend equipment need be located in one place.
EPG data is supplied from one or more local or remote EPG suppliers 18 to the RS-232 port of ISP 16 via a satellite link, modem link or other communication link. Similarly, text data from one or more text channel suppliers 20 is provided to another RS-232 port of ISP 16 via a satellite link, modem link or other communication link. In the preferred embodiment, the ISP 16 has a plurality of identical RS-232 ports for accepting data from a plurality of EPG suppliers 18 and text channel suppliers 20.
Also, as shown, one of these RS-232 ports can be used for the control link to the HEC 14. ISP 16 manages the EPG and text source databases in response to control signals from HEC 14 to provide EPG data and / or text channel data to selected viewers.
The HEC 14 also provides control data directly to the viewer's set top box via the RS-485 output port. Preferably, the control data from the HEC 14 may include a text channel definition transaction and an EPG definition transaction to indicate to the set top box which frequency to tune to EPG data and the like. The control data may include software for downloading to the viewer's set top box to reprogram the viewer's set top box as needed. The control data from the HEC 14 can be inserted into the vertical blanking interval of the selected cable television signal by scramblers 22, 24 and 26 daisy chained using known in-band techniques. The control data from the HEC 14 may be modulated with an out-of-band carrier or an in-band audio carrier for transmission. Preferably, the scramblers 22-26 are daisy chained so that the scramblers can be addressed individually or globally.
Similarly, EPG data and text channel data from the ISP 16 are provided to the viewer's television set top box via the RS-485 output port of the ISP 16. EPG data and text channel data are similarly inserted by the EPG scrambler 28 and text channel scramblers 30 and 32, respectively, into the vertical blanking interval of the selected cable television signal. Scramblers 22-32 may insert control data, EPG data, and text channel data into other parts of the video signal, such as horizontal blanking intervals, or may completely replace the video. Typically, the number of scramblers depends on the number of premium channels on which scrambling is used. Preferably, EPG scrambler 28 and text channel scramblers 30 and 32 are identical to control data scramblers 22-26 and are similarly daisy chained for individual or global addressing. As shown in FIG. 28, the scramblers 28-32 carry combined EPG and text data and display control transactions as described in more detail for FIG. 29 for all data streams used. Receive a single serial data channel. Each scrambler may be equipped with a memory for storing a predetermined amount of this data in internal memory in order to minimize the number of database accesses. Preferably, the scramblers 28-32 have sufficient internal memory to store a significant number of transactions. For example, scrambler 30 may have sufficient internal memory to score a daily sports score for display on a sports text channel. The data received and stored in the scramblers 28-32 is preferably in RS-485 format and the preferred embodiment protocol is SDLC. All data transactions to scramblers 28-32 are sent in a separate scrambler (a separate data stream that specifies a station address in the SDLC protocol), and the control data can be configured by transactions from ISP 16 in the data stream. Thus, based on the address of the scrambler, it is transmitted in a global data stream that is filtered in scramblers 28-32.The individual EPG data and text data streams are preferably such that they can be assigned as desired. In general, the scramblers 28 to 32 have a baud rate of at least 9600.
Preferably, the subscriber's set top box is a set top box 34 that includes an EPG memory 36 for storing EPG data from the ISP 16. For example, the EPG memory 36 may store one week or two weeks worth of EPG data for selective access by the viewer via the set top box 34 menu. This menu preferably allows the viewer to scroll through the EPG data stored in the EPG memory 36 using the viewer's television remote control keypad. The set top box 34 may also include a non-volatile template memory 38 for storing a template into which EPG data is to be inserted for display to the viewer on the viewer's television 40. In this aspect, the video signal containing the template display data need not be continuously retransmitted to the set top box 34, thereby saving additional bandwidth. Instead, the EPG data need only be updated every 30 minutes or when there is a program change. Of course, different set top boxes 34 may have various amounts of memory and processing power for such purposes, depending on the allowable memory cost in manufacturing the set top box 34.
As shown in FIG. 28, the set top box 34 may include a text data memory 42 for storing a page of text data for presentation on the screen. Thus, the next page of text data can be loaded into the text data memory 42 while a page of text data is being displayed to the subscriber.
The ISP 16 manages the flow of text data and EPG data from the data service provider to the viewer's set top box 34. ISP 16 accepts data only from one or more authorized text data sources and / or EPG data sources, processes the text data and EPG data in its internal database manager, and processes the processed data in a common data transaction format. This data is managed by formatting it into a scrambler for transmission to the set top box 34. The provision of EPG data and text data to the subscriber is controlled by the headend controller 14 via a control link.
An example of a suitable ISP 16 includes a plurality of RS-232 serial input ports for EPG data input and / or text data input and at least one RS-485 HDLC serial link at its output of the type used by the HEC 14. It has an IBM PS model 7546 personal computer. As shown in FIG. 28, the control link may be a single RS-232 serial port. The hardware components and software components of the ISP 16 are configured as shown in FIG.
One embodiment of ISP 16 is shown in FIG. 29 as a plurality of RS-232 ports that provide a common interface for EPG data and text channel data provided asynchronously by EPG supplier 18 and text channel supplier 20. EPG data and text channel data are transmitted to ISP 16 via a satellite link (when the interface operates in simplex mode) or by a modem (when the interface operates in half-duplex mode). Preferably, the data is transmitted at a baud rate of at least 1200.
The ISP 16 functions as a “gatekeeper” that allows access only by authorized data sources. Thus, when ISP 16 receives a message from EPG supplier 18 or text channel supplier 20, it first checks the data for authority. If the supplier is not authorized, the data is ignored. When the supplier is authorized to access ISP 16, ISP 16 performs the requested action and returns a command response message. If the communication link is simplex, the response is ignored. In this aspect, access to ISP 16 may be restricted by an authorization code, but access will also depend on whether the data provider provides EPG data or text data in the transmission protocol expected by ISP 16, as described below. Limited.
Messages sent between the EPG provider 18 or text channel provider 20 and the ISP 16 can be formatted to include a start of text byte, a block of ASCII character data, a checksum byte, and an ASCII return. This format is used in commands sent from the data supplier to the ISP 16 and responses sent to the data supplier. The checksum may be a 2-byte CRC of all bytes in the message field starting with the first character following the start of the text character and not including but up to the maximum checksum field. The checksum is transmitted in the message as a hexadecimal ASCII representation (4 bytes) of the CRC calculation. The data block can be configured differently depending on whether the input data is EPG data or text data.
EPG data from EPG supplier 18 may be used, for example, to establish program definition commands used to identify all data associated with a single program, categories for identifying different types of programs. Can be formatted according to an EPG command set including category definition commands used and category deletion commands used to delete categories not used to make room in the ISP 16 database for new programming categories is there. EPG data can be formatted on a per program basis with these commands.
Delimiter characters can be used for variable length fields such as titles and program description blocks to specify the length of the field. For example, NUL (0 in hex) means the field is null, SOH (1 in hex) means the field is valid, ETX (31 in hex) is the end of the current record Means.
Once the data to be transmitted with the program definition command is stored in the EPG database of ISP 16, the EPG data is formatted into a transaction for transmission to the set top box 34. This command may also be used to update the program definition as it overwrites the corresponding entry in the ISP 16 EPG database.
ISP 16 has no error (normal response), no service provider found (not authorized), no service type found (not authorized), category ID discovery, unrecognized command May respond to such commands from the EPG supplier 18 by sending appropriate responses such as checksum error, insufficient disk space, etc. Other EPG commands and command responses may be provided as desired.
Text channel data originates from many different text channel suppliers 20 and can reach ISP 16 via different communication links such as satellites, dial-up models, direct connect modems, or direct connection to system manager 12. it can.
ISP 16 may include multiple databases and database managers. There can be two types of databases maintained in ISP 16, as shown in FIG. 29, one for EPG data and one for text channel data. The EPG database is designed to collect data from each EPG supplier and sort each EPG program record by channel and date. A separate database collects text data from the associated text channel supplier 20 and formats the received text data for transmission on individual text channels using techniques such as those described below. Generated for each text channel. Each database generated is identified by the service code type and service provider listed in the program definition command used in the control link command provided from the HEC 14 to the ISP 16.
As shown in FIG. 29, the received command is examined by the router and formatter 43 for its command code, service provider, service and authority code type as appropriate. If the command is from an unauthorized data source, subsequent data is ignored. However, if the received data is from an authorized supplier, the router and formatter 43 routes the data to the appropriate database in the ISP 16. For example, if EPG data is received, it is routed to the EPG database 46 via the EPG database manager 44.
In one embodiment, EPG database manager 44 sorts the received EPG data by channel and date and stores the received EPG data in an appropriate location in EPG database 46 for later reproduction. The EPG database manager 44 may also perform garbage collection on the EPG database 46 when records are deleted. The EPG database manager 44 maps the more frequently used characters to fewer than the usual 8 bits used in normal ASCII, while being less frequently used, as is known to those skilled in the art. Data compression software routines such as the Huffman Compression Algorithm that give more bits to the character may be called. The number of bits used for a character is based on its probability of appearing in the data stream. Huffman coding is described in detail in an article entitled “Lossless Data Compression”, Byte, March 1991, pages 309-314, incorporated herein by reference. Is done. Such a routine is desired to maximize storage efficiency in the EPG database 46. Similarly, each text database manager can store text information in an associated text database and performs data compression.
