WO2017107453A1 - 一种视频内容推荐方法、设备和系统 - Google Patents

一种视频内容推荐方法、设备和系统 Download PDF

Info

Publication number
WO2017107453A1
WO2017107453A1 PCT/CN2016/089055 CN2016089055W WO2017107453A1 WO 2017107453 A1 WO2017107453 A1 WO 2017107453A1 CN 2016089055 W CN2016089055 W CN 2016089055W WO 2017107453 A1 WO2017107453 A1 WO 2017107453A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
viewing data
denotes
video content
matrix representing
Prior art date
Application number
PCT/CN2016/089055
Other languages
English (en)
French (fr)
Inventor
张雨薇
Original Assignee
乐视控股（北京）有限公司
乐视网信息技术（北京）股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 乐视控股（北京）有限公司, 乐视网信息技术（北京）股份有限公司 filed Critical 乐视控股（北京）有限公司
Priority to US15/250,629 priority Critical patent/US20170188102A1/en
Publication of WO2017107453A1 publication Critical patent/WO2017107453A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/258—Client or end-user data management, e.g. managing client capabilities, user preferences or demographics, processing of multiple end-users preferences to derive collaborative data
    • H04N21/25866—Management of end-user data
    • H04N21/25891—Management of end-user data being end-user preferences
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/251—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/252—Processing of multiple end-users' preferences to derive collaborative data
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/4508—Management of client data or end-user data
    • H04N21/4532—Management of client data or end-user data involving end-user characteristics, e.g. viewer profile, preferences
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies

Definitions

  • the embodiments of the present invention relate to the field of video technologies, and in particular, to a video content recommendation method, device, and system.
  • cross-border such as cross-screen, cross-platform, cross-network, cross-ecological chain, etc.
  • the recommendation system is considered to be an effective method to solve these problems. It mines the user's historical behavior, models the user's interest, and predicts the user's future behavior, thus establishing the relationship between the user and the content.
  • Video sites are also facing information overload issues. For example, YouTube currently has billions of video entries, and nearly 1,500 minutes of video is uploaded every minute, while professional video site Hulu also has nearly 200,000 high-quality videos. It is also difficult to find content that you are really interested in; it is also important to improve the user's viewing and staying time, so the recommendation system is necessary for video sites.
  • video websites are mainly divided into two types, one is a website based on user-generated content (UGC website), such as foreign YouTube and domestic Youku, potatoes, etc.; the other is a professional video content-based website, such as Hulu, Netflix and domestic Qiyi, Sina blockbuster, Sohu video, etc.
  • UGC website user-generated content
  • Netflix professional video content-based website
  • Hulu a website based on user-generated content
  • Netflix a professional video content-based website
  • Qiyi Sina blockbuster
  • Sohu video etc.
  • domestic UGC video sites such as Youku and Potato have also transformed into a model with both.
  • the content and user behavior of these two video sites are not The same, which will lead to a certain difference in the design of the corresponding recommendation system.
  • the UGC website has a large number of videos and rich content, but the quality is not good, the short video is mostly, and there is no good content data.
  • the general recommendation is based on a single video, and it will filter the video quality (such as deduplication and limit the minimum number of views); UGC content life cycle is relatively short, so the design of the recommendation system emphasizes timeliness, the latest video Recommend to the user, keep the freshness of the recommendation; at the same time, the content of the UGC website is more diverse and the user generally does not have a strong purpose, so the recommendation should be as diverse as possible and related to the user's recent behavior.
  • the content generally has good structured content data. Basically, the videos are organized in accordance with TV dramas or movies. Therefore, the recommendations are generally based on episodes rather than individual videos; the episodes are now playing or not. It is divided into the first broadcast (On-airshow) and the replayed drama (Library show). In comparison, the first broadcast is generally popular. The users have a lot of channels to know and have clear Catch-up requirements. The set is more suitable for recommendation; the length of the video is generally long and there are many videos in the episode, the user's acceptance cost is relatively high, and the recommended timing is also a problem that the professional video website needs to consider, such as weekend or holiday users are relatively idle and generally do not have the first broadcast. The drama is suitable for recommending some longer series.
  • the recommendation system design of video websites should follow some basic principles. For example, the system can give reasonable recommendations to win the trust of users; the system can respond instantly to the behavior of users; the logic of recommendation Transparent to users; encourage users to actively participate and feedback at the right time; provide sufficient information for the results of the recommendations.