Router and formatter 43 and database managers 44, 48 and 52 are all controlled by a configurator 56 that responds to control data from HEC 14. Configurator 56 is responsive to control commands from HEC 14 for comparison with incoming data and as EPG supplier 18 and text channel supplier 20 are added or subtracted from the system. In order to add / subtract database managers and databases, etc., the router and formatter 43 are provided with updated authority information.
Access to ISP 16 can be carefully controlled through the use of authorization codes. Further, the ISP 16 may maintain control over the information services provided to the viewer by storing EPG data and text data in a particular format in an appropriate database in the ISP 16.
Referring now to FIG. 30, the EPG database key may be a combination of data and time fields from EPG data and the number of channels. These fields are followed by a duration field, a repeat field, a program evaluation field, a program category field, a critical field, an attribute flag field, a program feature flag field, a text data compression flag, and finally text data. The text data field may further consist of several optional subfields with a delimiter between each field.
The EPG database manager 44 can access the EPG database 46 through shared library routines such as adding records, deleting records, and reading records. When the EPG database 46 is used, it can be fragmented as records are added and deleted, so that the EPG database manager 44 can further include a garbage collection routine for the EPG database 46. The text database is configured similarly, but does not require garbage collection.
In one embodiment, the EPG transaction formatter 58 reads the database records of the EPG database 46, formats them into a program-based transaction with a predetermined number of bytes that is transmitted to the EPG scrambler 28, and It is inserted into the vertical blanking interval and transmitted to the set top box 34. These transactions are then sent via the transaction arbitrator 64 to the EPG scrambler 28 shown in FIG. 28 for insertion into the appropriate video channel.
Transactions from transaction arbitrator 64 may be output to a single RS-485 output port of ISP 16 that is connected to multiple scramblers of the type used to scramble the premium cable channel. Transactions are segmented into EPG and text data streams for transmission to the EPG scrambler 28 (if the transaction includes EPG data) or to the text channel scramblers 30 and 32 (if the transaction includes text data). It becomes.
In one embodiment, the EPG transaction generated by EPG transaction formatter 58 is formatted into an SDLC frame as described above. A sample SDLC format for EPG transaction data is shown in FIG. In FIG. 31, the start flag outlines the start of the SDLC frame, the station address represents the scrambler to be addressed, the control byte is a command code that defines what is to be processed, and the information field is the figure. Including EPG data formatted as at 30, the frame check includes a CRC for all data between the start and end flags, and the end flag outlines the end of the SDLC frame. A transmission from the EPG transaction formatter 58 addresses a particular data stream, and a response from the EPG scrambler 28 identifies that data stream in the station address location. As described above, such transmission may or may not require a response from the EPG scrambler 28.
An EPG transaction typically includes a byte that specifies that subsequent data is from an EPG data stream, a control code byte that specifies whether a response from, for example, a scrambler is expected, and an EPG data block An EPG block addition command including two bytes describing the number, a flag indicating whether the EPG data is short-term or long-term data, the number of transactions constituting the EPG data block, and the actual transaction is included. The EPG transaction formatter 58 may also generate an EPG block delete command that specifies that data is to be deleted from the EPG data stream, a control code byte, and the number of EPG blocks to be deleted.
FIG. 32 shows a flowchart of one embodiment of software that may be used and implemented in the EPG transaction formatter 58. As shown, the software begins at step 500 and obtains the system time and date from the ISP system clock 63 at step 502. Next, the expired EPG data block is deleted from the memory of the EPG scrambler 28 at step 504. An expired EPG data block is defined as a data block that represents a program that has been fully broadcast prior to the current system time, or a program that was broadcast before the time window used for the EPG. At step 506, the current EPG data block having the time and date within the EPG time window is read from the EPG database 46. The current EPG data block is then formatted at step 508 into an EPG block add command and associated transaction. Next, the block / time map of the EPG transaction formatter 58 is updated at step 510. The block / time map preferably stores the time at which each EPG data block was sent to the EPG scrambler 28. Next, at step 512, a transaction representing EPG data is transmitted to the EPG scrambler 28. Next, at step 514, the EPG transaction formatter 58 makes the next EPG update (which should happen when the system time enters the new 30 minutes) or the next EPG change (which can happen at any time). Wait for. Upon receipt of such an update or change, control returns to step 502. Text transaction formatters 60 and 62 similarly generate text transactions for text data defined on a per screen basis (not per program) as described above. Thus, the add text screen command is similar to the add EPG block command except that the number of text channels and the number of screens are given instead of the number of EPG blocks and short / long term data bytes. Text transaction formatters 60 and 62 may require time from the scrambler so that proper pagination can be maintained.
FIG. 33 shows a flowchart of one embodiment of software that may be used and embodied in the text channel transaction formatters 60,62. As shown, the software begins at step 600 and reads a text screen record from text database 50 or 54 at step 602. The text screen is formatted into a text screen add transaction at step 604 and transmitted to the text channel scramblers 30 and 32 at step 606. Preferably, such transactions are formatted such that display characters are sent as display commands rather than as separate characters for each display coordinate of the text display screen. Next, at step 608, the text transaction formatters 60, 62 wait for the time period specified by the system manager 12 (if automatic pagination is used), after which the next text page is formatted and the text channel scrambler. 30 and 32. At the end of this time period, control returns to step 604 where the next text screen record in the text database 50, 54 is formatted and the text scrambler 30 for insertion into the vertical blanking interval of the particular video signal. , 32.
The text data may be sent to the screen by sending a separate character for each display location on the page. In other words, if the text screen contains 16 lines and 4 characters per line, the text screen is represented by sending 384 (24 × 16) characters, one for each display location on the display screen. . A blank space character is sent to indicate that the character is not in a particular text screen location.
In some embodiments, the text data is transmitted to the screen along with appropriate commands for controlling the display of the text data. For example, the first display command in the sequence identifies the subsequent data as text data and instructs the set top box 34 to fill the TV screen with a blue background or some other background or template in which the text is displayed. To do. The text data is then converted into a series of commands that together identify separate screens of text. As mentioned above, text data is grouped on a per screen basis, but this allows an appropriate delay mechanism to be incorporated into the display command, reducing the delay required between each text screen presentation. Provide.
For this purpose, the transaction formatters 60 and 62 scan the text data for actual characters, skip excess space in the text data, group the actual text, and perform vertical blanking of the field of the video signal. Software may be included for transmitting in a transaction of a specified size that conforms to the erase interval. Since the space is removed, the display command contains coordinates that specify the row and column address of the first display character on the screen, and multiple adjacent spaces until the transaction is full or multiple consecutive spaces are encountered. A character follows that character during the same transaction.
Attribute information such as underline, blinking or intensity reversal associated with a character can also be transmitted using these display commands. These display commands are used to read the text data for the text screen from the appropriate database, the display command is transmitted at the end of the text data for the text screen, and all data for that text screen is transmitted. Indicates that it has been done. Transaction formatters 60, 62 also include a wait loop or “timeout” command at the end of the transmission, which allows the viewer sufficient time to view the text screen before the text data for the next text screen is displayed. Is incorporated into the delay (which is on the order of 7 seconds), thereby giving automatic pagination of the text screen.
Automatic pagination allows the viewer to automatically advance from one text screen to the next without any viewer intervention. According to the automatic pagination scheme of the present invention, the cable operator can specify the duration between screens and transfer this information to the transaction formatters 60,62. Then, during operation, when the viewer selects a text channel, the selected text channel data is extracted from the vertical blanking interval of the video signal into which it is inserted and the text data is displayed on the text channel display. The current page of text data is displayed by associating it with the viewer's television channel designated for. The next screen of text data is displayed after a predetermined delay (about 7 seconds) that gives the viewer enough time to read the text data displayed on the current screen. This technique can replace the commonly used “Barker” channel, which uses a computer to generate text data that is then transmitted as a complete video channel by a cable television system.
In response to the control command from the HEC 14, the configurator 56 provides the updated authority information to the router and formatter 43 to be compared with the incoming data, and the EPG supplier 18 and the text channel supplier 20 add it. Database managers and databases etc. can be added / subtracted as they are subtracted from the system. A control link between the HEC 14 and the configurator 56 can be used to report the status of the ISP 16 to the system manager 12. In addition, the control link may accept text data from the system manager 12 to display system messages and the like.
In one embodiment, the interface between configurator 56 and HEC 14 is an RS-232 port with a data format fixed at, for example, 9600 baud. All control data is preferably transmitted as ASCII characters. Upon receipt of the message from the HEC 14, the configurator 56 examines the data and performs the requested action, and of the type described above for communication between the router and formatter 43 and the EPG supplier and text channel supplier. Returns a command response message in message format. Sample commands sent from the HEC 14 to the configurator 56 over the control link include date / time setting commands (for synchronization purposes), configuration request commands, status request commands, get category record commands, scrambler control commands and database control commands. including.