  • the relevant recommendation is to recommend the relevant video of the video that the user is watching or browsing to the user, that is, "the user who likes this video still likes it", and the result shows the important information of the video (title, thumbnail, average score, type, age, Time, profile, director, actor, etc.) and provide options for scoring or not interested to collect user feedback.
  • Personalized recommendation is to infer the user's interest based on the user's historical behavior, and to recommend a list of videos that the user is most likely to be interested in.
  • personalized recommendations use all of the user's historical behavior, including scoring, viewing, subscribing, searching, tagging, sharing, and commenting, not just the current single browsing or viewing behavior, so more accurate Reflect the user's interest.
  • User interests reflected in historical behavior may also be diversified, so the general personalized recommendation display knot It will be aggregated according to relatively independent points of interest (such as type, etc.).
  • the display form of personalized recommendation we aggregate the results according to the type of recommended video, and the user can also switch the type of interest; the result has very detailed video content information and the most useful user review information; the recommended results are explained; The user can directly feedback the recommendation result “already seen” or “interested or not”, and if interested, can also specifically collect the recommendation result or directly enter the viewing page.
  • Users can create their own channels, or the recommendation system automatically creates video program channels that match the user's interests based on the user's historical behavior.
  • the user can watch the video in the channel without interruption.
  • the system continuously collects feedback from the user (like, dislike, skip, read, etc.) to adjust the recommendation list in real time, so that the user can see more and more satisfied. result.
  • the personalized recommendation system has good development and application prospects. At present, almost all large-scale e-commerce systems, such as Amazon and eBay, use various forms of recommendation systems to varying degrees. Domestically, well-known shopping sites such as wheat bags, Vanke Eslite, Kuba, and Red kids have taken the lead in selecting the most advanced percentage recommendation engine system to build a personalized recommendation service system. In an increasingly competitive environment, personalized recommendation systems can effectively retain customers and improve the service capabilities of e-commerce systems. A successful recommendation system can bring huge benefits. On the other hand, various Web sites that provide personalized services also need the support of the recommendation system. The domestic recommendation system leader percentage technology has also contributed to the personalized content recommendation of Web sites. In today's information explosion, the implementation of personality Reading is imperative.
  • the method of recommending the system can be classified according to two dimensions of data and model. From the data used, the recommendation system can be divided into collaborative filtering system, content filtering system and social filtering system; from the perspective of the model used, it can be divided into neighborhood-based model, matrix decomposition model and graph model.
  • Collaborative filtering is the most famous method in the recommendation system. It mainly analyzes the user's interest and makes recommendations to the user through the historical behavior of the user.
  • the more commonly used collaborative filtering algorithm in video websites is Item CF.
  • the basic assumption is that users will like videos that are similar to their favorite videos. Therefore, when making recommendations to this user, you need to get the list of videos he likes from the user's historical behavior, and then find the video that is most similar to the user's favorite list from the remaining videos.
  • the core of this method is how to reasonably calculate the similarity between two videos. More commonly used cosine similarity or Pearson correlation coefficient, etc., in actual use, it needs to be corrected according to the specific circumstances. It is generally believed that the Item CF algorithm is relatively simple, easy to expand, high in accuracy, can be updated in real time and can be interpreted, and can handle explicit (scoring or interest) or implicit feedback (other behaviors such as viewing), so the actual video It is used in recommended systems such as Netflix, Hulu and YouTube.
  • An important disadvantage of the collaborative filtering method is that it cannot be cold-started, that is, it cannot be recommended for newly added videos or users. It is generally necessary to mix other recommended methods (such as content filtering, etc.) to deal with this problem.
  • the basic idea of content filtering is to recommend other videos that are similar in content to the videos they liked before. For example, if the user likes to watch "Two Shots", then the content filtering system will recommend other similar content works of Guy Ritchie, such as "Stolen and Robbery”; if the user likes "Naruto", the system will recommend " Naruto Shippuden or other Japanese animations of the hot blood. So the core of content filtering is how to calculate the content similarity between two videos.
  • calculating the content similarity of a video is to extract keywords from video content (such as title, type, region, company, age, director, actor, synopsis, user tag, comment, etc.) and then determine these key The weight of the word, so that the vector model of this video is obtained, and then the similarity of the two video vector models is calculated.