In one embodiment, in operation, ISP 16 monitors all input port data from EPG and text data service providers and builds a list of all available EPG and text data services. This list is sent to the system manager 12 using a configuration request command. This command includes the available service providers, the type of service from each provider (EPG or text data), the communication port associated with each service, the scrambler address or data stream for each service (EPG or text data), each Specify the authorization code from the supplier for each service, the time and date of the last update from the service provider, the time and date of the last update to the scrambler, the time and date of the latest EPG data in the EPG database, etc. To do. When this command is given, such information is provided to the system manager 12 for each service provider.
The status request command may include a flag that indicates whether there is an error present in the error log and whether the category list has changed since the last status request command was received. Next, error and category information may be extracted using a get error record command and a get category record command.
In one embodiment, the configuration commands are separated into EPG service configuration commands and text service configuration commands. The EPG service configuration command is for service provider, type of service, service is enabled or disabled, authorization code from EPG supplier 18, scrambler data stream for short-term data, long-term data Scrambler data stream, length of short-term data in time (1-255), and length of long-term data in time (1-4096). The text service configuration command includes the service provider, the type of service (weather, sports, etc.), whether this service should be enabled or disabled, the authorization code from the text channel supplier 20, data about text data Specifies the pagination delay time in seconds before the next page of text data replaces the current page of text data for stream or scrambler address, number of channels, and automatic pagination.
The scrambler control command, for example, specifies the amount of scrambler memory in kilobytes that is used when the scrambler is replaced and needs data to be reloaded into its memory, and the amount of scrambler memory A scrambler configuration command.
In some embodiments, the database control command can be, for example, a database clear command used to clear a database associated with a particular service, and a database used to delete a database associated with a particular service. Contains a delete command. Other database control commands such as a category map download command may be provided to establish a list of designated categories of program data in the EPG data.
FIG. 34 illustrates one embodiment of a set top box 34 that can be used with the present invention. As shown, the set top box 34 reads commands from the EPG memory 36, the template memory 38, the text page memory 42, the set top box 700, and the vertical blanking interval of the incoming video signal. A set top processor 702 for performing For example, if the incoming command is a text channel definition or EPG definition command from HEC 14, an appropriate update of bitmap 704 is performed. Similarly, if the incoming command is a display command including EPG data, the data is stored in the EPG memory 36 and a menu by the TV remote control unit 706 and the remote receiver 708 from which the user requests display of the EPG data. When the selection is made, it is displayed together with the template stored in the template memory 38. If no template memory is provided, the template data may be sent as part of an EPG display command. On the other hand, if the incoming command is a display command containing text data, the page of data is stored in the text page memory 42 to present one page at a time on the display. If automatic pagination is enabled, the text page memory can be automatically updated every few seconds due to the delay incorporated into the display commands from the text formatters 60,62. Alternatively, a user may be able to manually access text data in memory. If manual access is provided, the text page memory holds at least the currently displayed text page, the previous text page, and the subsequent text page to give the user the ability to scroll through the text data. It is preferable. In any case, the set top processor 702 preferably has the ability to request the next text page while the current page is displayed, so that the next text page is displayed for display at the end of the screen delay time. Already loaded. Next, the selected text, EPG, or video signal is modulated by modulator 710 and displayed on television screen 40 on the channel specified in bitmap 704.
The bit map 704 of the set top processor 702 of the set top box 34 specifies the frequency that must be tuned by the box 700 to receive the desired text data, and displays the received text information for display. Associate with the specified cable channel. This information is received from the HEC 14 in the aforementioned text channel definition transaction. For example, the viewer may specify via television remote 706 that he wants to watch a sports text data channel indicating that the program guide is available by tuning set-top box 34 to channel 181. The set top processor 702 then checks the bitmap 704 for channel 181 and extracts the sports text data for channel 181 of the viewer from channel 29's vertical blanking interval for channel 29. The set top processor 702 sets the set top box 100 to tune channel 29, but the video signal on channel 29 is not displayed. Instead, the video screen is erased by the set top processor 702 and the text data extracted from the vertical blanking interval by the set top processor 702 is displayed. The necessary descrambling of the received video is performed by the set top processor 702. Thereby, the viewer perceives that many more “virtual” channels are available even though a separate video channel is not used to transmit the text data.
In one embodiment of the present invention illustrated in FIG. 42, electronic content is tagged and aggregated on a server, resulting in electronic content with targeted information. Suitable electronic content includes, but is not limited to, advertisements, news segments, video programs, audio programs, WEB pages, and the like. Electronic content with targeted information is transmitted to a broadcast server and widely broadcast. Set top box 34 generates a profile representing different people in the household. The set top box 34 matches the generated profile with the tags in the broadcast stream and determines which of the tagged content should be downloaded. Of the broadcast content, all subsets associated with individuals in the household are downloaded and displayed to the user.
For example, the set-top box 34 may be used to match tags in a broadcast stream with previously generated demographic profiles of viewers to determine which of the tagged content should be downloaded. it can. Individual viewers receive only from the set top box 34. The set top box 34 matches the tags in the broadcast stream with the demographic profile of the generated viewer and determines which of the tagged content is to be downloaded.
Some examples of demographic characteristics include, but are not limited to, the user's gender, race, age and income. The output of the representative sample viewing habit analysis is the basis for determining the demographic characteristics of individual users.
As shown, the set top box 34 reads incoming tagged and assembled electronic content commands and takes appropriate actions. For example, if the incoming command is a text channel definition or EPG definition command from the HEC 14, the bitmap 704 is updated appropriately. Similarly, if the incoming command is a display command including EPG data, the data is stored in the EPG memory 36, and the menu selection by the TV remote control unit 706 and the remote receiver 708 that the user requests to display the EPG data is performed. When done, it is displayed with the template stored in the template memory 38. If no template memory is provided, the template data may be sent as part of an EPG display command. If the incoming command is a display command containing text data, the page of data is stored in the text page memory 42 to present one page at a time on the display. If automatic pagination is enabled, the text page memory can be automatically updated every few seconds due to the delay incorporated into the display commands from the text formatters 60,62. Alternatively, the user may be able to manually access text data in memory. If manual access is provided, the text page memory holds at least the currently displayed text page, the previous text page, and the subsequent text page to give the user the ability to scroll through the text data. Is preferred. In any case, the set top processor 702 preferably has the ability to request the next text page while the current page is displayed, so that the next text page is displayed at the end of the screen display time. Already loaded for. The tagged and assembled electronic content is then modulated by a modulator 710 and displayed on the television screen 40 on the channels specified in the bitmap 704.
The bitmap 704 of the set top box 34 associates the received tagged and assembled electronic content with the designated cable channel for display. Next, the set top processor 702 examines the bitmap 704 for channel 181 and determines that it must tune the frequency for channel 29 in order to extract it. The necessary descrambling of the received tagged and assembled electronic content is performed by the set top processor 702.
The present invention also provides a mechanism for automatically generating multiple profiles corresponding to multiple users with or without information about themselves explicitly provided by the users. FIG. 35 illustrates an embodiment of the present invention that results in automatic generation of multiple profiles and automatic identification of profiles when they are active on the device.
In some embodiments, the number of profiles to be generated is initially determined. Several methods can be used to determine the number of profiles. In one embodiment of the invention, the number of profiles is optionally determined by applying a minimum message length (MML) criterion. The process of applying the MML criteria to determine the optimal number of clusters is described in the reference (ref.). In one embodiment of the invention, the device user explicitly specifies the number of profiles to be generated.
The system monitors user interaction with the device. User interaction may include, but is not limited to, a request to change channels, request more information, configure device parameters, delete or record programs from storage. All actions by user interaction are recorded in a history database that stores a history of viewer behavior. Data records in the history database that describe viewer behavior can be in different formats. FIG. 37a shows one possible format for the data record. The data record may include information about the action, the time when the action occurred, and some parameters that further describe the action. In one embodiment of the invention, only actions that meet certain criteria are recorded in the history database. For example, all user actions related to viewing a channel can be ignored if the viewing duration is less than a configurable threshold, and all of the viewing of a particular channel, such as a notice channel, can be ignored. User behavior can be ignored. In default usage of the device, the user does not identify himself during each usage session. A usage session in this context can be a period in which there is user activity between two periods of inactivity, or a period in which the device is used between two periods when the device is not in use. In order to give the user a very personalized environment, it is up to the device to determine who the actual user is. In such a scenario, the user behavior record does not include information about who the user is. In some embodiments of the invention, the user can explicitly identify all or part of the usage session by using some mechanism provided by the system. These mechanisms may include pressing a series of keys or selecting a username from a menu and logging in as that user. In this embodiment of the invention, the user may choose to be identified during certain sessions and not choose to be identified during any other session. When a user identifies, some user action record may contain the user's name or some other identifying data as a parameter that specifies that the user was logged in during the session in which the user action record was generated Can have.