  • keywords such as title, type, region, company, age, director, actor, synopsis, user tag, comment, etc.
  • social filtering is that the user's preferences may be influenced by his friends in the social network.
  • social network recommendations are getting more and more attention.
  • the video search site Clicker uses Facebook's friend relationship to make recommendations; of course, another advantage of using Facebook is that video sites can get more information from users. In particular, some of the Like information outside the station will also help improve the quality of the recommendations.
  • a complete recommendation system includes at least the log system, recommendation engine and display interface design.
  • the logging system primarily collects user behavior and feedback on the recommendation system.
  • the recommendation engine also separates the line and the online part: the offline system is mainly responsible for generating the video correlation matrix, which is stored in the database for real-time query and invocation by the online system; the online system is responsible for responding to the user's request in real time, extracting and analyzing the user behavior online and generating the final Recommended results.
  • the offline part of the recommendation engine utilizes the collected user behavior log: calculates a series of association matrices (such as the two-two similarity between videos, the association between movie themes and videos, etc.); calculates pairs of global or certain groups of users.
  • Recommend system feedback (such as the weight of user behavior, the weight of the recommendation algorithm, etc.).
  • the embodiment of the present invention provides a video content recommendation method, device, and system, which are used to solve the defect that the user cannot perform fine video content recommendation on different channels in the prior art.
  • An aspect of the present application provides a video content recommendation method, including: analyzing user history viewing data to obtain various personalized preference parameters; and performing user history viewing data according to various personalized preference parameters and user individual characteristic information. Organizing and intersecting, obtaining grouped viewing data according to personalized preference parameters and user individual characteristic information; grouping according to an independent variable matrix representing the historical viewing data, an independent variable matrix representing the individual characteristic information of the user, and a dependent variable matrix pair representing the group viewing data Viewing data for processing; based on the processing result, the channel corresponding to each video content is obtained as an independent coefficient; the corresponding coefficient obtained by using the channel as an independent variable is converted into a recommended weight coefficient Wi; according to different weight coefficients Wi Make recommendations for the corresponding video content.
  • a video content recommendation device including: a processor, a transmitter, and a receiver;
  • a receiver configured to receive user viewing data
  • the processor is configured to analyze the user historical viewing data to obtain various personalized preference parameters; and to perform the user historical viewing data according to various personalized preference parameters and individual characteristic information of the user. And intersecting, obtaining grouped viewing data according to personalized preference parameters and user individual characteristic information; grouping according to an independent variable matrix representing the historical viewing data, an independent variable matrix representing the individual characteristic information of the user, and a dependent variable matrix pair representing the group viewing data Viewing data for processing; based on the processing result, the channel corresponding to each video content is obtained as an independent coefficient; the corresponding coefficient obtained by using the channel as an independent variable is converted into a recommended weight coefficient Wi; according to different weight coefficients Wi Determining the corresponding recommended video content;
  • a sender configured to send recommended video content.
  • Yet another aspect of the present application provides a video content recommendation system including a network transmission system and the above-described video content recommendation device.
  • the video content recommendation method, device and system provided by the embodiments of the present application can perform fine video content recommendation on different channels.
  • the embodiment of the present application provides a computer readable recording medium on which a program for executing the above method is recorded.
  • FIG. 1 is a schematic structural diagram of a digital cable television broadcasting system according to an embodiment of the present application.
  • FIG. 2 is a schematic structural diagram of a UMTS communication system according to another embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of a video content recommendation device according to another embodiment of the present application.
  • FIG. 4 is a schematic flowchart diagram of a video content recommendation method according to another embodiment of the present application.
  • the video content recommendation methods and devices described herein may be configured into various video systems, such as cable television systems, video websites, e-commerce websites, and the like.
  • the video content recommendation method and apparatus described herein can be implemented by various terminals, such as computers, television dramas, and wireless terminals having video playback systems.
  • the wireless terminal having the video playing system may be a device that provides voice and data connectivity to the user, a handheld device with wireless connection function, or other processing device connected to the wireless modem.
  • a wireless terminal having a video playback system can communicate with one or more core networks via a radio access network (eg, RAN, Radio Access Network), and the wireless terminal having a video playback system can be a mobile terminal, such as a mobile phone and having a mobile
  • the computer of the terminal for example, can be a portable, pocket-sized, handheld, computer-integrated or in-vehicle mobile device with a video playback system that exchanges language and/or data with the wireless access network.