By analyzing a user behavior history database generated during a period of time, the present invention provides a mechanism for automatically generating multiple profiles. Each of these profiles may correspond to a single user's overall preferences, a single user's group of related preferences, a group of user's related preferences, or a collective preference of a group of users. The mechanism for generating multiple profiles is represented in FIG.
A set of consecutive user action records are grouped together to form a usage pattern record. The usage pattern record may be in the form of an array of user behavior records. Only user actions that occur continuously are grouped into a single usage pattern record. Usage pattern records can be formed using a number of methods, some of which include:
1. Group all user behavior records in a single usage session into a single usage pattern record. This is represented graphically in FIG. 38a.
2. All user behavior records in a single usage session are grouped into one or more usage pattern records such that each usage pattern record has a predetermined number of user behavior records. This is represented graphically in FIG. 38b.
3. One predetermined number of consecutive user behavior records in a usage session such that each usage pattern record has a plurality of user behavior records that overlap some of the user behavior records in adjacent usage pattern records Group into the above usage pattern records. This is represented graphically in FIG.
Each usage pattern record is associated with a point in an N-dimensional space, and each axis in the N-dimensional space represents a parameter that is important when specifying a plurality of profiles. The N dimension of this space is called the cluster dimension. This N-dimensional space parameter is either manually determined by those skilled in the art identifying the most meaningful parameters in identifying multiple profiles, or by automatically identifying profile aspects that differ significantly between profiles. It is selected by. In one embodiment of the invention, a set of these parameters is identified and configured when the device is configured during initialization, and by looking at different aspects in multiple profiles stored in the device, a new set is created. Added periodically. Various combinations of initial parameters may be used. In one embodiment, these initial parameters are the channel name and view count in the channel lineup. In this example, these initial parameters are possible values for the program description field of the television program. In some embodiments, the rate of channel change is also a parameter. The association of the usage pattern record with the point in the space defined by the cluster dimension can be performed by many methods. One method determines the amount of a particular feature defined by the clustering parameters indicated in the usage pattern record, such as the number of channel changes in the usage pattern, and uses this as the value of the corresponding dimension axis. One method determines the rate of consumption of a particular feature in the usage pattern record, for example, the rate of consumption for “NBC” in the usage pattern record is 5, which means that the program on “NBC” is this It shows that it was seen 50% of times during the period of the usage pattern record.
All points in the space defined by the cluster dimension are clustered into multiple clusters using standard clustering algorithms. Clustering can be performed using any clustering algorithm. In one embodiment, EM clustering is applied to group points in the cluster dimension space into a predetermined number of clusters. The number of profiles generated determines the number of clusters formed. Those skilled in the art of artificial intelligence and clustering, particularly EM clustering, will be familiar with these clustering techniques. Each cluster formed as a result of clustering represents a single profile. The clustering process provides a blend weight for each of the clusters as one of the outputs. Using the blend weight for the cluster, the percentage of time that the profile was active out of the total amount of time the device was used can be calculated.
In one embodiment of the invention, clustering is performed periodically using user behavior records stored in a history database. In one embodiment, user behavior records are periodically removed from the history database based on certain criteria such as record age, record size, and record relevance.
The invention also provides a mechanism for predicting a profile that is active at any given time. FIG. 40 shows a method for identifying a profile that is currently active on the device. The current user behavior is monitored and a user behavior record is generated. A usage pattern record is generated using the latest user action record. The usage pattern record is associated with a point in the cluster dimension. Using information about clusters representing multiple profiles, the probability that each of the clusters is active for a given usage pattern is calculated. This is calculated using Bayesian probability theory, which can be used to calculate posterior probabilities using prior probabilities and blend weights. The probability that a cluster is active is calculated using the probability of a usage pattern record (P / use_pattern profile), the probability that the profile is active, and the probability that a usage record occurs for a given profile. In this context, the mathematical expression for Bayesian probability theory is:
P (profile / usage_pattern) = (P (usage_pattern, profile) * P (profile)) P (usage_pattern)
In this equation, the probability (P (use pattern, profile)) that a use pattern record occurs for a given profile can be calculated by knowing the probability distribution governing the cluster. In one embodiment, the cluster probability distribution can be assumed to be a Gaussian probability distribution, so P (utilization_pattern profile) can be calculated using the cluster center and cluster variation. Cluster center and cluster variation are outputs generated by the clustering process performed to generate a plurality of profiles.
In some embodiments, switching the device on may be one of the recorded user actions, and “time of use” is one of the cluster dimensions. In this example, as soon as the user switches on the device, the probability that any cluster is active can be determined without further user action. As the user performs more user actions, the current usage pattern record is refined to include more user action records, and the probability of a profile that can be active is determined by adding each new user action record. Refined. As more user behavior records are added to the current usage pattern, the oldest set of user behavior records may be removed from the current usage pattern record. This process ensures that the impact of individual user behavior in identifying activity profiles decreases over time.
In one embodiment of the invention, the clusters generated by the clustering procedure are used to generate multiple profiles using the method depicted in FIG. In this embodiment, clusters are used to calculate the probability of each cluster being active during each usage pattern record. User preferences for each profile are generated using the process described above using user behavior weighted by the probability of the corresponding cluster being active.
In yet another embodiment of the invention, multiple profiles may be generated for each user. Each of such multiple profiles may be used to represent a user's viewing preferences at various time periods such as weekdays or weekends or morning or evening. The profile also follows other variables, such as showing more sports during the football season, showing more gardening programs in the spring, and showing more political reviews during the primary election every four years. There may be various. Thus, when the system of the present invention is activated and the user who activated the system is recognized, the system can further determine the date and / or week and access the appropriate profile for the user.
Referring to FIG. 43, in another preferred embodiment of the present invention used to deliver advertising content to viewers specifically targeted based on individual viewer profiles, the broadcast network may have many different Content is gathered from the content providers and combined at the network broadcast head end 29 to be a broadcast signal transmitted to and received by the television control system 2 of the present invention. The control system 2 sorts the content received from the broadcast network based on different viewer profiles of a particular household, edits the list of preferred programs available to each user, and records the programs of interest. Make programs that you can remember and watch when you need them. Equally important, the control system 2 receives and sorts the various advertising content and stores the content most relevant to the household audience based on the demographic profile and the targeted audience of the advertising content.
Even more specifically, as shown in FIG. 43, the content providers include an EPG provider 18, a text channel provider 20, a traditional programming provider 23, and an alternative content provider 25. As described below, alternative content refers to highly targeted programming that can be delivered through the use of the control system 2 described herein. Traditional programming includes, but is not limited to, all types of programming currently available, including movies, news, sporting events, educational programs, paper view programs, and shopping channels. Advertising content providers 19 are programming suppliers 23 with promotional materials such as short video clips or graphics and text information for inclusion in their traditional programming content in a manner familiar to TV viewers around the world. To provide. However, in a novel method of delivering advertising content that is uniquely available through the use of the present invention, the advertising content can be provided to alternative content suppliers (see description below), each of the personal advertising messages May be profiled by the profiler 21 to develop metadata describing the. As discussed throughout this specification, metadata refers to demographic characteristics that describe the audience to which each advertisement is specifically targeted.
The profiler 21 may be a software-based system or may rely on human intervention to develop metadata that views and analyzes each advertising message to describe the advertisement. Alternatively, the advertising content supplier 19 may provide this information along with the advertisement. The metadata is developed based on parameters of interest to the client where the ad is developed and typically includes age, gender, ethnic background, education, occupation, and other demographic information related to the viewer. Including. Thus, the metadata will describe the viewer to whom each advertising message is targeted for viewer demographic information defined by the parameters described above. The metadata generated in this way is added to each advertisement, and the advertisement tagged in this way is given to the broadcast network and included in the broadcast signal.
Tagged advertisements are preferably provided to multiple channels, where each channel may have a specific type of advertisement for the target audience. Thus, each channel may be defined according to one demographic parameter or often a repeated combination. For example, a channel having an advertisement for men, a channel having an advertisement for women, a channel having an advertisement for male sports fans, and the like. Alternatively, each channel may have all types of advertisement flows, where each type of advertisement exists in a ratio that depends on the demographic parameters to which it is targeted. Thus, on weekend afternoons, there are more ads targeted to male sports fans than ads targeted to small children. This is because many sports events are broadcast at such times and small children are more likely to play outside than watching TV indoors at such times. Providing tagged advertising content via multiple channels is not a requirement of the present invention, but the pure amount of available advertising content may actually require multiple channels. Further, providing the channels in a pre-classified manner may reduce the amount and complexity of classification performed by the control system 2 (described in more detail below).
As shown in FIG. 43, the signal broadcast by this network may further include information provided by the administrator of the service provided by the present invention, such as Metabyte Corporation, the assignee of this application. As will be explained in more detail elsewhere in this application, this information may include software updates of the control system 2.
Still referring to FIG. 43, the network broadcast signal is communicated to and received by the control system 2. The broadcast signal may be encoded analog or digital and may be transmitted via cable, satellite, telephone line, or any other feasible aspect. Practical considerations will dictate broadcasting network signals in digital form from the many channels and base technologies typically available. This requires that both the network head end 29 and the control system 2 incorporate ADC and DAC circuitry in a manner well known to those skilled in the art. Furthermore, although not shown in the figure, the control system 2 is required to separate a number of multiplexed signals in the broadcast signal and select any one of these signals (ie, the channel) as required. It will be appreciated that a channel conditioning circuit needs to be incorporated.
When the broadcast signal is received by the control system 2, it is supplied to the various components of the system. The program switch 114 directs the channel signal to the TV monitor 108 in response to viewer input provided via a remote control or similar unit to select the desired channel. As described above, the preference agent 110 monitors the viewing selection of various viewers using the control system 2 and generates a viewing profile for each viewer stored in the preference database 116. Based on these profiles, the preference agent 110 classifies incoming programming content as described in the EPG information, such as a “Top 10” list of viewing choices available at any given time for each viewer, etc. Edit the list of videos and record in the recording manager 112 the top ranked programs broadcast at any given time (including any programs selected by the viewer for recording) and store them Instruct to store in memory device 35.
The preference agent further includes software that allows a demographic profile for each viewer to be generated based on the viewer's viewing profile and a particular algorithm or associated rule. These algorithms may be adjusted over time as the model used by system administrator 27 improves and its accuracy improves. For this purpose, the system update information channel included in the broadcast signal may include periodic software updates including new preference database parameters that may need to be included at the request of the advertisement provider 19. Thus, in one embodiment, the control system 2 of the present invention is designed to meet any new demands that may arise as advertising content providers become familiar with the system and the process of customizing the narrowly focused targeted advertisement. May be improved remotely. The demographic profile generated for each viewer is stored in the demographic database 31, which ensures the viewer's privacy to be located in the control system 2.
The preference agent 110 also classifies the advertising content that flows through multiple advertising channels included in the broadcast signal and includes metadata included in each advertisement to describe the demographic profile of the viewer and the target audience for a particular advertisement. To store and / or display specific advertisements. As will be described later, the control system 2 may operate the advertising content using any of various methods.
In one embodiment of the invention, each of the advertising channels has the same type of advertisement. The preference agent determines which viewers are watching TV at any given time and stores the tagged advertisement targeted to that particular viewer in the stored advertisement memory device 33. At the appropriate time in the program that the viewer is watching, such as during a commercial break that is typically inserted into most TV programs, the preference agent can be selected by accessing the program source switch 114 to access the stored advertisement. The TV monitor 108 is instructed to play the advertisement. When the program that the viewer is watching contains information about the length of the commercial break, the preference agent may select a stored advertisement of the appropriate length for insertion into the assigned time frame. The preference agent may further ensure that all stored advertisements are displayed equally by remembering previously played advertisements. Alternatively, a tagged advertisement may include a desired number of times that the advertisement provider wants the advertisement to be broadcast on any given day, or sometimes a specific time at which the advertisement should be shown. Thus, after some advertisements are memorized, for example, enough for 24 hours, the preference agent reviews all memorized advertisement metadata and when and what to show these advertisements to the viewer You may develop a strategy that includes how to show times.
Alternatively, the preference manager may display a specific number of advertisements and a recording manager to record the selected program when the displayed stored advertisement is longer than the time frame allocated for the commercial break. 112 may be instructed. Therefore, the control system 2 of the present invention can manipulate the broadcast schedule of the program to some extent by changing the amount of advertisement content received by the viewer.
In another embodiment, the advertising channel may only operate during a limited time of the day, typically during off-peak hours (such as 2 am), during which all advertisements for the next day May be downloaded and stored. That is, the advertisement channel may be turned on at a particular time and all advertisements may be run through the control system during one or two hours. Alternatively, a regular programming channel may pause programming for one or two hours at a convenient time and serve advertisements for the next day. In yet another alternative, a dedicated connection such as a telephone line or internet connection may be used for periodic downloading of advertisements.
In an alternative embodiment, multiple ad channels have a mix of ads so that at any given time, a preference agent can target at least one ad targeted to a viewer from one of these channels. You may have the option to select. In this embodiment, the control system can store only one advertisement at any given time to ensure continuity between the program being viewed and the advertisement. For example, a commercial break of a program may begin at a time that does not correspond to the start of advertising on any advertising channel. In this case, the advertisement stored in the control system is directed to the TV monitor through the program source switch, while another targeted advertisement is stored simultaneously for subsequent display. When a commercial break happens to coincide with the start of a targeted ad for one of the ad channels, the preference agent may simply cause the program source switch to point a particular channel to the TV monitor while another ad Another advertisement from the channel is remembered.
In another embodiment, each regular programming channel may have multiple multiplexed versions of an advertisement. When a commercial break begins, the preference agent selects the most appropriate version of the advertisement and directs it to the program source switch. This embodiment requires additional circuitry to demultiplex the various versions of each advertisement and apply the specific version selected by the preference agent to the program source switch.
The new method of delivering targeted advertisements provided by this invention is particularly advantageous for the advertising industry. In one aspect, the present invention allows greater freedom in advertising production. This is because advertising no longer needs to appeal to a wide variety of viewers. In addition, because alternative advertisements are delivered in real time, all viewers will receive personalized advertisements at the same time compared to the prior art, which requires five different time frames to broadcast five different advertisements. Provides significant time savings. This realizes cost savings, which can be especially noticeable during very popular events such as the Super Bowl.
As described above, the preference agent 110 may classify the available channels and select the 10 (or any other number) currently playing programs that most closely match the viewer's viewing profile. In addition, the preference agent may build a customized list of future programs based on the viewer's profile and any additional criteria specified by the user. In the preferred embodiment, the user can fully customize his viewing profile, including values assigned to the different parameters that make up his or her viewing profile. In an alternative embodiment, the user may even be able to identify the type of advertisement he likes. Such a configuration can create an alternative billing device so that the user can see the ad in exchange for being able to identify the type of ad he wants to receive and possibly in some rewards. I agree.
Another new possibility that may arise with this invention is to always show the most interesting programs broadcast at any given time and / or programs that fit very close to the previously recorded and stored viewer profile. Creation of a “personal channel”. Thus, by turning on the personal channel, the viewer is always assured of the most personalized and interesting viewing experience available based on the program broadcast during a particular prior period that the viewer can specify. . Thus, a personal channel may show a collection of movies, news segments, sporting events, and any other programming content that has been broadcast over the past 24 hours and fits the viewer's profile to the extent specified. Good.
An additional component of the control system 2 that may be incorporated only into the selected control system is to delete any personal information from the demographic database and use this anonymous information to generate a demographic database A privacy filter 37 that communicates to the system administrator 27 for the purpose of maintaining and updating. Such a control system model requires a feedback line to communicate demographic information to the system administrator, which in the preferred embodiment is a telephone line or a dedicated internet line such as DSL, cable, etc. The system administrator may offer a reward or other reward to the user to convince the user to supply this type of information. By carefully selecting users from a wide range of demographic profiles, system administrators can use the information gathered in this way to compare these users' actual demographic data with demographic profiles developed by preference agents. By doing so, the model used to develop the demographic profile of the viewer based on his or her viewer profile can be improved.
Due to the narrowly focused nature of advertisements delivered by the present invention, targeted advertisements developed for delivery by the control system of the present invention will have new configurations such as coupons or highly targeted descriptions. It may contain elements. Thus, targeted advertisements developed for distribution through the present invention may include specific information expressed in a specific language that is specifically intended to appeal to the target audience.
The invention further enables the creation of highly targeted content other than advertisements that can only be delivered to very specific viewers. Thus, movies, shows, religious programs, video magazines, infomercials, etc., without the restrictions typically imposed on their content developers when the program reaches or is at least accessible to a very wide audience, It can be developed to reach a very specific audience. This embodiment requires that such specific content is supplied via a specialized channel that is not directly matched by a conventional TV tuner and can only be accessed through the control system 2. Such highly targeted content may be provided by alternative content provider 25 or may be developed for alternative distribution by conventional programming provider 23. By using this invention, a new content content provider can not only reach very specific viewers but also automatically exclude certain parts of the viewers remotely from not accessing this material A simple distribution medium. Such alternative content may be broadcast exclusively on the viewer's personal channel as described above.
Thus, the present invention enables the delivery of highly targeted content to viewers including not only advertisements but also other programming and content. Referring to FIG. 44, a timeline 800 that tracks the progress of a typical television program 850 is shown. In conventional linear programming currently available to TV viewers from broadcast and cable / DSS stations, program 850 consists of scenes such as scene 1 and scene 2 and advertisements such as Ad1 and Ad2. According to conventional linear programming, the scenes are shown in order as predetermined by the program producer, with the display of advertisements interspersed there as determined by the headend operator. According to the present invention, customized linear programming is enabled by the development of demographic and / or viewer preference profiles for each user. The customized linear program 860 may also include scenes with scattered advertisements, but some or all of the scenes and advertisements may be targeted to various demographic or viewing features of the user. Scene 2 may actually have two different versions filmed by the TV program producer, Scene 2 and Scene 2a, each version of that scene being more appropriate for a particular target audience, or more It is interesting or less unpleasant. Thus, in one possible embodiment, scene 2a may be a less violent or less explicit version than scene 2. The system of the present invention receives a TV signal having both scenes and selects among the available alternatives based on the profile of the currently viewing user and the target data of the alternative scene. Therefore, the program 860 shown in FIG. 44 is linear as including scene 1, alternative scene 2a, alternative scene 3b, alternative Ad1a, alternative Ad2c, scene 4, Ad3, and alternative Ad4b. Shown to the user in a seamless manner.