  • a personal communication service with a video playback system
  • a cordless phone for example, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a personal digital assistant (PDA, Personal) Digital Assistant
  • SIP Session Initiation Protocol
  • WLL Wireless Local Loop
  • PDA Personal digital assistant
  • a wireless terminal having a video playback system may also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile station, a remote station, an access point, a remote terminal, an access terminal, a user terminal, a user agent, a user equipment, or a user equipment.
  • system and “network” are used interchangeably herein.
  • the term “and/or” in this context is merely an association describing the associated object, indicating that there may be three relationships, for example, A and / or B, which may indicate that A exists separately, and both A and B exist, respectively. B these three situations.
  • the character "/" in this article generally indicates that the contextual object is an "or" relationship.
  • the video content recommendation device of the embodiment of the present application is a video content providing device or a video content providing server, and the video content providing device or the video content providing server provides the video content to the user terminal through a digital cable television broadcasting system or a wireless network communication system.
  • the user terminal can be a computer, a PDA, a television, a mobile phone TV, a car mobile TV, and the like.
  • Analog TV has three standards: NTSC, PAL and SECAM.
  • digital television forms three different digital television standards in the United States, Europe, and Japan.
  • the standard in the United States is the Advanced Television System Committee (ATSC) standard; the European standard is the Digital Video Broadcasting (DVB) standard; the Japanese standard is Integrated Services Digital Broadcasting (ISDB). standard.
  • ATSC Advanced Television System Committee
  • DMB digital multimedia mobile broadcasting
  • ISDB Integrated Services Digital Broadcasting
  • China has also developed relevant standards: China Mobile Multimedia Broadcasting (CMMB) standard, digital multimedia mobile broadcasting (DMB, Digital Multimedia) Broadcasting) standard.
  • CMMB China Mobile Multimedia Broadcasting
  • DMB digital multimedia mobile broadcasting
  • the DVB transmission system involves all transmission media such as satellite, cable TV, terrestrial, SMATV, MMDS. Their corresponding DVB standards are: Digital Satellite Broadcasting System Standard (DVB-S), Digital Cable Broadcasting System Standard (DVB-C), Digital Terrestrial Television Broadcasting System Standard (DV
  • a schematic diagram of a digital cable television broadcasting system includes a video content providing server 10, a front end system 11, a network system 12, and a user terminal 13, wherein
  • the video content providing server 10 is configured to provide video content
  • the front end system 11 is the core of the entire digital cable television system
  • the network system 12 is the basic platform of the system
  • the user terminal 13 is the final result.
  • the front-end system 11 is an information source and a switching center of a cable television network, and is generally composed of a digital satellite receiver, a video server, a codec, a multiplexer, a QAM modulator, various management servers, and a control network portion.
  • the digital TV front end system 11 can generally be divided into four main parts: a signal input part, a signal processing part, a signal output part, and a system management part, each of which has its specific function, and finally constitutes a complete digital TV front end system.
  • the input part receives various programs from different networks, such as satellites, open channels, and the like, and some local analog TV programs are encoded and compressed, and formed by a video server, and the received signals are converted into a unified format.
  • Signal processing part receives various programs from different networks, such as satellites, open channels, and the like, and some local analog TV programs are encoded and compressed, and formed by a video server, and the received signals are converted into a unified format.
  • Signal processing part is also be used to convert signals into a unified format.
  • the signal processing part includes: descrambling, multiplexing, SI processing, etc., which is the core of the digital front end.
  • the main part of the signal processing is to perform descrambling, interception, multiplexing and other processing on all programs.
  • Service information is updated at any time to ensure proper operation of the set-top box and all application data is properly inserted.
  • the management of the signal processing part requires an integrated management system.
  • ASI Asynchronous Serial Interface
  • ASI Asynchronous Serial Interface
  • the signal output portion After the signal output portion receives the information that has been processed by the signal processing portion, it becomes the signal format required for the transmission network, and a typical 64QAM modulator is used for the cable television network. In the use of the modulator, it is very important to debug the setting of the output level and frequency.
  • the various management servers in the system management part mainly complete some user information management and billing work, as well as the management and security of video materials.
  • the control network part mainly completes various information transfer work in various servers and exchange of film and television materials and data in the background.
  • the network portion 12 includes various optical transmitters.
  • the topology of the various optical transmitters generally has a star structure, a tree structure, and a star tree type hybrid structure, and a two-stage optical link cascaded double star structure.
  • the user terminal 13 may be composed of a digital set top box (STB) and a display, or a network terminal and a display, and utilizes a cable television network as a transmission platform, so that the user enjoys all-round information services such as digital television and data broadcasting.
  • STB digital set top box
  • the user terminal 13 may be composed of a digital set top box (STB) and a display, or a network terminal and a display, and utilizes a cable television network as a transmission platform, so that the user enjoys all-round information services such as digital television and data broadcasting.
  • STB digital set top box
  • the video content providing device or the video content providing server can be combined not only with the digital cable television broadcasting system, but also with various communication systems for providing users with all-round information services such as digital television and data broadcasting.
  • GSM Global System for Mobile Communications
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • WCDMA Wideband Code Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDM Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • GPRS General Packet Radio Service
  • UMTS Universal Mobile Telecommunications
  • LTE Long Term Evolution
  • the UMTS communication system includes: an access network and a core network that communicate with each other, wherein the access network includes a plurality of base stations 21 and a plurality of radio network controllers 22, and the core network is divided into a circuit domain (CS domain) and a packet domain (PS domain).
  • the CS domain is mainly a voice service, and is composed of an interconnected Mobile Switching Center (MSC) server and a Media Gateway (MGW), wherein the MSC server includes an interconnected access mobile switching center (VMSC, Visited).
  • MSC Mobile Switching Center
  • MGW Media Gateway
  • the PS domain is mainly a mobile data service, and is mainly composed of an interconnected Serving GPRS Support Node (SGSN) 26 and a Gateway GPRS Support Node (GGSN) 27.
  • the MGW 25 is also connected to a PSTN (Public Switched Telephone Network) 28 or the like.
  • the GGSN 27 is connected to the video content providing server 20 via the Internet 29.
  • the video content providing server 20 is configured to provide video content and is provided by the UMTS communication system.
  • the terminal provides video content.
  • the video content providing servers 10 and 20 have the same structure, and the video content providing servers 10 and 20 can also serve as a video content recommending device, for example, as shown in FIG. 3, which is a video content in another embodiment of the present application.
  • FIG. 3 is a video content in another embodiment of the present application.
  • the video content recommendation device includes a processor 301, a transmitter 302, a receiver 303, a communication interface 304, a memory 305, and a communication bus 306; wherein the processor 301 and the transmitter 302 Receiver 303, communication interface 304, and memory 305 complete communication with one another via communication bus 306.
  • the processor 301 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.
  • CPU central processing unit
  • ASIC Application Specific Integrated Circuit
  • the memory 305 is configured to store program code, the program code including computer operating instructions.
  • the memory 305 may include a high speed RAM memory and may also include a non-volatile memory such as at least one disk memory.
  • Communication interface 304 is configured to enable connection communication between these devices.
  • the receiver 303 is configured to receive user viewing data.
  • the processor 301 is configured to execute program code, configured to analyze user historical viewing data, to obtain various personalized preference parameters, and to organize and cross the user historical viewing data according to various personalized preference parameters and individual characteristic information of the user. Obtaining grouped viewing data according to personalized preference parameters and user individual characteristic information; processing grouped viewing data according to an independent variable matrix representing historical viewing data, an independent variable matrix representing user individual characteristic information, and a dependent variable matrix representing group viewing data And based on the processing result, the corresponding channel of each video content is obtained as an independent coefficient; the corresponding coefficient is obtained by using the channel as an independent variable to a recommended weight coefficient Wi; and the corresponding recommendation is determined according to different weight coefficients Wi; Video content.
  • the sender 302 is configured to send the recommended video content.
  • the processor 301 is configured to process the group viewing data according to the argument matrix representing the historical viewing data, the argument matrix representing the individual characteristic information of the user, and the dependent variable matrix representing the group viewing data. Specifically, the processor 301 is configured to represent the history. The independent variable matrix of the viewing data, the independent variable matrix representing the individual characteristic information of the user, and the dependent variable matrix representing the group viewing data are placed in the mixture The effect model formula (1) is combined, and the group viewing data is calculated according to the mixed effect model formula (1).