The same adaptation and selection process occurs for scenes 3, 3a and 3b, and advertisements. Ad1 may be provided with four variants, each targeting a different demographic characteristic such as the user's gender, income, geographic location, or any other desired characteristic. As described above, providing such targeted advertisements provides a more enjoyable viewing experience for the user and thus increases the likelihood that the user will see the advertisement without changing or leaving the channel.
It is important to point out that the targeted advertising methods enabled by the present invention are different from providing advertising messages by programming guides. The present invention makes it possible to show advertisements in a seemingly conventional linear fashion, and the broadcast of a television program is periodically paused by the headend to broadcast the advertisement. The advertisements shown are targeted to individual users watching a particular TV set, but the user is not prompted to select alternative advertisements in any way, and the set-top box is in any other way You will not be informed of the fact that you are receiving alternative advertising. This is a technique for providing targeted advertisements that is different from providing targeted advertisements by a programming guide. Because in conventional programming guide situations, the TV screen is typically divided into at least two different areas, one area showing the program listing while the other area displaying advertisements. Thus, in the program guide situation, the user receives no program organization at all and receives only program listings and advertisements.
As shown in FIG. 44, the alternative scenes and advertisements preferably occupy the same time frame within the program timeline by being the same length. However, alternative length advertisements and scenes such as those included in television program 870 can also be provided. In this embodiment, the set-top box calculates the length of each scene and advertisement and additionally acts as a scheduler by scheduling the scene and advertisement, thereby allowing the user to have a seamless or non-overlapping seam. Show no apparently linear programs. In this example, both scenes and advertisements may include any combination of broadcast and previously recorded stored scenes and advertisements. For example, in the program 870, the alternative scene 2a may be shorter than the alternative scene 2, so that it is necessary to show the scenes 3b and Ad1a at an earlier time. In this case, an additional AdA may be indicated between Ad1a and Ad2c, where AdA is that the target profile matches the user's demographic profile and that is an appropriate amount of time. From the set-top box.
Providing customized programs also allows reaching a wider audience by providing more specific to specific features and characteristics of the general audience population. In addition, by providing such customized content, program producers can explore different variants of the same concept (storyline) and edit different scenes for different audiences, which is much greater for them. Artistic and expressive freedom may be possible. Another possible use of customized linear programming enabled by the present invention is to provide highly personalized services such as news, weather, stock market quotes, shopping events, instructional videos, and the like.
The popularity and acceptance of the system of the present invention is highly dependent on the cost to the end user or viewer. Thus, in one preferred financial arrangement, the user pays a set fee for the control system or hardware, connects the control system to his TV and incoming cable or satellite antenna line, and beyond Enjoy the personalized services offered by the control system without paying costs. The system administrator provides tagged advertisements for broadcasting over the broadcast network and charges the advertising client for a fee. The actual advertisement may be provided by the advertising content supplier or the advertising client itself (eg, the truck manufacturer), and the system administrator tags the metadata. The fee may be based on the total number of target viewers that the installed control system and / or advertising client wants to reach. Thus, when a client wishes to broadcast an advertisement that is tagged to reach a relatively wide range of viewers who have purchased and installed the control system of the present invention, the fee charged is more narrowly focused. It is relatively higher than the fee charged for the combined advertisement.
Although the invention has been described in accordance with the requirements of patent law, those skilled in the art will understand how to modify and change the invention to meet their specific needs or conditions. Such modifications and changes can be made without departing from the scope and spirit of the invention as set forth in the appended claims.

The claims at the time of filing the original application (Japanese Patent Application No. 2001-52064) are described below.
A method for delivering a customized linear video programming to each of a plurality of individual viewers comprising:
Processing information indicating the preferences of each of a plurality of viewers to develop viewer feature information for each viewer;
Constituting a set of video programming segments for each viewer, at least one of the video programming segments being selected from a plurality of available video programming segments for linear delivery to the viewer according to viewer feature information Generating an apparently linear program.
[Claim 2] The step of configuring a set of video programming segments includes:
The method of claim 1, further comprising generating an apparently linear program by selecting at least one broadcast video programming segment for linear delivery to a viewer simultaneously with the broadcast of the video segment.
[Claim 3] The step of configuring a set of video programming segments includes:
The method of claim 1, further comprising selecting at least one video programming segment stored on the storage medium for linear delivery to a viewer to generate an apparently linear program.
[Claim 4] The step of processing information indicating each preference of a plurality of viewers includes:
The method of claim 1, comprising processing information indicating a television program viewing preference of each of a plurality of viewers.
[Claim 5] The step of processing information indicating each television program viewing preference of a plurality of viewers includes:
5. The method of claim 4, comprising processing information indicative of a television program that each of a plurality of viewers watches.
[Claim 6] The step of processing information indicating each television program viewing preference of a plurality of viewers includes:
5. The method of claim 4, comprising processing information indicative of a television program recorded by each of a plurality of viewers.
[Claim 7] The step of processing information indicating each television program viewing preference of a plurality of viewers includes:
5. The method of claim 4, comprising processing information indicative of a television program that each of the plurality of viewers does not watch.
[Claim 8] The step of processing information indicating each TV program viewing preference of a plurality of viewers includes:
5. The method of claim 4, comprising processing information indicating television program guide information requested by each of a plurality of viewers.
[Claim 9] The step of processing information indicating each television program viewing preference of a plurality of viewers includes:
5. The method of claim 4, comprising processing information indicative of television program guide information not required by each of a plurality of viewers.
[Claim 10] The step of processing the information includes:
10. A method according to any of claims 5 to 9, comprising the step of processing electronic program guide information.
[Claim 11] The step of processing information indicating each preference of a plurality of viewers includes:
The method of claim 1, comprising processing information indicative of a preference of each of a plurality of viewers provided by each viewer in response to an inquiry.
[Claim 12] The step of processing information indicating each TV program viewing preference of a plurality of viewers includes:
The method of claim 1, comprising processing information indicative of preferences other than television program viewing preferences of each of a plurality of viewers.
[Claim 13] The step of processing information indicating a preference other than a television program viewing preference of each of a plurality of viewers includes:
The method of claim 12, comprising processing information indicative of a music preference for each of a plurality of viewers.
[Claim 14] The step of processing information indicating a preference other than the television program viewing preference of each of a plurality of viewers includes:
The method of claim 12, comprising processing information indicating a reading preference of each of a plurality of viewers.
[Claim 15] The step of processing information indicating a preference other than the television program viewing preference of each of a plurality of viewers includes:
The method of claim 12, comprising processing information indicative of shopping preferences for each of a plurality of viewers.
[Claim 16] The step of processing information indicating a preference other than the television program viewing preference of each of a plurality of viewers includes:
Demonstrate non-viewing preferences of each of multiple viewers, taken from a group of sources including online music clubs, online book clubs, online hobby clubs and organizations, and online retailers and traders The method of claim 12, comprising processing the information.
[Claim 17] The step of processing information indicating the preferences of each of a plurality of viewers to develop viewer characteristic information for each viewer comprises:
13. A method according to any of claims 1, 4, 11 or 12, comprising the step of processing information indicative of the preferences of each of a plurality of viewers to develop a television program viewing preference profile for each viewer.
[Claim 18] The step of processing information indicating each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer comprises:
18. Processing information indicating each preference of a plurality of viewers according to a prediction model based on a viewing habit of a representative sample of the general population, and developing a television program viewing preference profile for each viewer. The method described.
19. Processing the television program viewing preference profile developed for each viewer according to a probability model based on the viewing habits of a sample of the general population,
The method of claim 18, comprising calculating a maximum posterior probability value for the parameters of the prediction model by constructing a Bayesian network and predicting television program viewing preferences for each viewer.
[Claim 20] The step of processing information indicating each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer comprises:
The method of claim 17, comprising deducing each viewer's hidden features by processing information indicative of the preferences of each of the plurality of viewers.
[Claim 21] The step of processing information indicating each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer comprises:
The method of claim 17, comprising deducting each viewer's associated features by processing information indicative of the preferences of each of the plurality of viewers.
22. The method of claim 17, further comprising processing a television program viewing preference profile developed for each viewer to develop demographic information for each viewer.
23. The step of developing a television program viewing preference profile developed for each viewer to develop demographic information for each viewer comprises:
23. Processing a television program viewing preference profile developed for each viewer according to a prediction model based on viewing habits of representative samples of the general population to develop demographic information for each viewer. The method described in 1.