  • the sum of the weighting factor Wi sequences is 100%.
  • the mixed effect model referred to as the hybrid model, also known as the mixed variance component model, the variance analysis model III, includes both the fixed effect model and the statistical model of the random effect.
  • the fixed effect is similar to the standard regression coefficient and is directly estimated by historical data regression.
  • Random effects are not direct estimates (although they may be taken from post hoc estimates), but are summed up from their variance and covariance estimates. Random effects are presented in the form of random intercepts or random coefficients, and the organization of the data may include multiple levels of nested packets. Thus, in the literature, the mixed effects model is also referred to as a multilevel model and a layered model. The mixed effect command of the mixed effect model used to fit the variation of the reaction distribution is conditioned on a random effect that conforms to the normal distribution.
  • the complexity and comprehensiveness of the mixed-effects model can effectively capture the stable characteristics of the data (fixed effects) and the characteristics of random changes (machine effects).
  • the processor 301 performs data calculation using the following mixed effect model formula (1).
  • ⁇ denotes a fixed effect and b i denotes a random effect, wherein ⁇ and b i are coefficients corresponding to the channel as an independent variable;
  • Y i represents a matrix of dependent variables, representing group viewing data, for example, for a user Video group (to be subdivided into different types of attribute videos);
  • X i represents an independent variable matrix, representing individual user characteristic information, such as user's individual characteristic information (eg, age, gender, income, etc.);
  • ⁇ i represents error term Matrix, which is a built-in hybrid effect model, does not need to be manually defined;
  • Z i represents another set of attributes that differ from X i 's independent variable matrix, such as user history viewing data;
  • K i represents a weight coefficient, after a series of pre-sets Satisfies
  • the mixed effect model formula (1) also needs to meet the following requirements:
  • ⁇ i ⁇ N(0, ⁇ i ) denotes that ⁇ obeys the standard normal distribution, where ⁇ i ⁇ N(0, ⁇ i ) represents the corresponding standard positive distribution, and ⁇ i represents the addition operation.
  • cov denotes a covariance
  • model result of the mixed effect model formula (1) is also defined by the density equation of the dependent variable Yi, defined as equation (2):
  • f(y i ) denotes the density equation expression symbol
  • y i denotes the element in the dependent variable Yi
  • b i ) denotes the density equation for expressing f(y) with b
  • f(b i ) Represents the density equation for b
  • d denotes the differential symbol.
  • FIG. 4 is a video content recommendation according to another embodiment of the present application.
  • a video content providing server that is, a video content recommendation device
  • FIG. 4 is a video content recommendation according to another embodiment of the present application.
  • step 401 the user history viewing data is analyzed to obtain various personalized preference parameters.
  • the video content providing server analyzes the user history viewing data, obtains the type of video that the user prefers, and the time distribution of various types of favorite viewing.
  • Step 402 Organize and cross the user history viewing data according to various personalized preference parameters and user individual characteristic information, and obtain group viewing data according to the personalized preference parameter and the user individual characteristic information.
  • the video content providing server sorts and crosses the user historical viewing data according to the obtained personalized preference parameters, and combines the user's individual characteristic information to obtain a corresponding grouping.
  • the user individual characteristic information includes the user's age, income, and social platform activity.
  • Historical data such as viewing time and matching video type/length, key element data (such as actors, languages, video stream versions, etc.) that are biased by users of different ages and even education levels.
  • Step 403 the group viewing data is processed according to an independent variable matrix representing the historical viewing data, an independent variable matrix representing the individual characteristic information of the user, and a dependent variable matrix representing the group viewing data.
  • an independent variable matrix representing historical viewing data, an independent variable matrix representing user individual characteristic information, and a dependent variable matrix representing group viewing data are placed in the mixed effect model formula (1), and The group viewing data is calculated according to the mixed effect model formula (1), wherein the mixed effect model formula (1) refers to the foregoing.
  • Step 404 Based on the processing result, the channel corresponding to each video content is used as an independent variable to obtain a corresponding coefficient.
  • the corresponding channel of each video content is obtained as an independent variable, and the corresponding coefficient of the channel as an independent variable is ⁇ and b in the mixed effect model formula (1). i , where ⁇ represents a fixed effect and b i represents a random effect.