24. Processing the television program viewing preference profile developed for each viewer according to a prediction model based on viewing habits of representative samples of the general population,
For each viewer according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for a television program that attracts viewers that show a higher proportion of demographic features than shown by a representative sample of the general population 24. The method of claim 23, comprising processing a developed television program viewing preference profile.
25. A predictive model for predicting each demographic feature of a viewer based on viewer preference for a television program that attracts viewers that exhibit a higher proportion of demographic features than shown by a representative sample of the general population Processing the TV program viewing preference profile developed for each viewer according to:
Audiences that attract viewers with a higher percentage of demographic features than shown by a representative sample of the general population and that show a higher percentage of other demographic features than shown by a representative sample of the general population TV developed for each viewer according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for television programs that exhibit minimal demographic feature correlation with other television programs that attract viewers 25. The method of claim 24, further comprising processing a program viewing preference profile.
[Claim 26] The step of processing information indicating the preferences of each of a plurality of viewers to develop viewer characteristic information for each viewer includes:
The method of claim 1, comprising processing information indicative of the preferences of each of a plurality of viewers to develop demographic information for each viewer.
[Claim 27] The step of processing information indicating the preferences of each of a plurality of viewers to develop demographic information for each viewer comprises:
27. The method of claim 26, comprising: processing information indicative of each viewer preference according to a prediction model based on viewing habits of a representative sample of the general population to develop demographic information for each viewer.
[28] The step of processing information indicating each viewer's preference according to a prediction model based on a viewing habit of a representative sample of the general population,
Each viewer's 28. The method of claim 27, comprising processing information indicative of preferences.
29. A predictive model for predicting each demographic feature of a viewer based on viewer preference for a television program that attracts viewers that exhibit a higher proportion of demographic features than shown by a representative sample of the general population Processing the information indicating the preferences of each viewer according to
Audiences that attract viewers with a higher percentage of demographic features than shown by a representative sample of the general population and that show a higher percentage of other demographic features than shown by a representative sample of the general population Process information indicating each viewer's preference according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for television programs that exhibit minimal demographic feature correlation with other TV programs that attract viewers 30. The method of claim 28, further comprising:
[30] The step of developing information about each preference of a plurality of viewers to develop demographic information for each viewer comprises:
27. The method of claim 26, further comprising developing a television program viewing preference profile for each viewer according to viewer demographic information.
[Claim 31] The step of processing information indicating each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer comprises:
31. A method according to claim 17 or 30, further comprising the step of processing information indicative of each preference of a plurality of viewers to develop a plurality of television program viewing preference profiles for each viewer.
[Claim 32] The step of processing information indicating each preference of a plurality of viewers to develop a plurality of television program viewing preference profiles for each viewer comprises:
Processing information indicating respective preferences of a plurality of viewers to develop a plurality of television program viewing preference profiles for each viewer, each profile being a viewer's television program in a different time zone, day of the week, or season 32. The method of claim 31, wherein the viewing preference is described.
[Claim 33] The step of processing information indicating each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer comprises:
31. A method according to claim 17 or 30, further comprising the step of processing information indicative of each preference of a plurality of viewers accessing the same television device to develop a plurality of television program viewing preference profiles for each viewer.
34. The step of constructing a set of video programming segments comprises:
Configuring a set of video programming segments for viewers accessing the same television equipment, wherein at least one of the video programming segments is selected from a plurality of available video programming segments, and the viewers according to viewer feature information 34. The method of claim 33, comprising generating an apparently linear program for linear delivery to.
35. Constructing a set of video programming segments for each viewer comprises:
Configuring a set of video programming segments for each viewer, wherein at least one of the video programming segments is selected from a plurality of available video programming segments and recorded according to a viewer's television program viewing preference profile. 31. A method according to claim 17 or 30, comprising.
36. The step of constructing a set of video programming segments comprises:
The method of claim 1, comprising configuring a set of video programming segments selected from a group of video programming segments including advertising, entertainment, news, weather, finance, sports, education, and shopping programming. .
37. The step of configuring a set of video programming segments comprises:
Forming a set of video programming segments for each viewer, wherein at least one of the video programming segments is selected from a plurality of available video programming segments and further comprising recording according to viewer characteristic information. Item 37. The method according to Item 36.
38. The step of configuring a set of video programming segments comprises:
Constitutes a set of video programming segments for each viewer, at least one of the video programming segments being selected from a plurality of available video programming segments and viewing the viewer into at least one dedicated channel according to the viewer characteristic information 38. The method of claim 36, further comprising generating an apparently linear program for linear delivery to.
39. The step of configuring a set of video programming segments comprises:
38. The method of claim 36, further comprising providing the user with a list of sets of video programming segments for the user to select from.
40. The step of configuring a set of video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer The method of claim 1, wherein the content is customized according to feature information.
41. The step of constructing a set of video programming segments includes:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer The method of claim 1, wherein the targeted information is indicated in the feature information.
42. Selecting one or more video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program comprising video programming 42. The method of claim 41, wherein the content is targeted to audience feature information by a segment provider.
43. The step of configuring a set of video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer The method of claim 17, wherein the content is customized according to a television program viewing preference profile.
44. The step of configuring a set of video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer The method of claim 17, wherein the content is targeted to a television program viewing preference profile.
45. The step of selecting one or more video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program comprising video programming 45. The method of claim 44, showing content targeted to a viewer television program viewing preference profile by a segment provider.
46. The step of selecting one or more video programming segments comprises:
Selecting one or more video programming segments from a plurality of available advertising video programming segments to generate an apparently linear program for linear delivery to a viewer, said program comprising advertising video 46. The method of claim 45, wherein the advertising content is targeted to a viewer television program viewing preference profile by a programming segment provider.
47. The step of constructing a set of video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer 27. The method of claim 26, showing content customized according to demographic information.
48. The step of configuring a set of video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to the viewer, the program comprising the viewer 27. The method of claim 26, wherein the targeted content is shown in demographic information.
49. The step of selecting one or more video programming segments comprises:
Selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program comprising video programming 49. The method of claim 48, wherein content targeted to audience demographic information by a segment provider is shown.
50. The step of selecting one or more video programming segments comprises:
Selecting one or more video programming segments from a plurality of available advertising video programming segments to generate an apparently linear program for linear delivery to a viewer, said program comprising advertising video 50. The method of claim 49, wherein the advertising content is targeted to audience demographic information by a programming segment provider.
[Claim 51] The step of configuring a set of video programming segments includes:
44. The method comprising configuring a set of video programming segments selected from the group of video programming segments including advertising, entertainment, news, weather, finance, sports, education, and shopping programming. 48. The method according to any one of 47.
52. A system for delivering customized linear video programming to each of a plurality of individual viewers comprising:
A processor for processing information indicating the preferences of each of a plurality of viewers to develop viewer feature information for each viewer;
Constituting a set of video programming segments for each viewer, at least one of the video programming segments being selected from a plurality of available video programming segments for linear delivery to the viewer according to viewer feature information A system including a programmer for generating an apparently linear program.
[Claim 53] The programmer is:
53. The system of claim 52, further comprising a programmer for selecting at least one broadcast video programming segment for linear delivery to a viewer concurrently with the broadcast of the video segment to generate an apparently linear program.
54. The programmer is:
53. The system of claim 52, further comprising a programmer for selecting at least one video programming segment stored on the storage medium for linear delivery to a viewer to generate an apparently linear program.
(55) The processor comprises:
53. The system of claim 52, including a processor for processing information indicative of each television program viewing preference of a plurality of viewers.
(56) The processor comprises:
56. The system of claim 55, including a processor for processing information indicative of a television program that each of a plurality of viewers watch.
(57) The processor comprises:
56. The system of claim 55, including a processor for processing information indicative of a television program recorded by each of a plurality of viewers.
(58) The processor comprises:
56. The system of claim 55, including a processor for processing information indicative of television programs that each of the plurality of viewers does not watch.
[Claim 59] The processor comprises:
56. The system of claim 55, including a processor for processing information indicative of television program guide information requested by each of a plurality of viewers.
60. The processor comprises:
56. The system of claim 55, including a processor for processing information indicative of television program guide information not required by each of the plurality of viewers.
(61) The processor comprises:
61. A system according to any of claims 56 to 60 including a processor for processing electronic program guide information.
(62) The processor comprises:
53. The system of claim 52, including a processor for processing information indicative of each of a plurality of viewer preferences provided by each viewer in response to an inquiry.
(63) The processor comprises:
53. The system of claim 52, including a processor for processing information indicative of preferences other than television program viewing preferences of each of a plurality of viewers.
[Claim 64] The processor comprises:
64. The system of claim 63, comprising a processor for processing information indicative of a music preference for each of a plurality of viewers.
(65) The processor comprises:
64. The system of claim 63, comprising a processor for processing information indicative of the reading preference of each of a plurality of viewers.
[Claim 66] The processor includes:
64. The system of claim 63, comprising a processor for processing information indicative of each shopping preference of a plurality of viewers.
[67] The processor comprises:
A preference other than the viewing preference of each of multiple viewers obtained from online music clubs, online book clubs, online special hobby clubs and organizations, and a group of sources including online retailers and traders. 64. The system of claim 63, comprising a processor for processing the indicated information.
68. The processor,
64. A system according to any of claims 52, 55, 62 or 63, comprising a processor for processing information indicative of each preference of a plurality of viewers to develop a television program viewing preference profile for each viewer.
69. The processor comprises:
A processor for processing information indicating a preference of each of a plurality of viewers according to a prediction model based on a viewing habit of a representative sample of the general population, and developing a television program viewing preference profile for each viewer. 68. The system according to 68.
[Claim 70] The processor comprises:
70. The system of claim 69, further comprising a Bayesian network for calculating a maximum posterior probability value for a parameter of the prediction model to predict a television program viewing preference for each viewer.
71. The processor comprises:
69. The system of claim 68, comprising a processor for deducing each viewer's hidden features by processing information indicative of the preferences of each of the plurality of viewers.
72. The processor comprises:
69. The system of claim 68, comprising a processor for deducing each viewer's associated features by processing information indicative of the preferences of each of the plurality of viewers.
73. The system of claim 68, further comprising a processor for processing the television program viewing preference profile developed for each viewer to develop demographic information for each viewer.
74. The processor comprises:
Including a processor for processing a television program viewing preference profile developed for each viewer according to a prediction model based on viewing habits of a representative sample of the general population to develop demographic information for each viewer Item 74. The system according to Item 73.
[75] The processor comprises:
For each viewer according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for a television program that attracts viewers with a higher proportion of demographic features than shown by a representative sample of the general population 75. The system of claim 74, comprising a processor for processing a developed television program viewing preference profile.
[76] The processor comprises:
Audiences that attract viewers with a higher percentage of demographic features than shown by a representative sample of the general population and that show a higher percentage of other demographic features than shown by a representative sample of the general population TV developed for each viewer according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for television programs that exhibit minimal demographic feature correlation with other television programs that attract viewers 76. The system of claim 75, including a processor for processing a program viewing preference profile.
[Claim 77] The processor comprises:
53. The system of claim 52, including a processor for processing information indicative of the preferences of each of a plurality of viewers to develop demographic information for each viewer.
78. The processor comprises:
78. The system of claim 77, comprising a processor for processing information indicative of each viewer's preferences according to a prediction model based on viewing habits of a representative sample of the general population to develop demographic information for each viewer. .
(79) The processor comprises:
Each viewer's 79. The system of claim 78, including a processor for processing information indicative of preferences.
80. The processor comprises:
Audiences that attract viewers with a higher percentage of demographic features than shown by a representative sample of the general population and that show a higher percentage of other demographic features than shown by a representative sample of the general population Process information indicating each viewer's preference according to a predictive model that predicts each demographic feature of the viewer based on the viewer's preference for television programs that exhibit minimal demographic feature correlation with other TV programs that attract viewers 80. The system of claim 79, comprising a processor for performing.
81. The processor comprises:
78. The system of claim 77, including a processor for developing a television program viewing preference profile for each viewer according to viewer demographic information.
82. The processor comprises:
82. A system according to claim 68 or 81, comprising a processor for processing information indicative of each preference of a plurality of viewers to develop a plurality of television program viewing preference profiles for each viewer.
(Claim 83) The processor comprises:
Including a processor for processing information indicative of each preference of the plurality of viewers to develop a plurality of television program viewing preference profiles for each viewer, each profile of the viewer at a different time zone, day of the week, or season The system of claim 82, wherein the system describes television program viewing preferences.
(84) The processor comprises:
82. A system according to claim 68 or 81, comprising a processor for processing information indicative of each preference of a plurality of viewers accessing the same television equipment to develop a plurality of television program viewing preference profiles for each viewer. .
(85) The programmer:
Configuring a set of video programming segments for viewers accessing the same television equipment, wherein at least one of the video programming segments is selected from a plurality of available video programming segments, and the viewers according to viewer feature information 85. The system of claim 84, comprising a programmer for generating an apparently linear program for linear delivery to.
(86) The programmer is:
Configuring a set of video programming segments for each viewer, wherein at least one of the video programming segments is selected from a plurality of available video programming segments and for recording according to a viewer's television program viewing preference profile 82. A system according to claim 68 or 81, comprising a programmer.
(87) The programmer is:
53. A programmer for configuring a set of video programming segments selected from a group of video programming segments including advertising, entertainment, news, weather, finance, sports, education, and shopping programming. System.
88. The programmer,
Comprising a set of video programming segments for each viewer, wherein at least one of the video programming segments is selected from a plurality of available video programming segments and includes a programmer for recording according to the viewer characteristic information; 90. The system of claim 87.
(89) The programmer:
Constitutes a set of video programming segments for each viewer, at least one of the video programming segments being selected from a plurality of available video programming segments and viewing the viewer into at least one dedicated channel according to the viewer characteristic information 90. The system of claim 87, comprising a programmer for generating an apparently linear program for linear delivery to.
90. The programmer,
90. The system of claim 87, comprising a programmer for providing the user with a list of sets of video programming segments for the user to select from.
(91) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 53. The system of claim 52, wherein the system shows content customized according to the person feature information.
(92) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 53. The system of claim 52, wherein the targeted feature information is indicative of the targeted content.
(93) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being a video 94. The system of claim 92, showing content targeted to audience feature information by a programming segment provider.
(94) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 69. The system of claim 68, wherein the system shows content customized according to a consumer television program viewing preference profile.
(95) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 69. The system of claim 68, showing content targeted to a consumer television program viewing preference profile.
(96) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being a video 96. The system of claim 95, showing content targeted to a viewer television program viewing preference profile by a programming segment provider.
(97) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available advertising video programming segments to generate an apparently linear program for linear delivery to a viewer, the program comprising: 99. The system of claim 96, wherein the advertising content is targeted to a viewer television program viewing preference profile by a provider of advertising video programming segments.
(98) The programmer:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 78. The system of claim 77, showing content customized according to the demographic information.
[Claim 99] The programmer is:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being viewed 78. The system of claim 77, wherein the system shows content targeted for demographic information.
[Claim 100] The programmer is:
A programmer for selecting one or more video programming segments from a plurality of available video programming segments to generate an apparently linear program for linear delivery to a viewer, the program being a video 100. The system of claim 99, showing content targeted to viewer demographic information by a programming segment provider.
[Claim 101] The programmer:
A programmer for selecting one or more video programming segments from a plurality of available advertising video programming segments to generate an apparently linear program for linear delivery to a viewer, the program comprising: 101. The system of claim 100, showing advertising content targeted to audience demographic information by a provider of advertising video programming segments.
(102) The programmer:
93. A programmer for configuring a set of video programming segments selected from a group of video programming segments including advertising, entertainment, news, weather, finance, sports, education, and shopping programming. , 94 or 98.

Claims (4)

A method for delivering a customized video program comprising a collection of video segments to each of a plurality of individual viewers, comprising:
Receiving a first video program stream, wherein each of the first video program stream and the other video program stream includes a plurality of video segments, and for at least one segment of each video program stream, A set of segments including a plurality of selectable segments is provided; and
Processing information indicating the preferences of each of the plurality of viewers to determine viewer feature information for each of the viewers;
Selecting a particular segment based on viewer feature information for each of the viewers for a segment provided with a segment set including a plurality of selectable segments;
A customized video program for a particular viewer, including segments from the first video program stream and at least one other video program stream, each segment included in the customized video program having that particular viewing Satisfying the viewer characteristic information for a viewer, wherein segments from one video program stream have content different from content from other video program streams, and in the forming step, As long as the resulting customized video program length is the same as the length of the first video program stream, the length of the individual segments is not taken into account ,
Method. Selecting at least one video segment from one of a plurality of segments and a plurality of broadcast segments stored in a storage medium, and inserting the selected video segment into a video sequence to create the customized video The method of claim 1, wherein the program is generated. A system for delivering a customized video program comprising a collection of video segments to each of a plurality of individual viewers,
A receiver for receiving a first video program stream and at least one other video program stream different from the first video program stream, wherein each video program stream includes a plurality of video segments, and each video A receiver provided with a segment set including a plurality of selectable segments for at least one segment of a program stream;
A processor that processes information indicating a preference of each of the plurality of viewers to determine viewer feature information for each of the viewers;
Means for selecting a specific segment based on viewer feature information for each of the viewers for a segment provided with a segment set including a plurality of selectable segments;
A customized video program for a particular viewer, including segments from the first video program stream and at least one other video program stream, each segment included in the customized video program having that particular viewing Means for satisfying the viewer characteristic information for a viewer, segments from one video program stream have content different from content from other video program streams, and the means for forming The system does not take into account the length of individual segments as long as the resulting customized video program length is the same as the length of the first video program stream.   Selecting at least one video programming segment from one of a plurality of segments and a plurality of broadcast segments stored in a storage medium, and inserting the selected video segment into a video sequence to create the customized video program The system of claim 3 further comprising means for generating.

Images