  • step 405 the corresponding coefficient obtained by using the channel as an independent variable is converted into the recommended weight coefficient Wi.
  • Step 406 Perform recommendation of the corresponding video content according to different weight coefficients Wi.
  • the operational details and rules of steps 402-405 are independent. Grouping in accordance with industry rules based on different user characteristics and content library content. Finally, according to the custom mapping rule, the weight coefficient Wi sequence with the sum of 100% is converted into a real-time video content combination recommendation according to the weight value Wi.
  • the present application also provides a computer readable recording medium on which a program for executing the above method is recorded.
  • the computer readable recording medium includes any mechanism for storing or transmitting information in a form readable by a computer (eg, a computer).
  • a machine-readable medium includes read only memory (ROM), random access memory (RAM), magnetic disk storage media, optical storage media, flash storage media, electrical, optical, acoustic, or other forms of propagation signals (eg, carrier waves) , infrared signals, digital signals, etc.).
  • the disclosed system, apparatus, and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative
  • the division of the module or unit is only a logical function division, and the actual implementation may have another division manner, for example, multiple units or components may be combined or integrated into another system, or some Features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, ie may be located in one Places, or they can be distributed to multiple network elements. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the integrated unit if implemented in the form of a software functional unit and sold or used as a standalone product, may be stored in a computer readable storage medium.
  • a computer readable storage medium A number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) or a processor to perform all or part of the steps of the methods described in various embodiments of the present application.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Computing Systems (AREA)
  • Computer Graphics (AREA)
  • Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)

Abstract

一种视频内容推荐方法、设备和系统，视频内容推荐设备对用户历史观看数据进行分析，得到各种个性化喜好参数；根据各种个性化喜好参数和用户个体特性信息，对用户历史观看数据进行整理和交叉，得到按照个性化喜好参数和用户个体特性信息的分组观看数据；根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理；基于处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；按一定的比例折算成推荐权重系数Wi；根据不同的权重系数Wi进行对应视频内容的推荐。通过本申请的视频内容推荐方法、设备和系统，可以对不同频道进行精细的视频内容推荐。

Description

一种视频内容推荐方法、设备和系统
本申请要求于2015年12月23日提交中国专利局、申请号为201510980566.X，发明名称为“一种视频内容推荐方法、设备和系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及视频技术领域，尤其涉及一种视频内容推荐方法、设备和系统。
背景技术
近几年，互联网、网络视频、IT、家电甚至于消费类电子产品等企业纷纷通过多种跨界(如，跨屏、跨平台、跨网络、跨生态链等)向传统广电领域渗透。
随着上述企业不断的“攻城掠地”，无论是用户群体还是其市场份额，传统广电机构终于意识到危机的来临，部分地区的广电机构“领头羊”们开始紧盯着全球科技的新趋势、新技术，也想借此机会在云计算、大数据、云存储、3D、全息影像等方面发力。
随着互联网特别是社会化网络的快速发展，我们正处于信息过载的时代。用户面对过量的信息很难找到自己真正感兴趣的内容，而内容提供商也很难把优质的内容准确推送给感兴趣的用户。推荐系统被认为是解决这些问题的有效方法，它对用户的历史行为进行挖掘，对用户兴趣进行建模，并对用户未来的行为进行预测，从而建立了用户和内容的关系。
视频网站同样也面临着信息过载的问题，比如YouTube目前有数十亿视频条目，而且每分钟都有近1500分钟时长的视频被上传，而专业视频网站Hulu也有近20万的高质量视频，用户也很难从中找到自己真正感兴趣的内容；同时提高用户的观看和停留时间也非常重要，所以推荐系统对于视频网站来说是有必要的。
目前视频网站主要分为两种，一种是以用户产生内容为主的网站(UGC网站)，比如国外的YouTube和国内的优酷、土豆等；另一种是专业视频内容为主的网站，比如国外的Hulu、Netflix和国内的奇艺、新浪大片、搜狐视频等。为了获得广告商和资本市场的青睐，国内UGC视频网站比如优酷和土豆也纷纷转型为两者兼备的模式。这两种视频网站的内容和用户行为都是不 一样的，从而会导致相应的推荐系统的设计也会有一定差别。
相比较而言，UGC网站的视频数量多，内容比较丰富，但是质量良莠不齐，以短视频居多，而且没很好的内容数据。一般的推荐是基于单个视频的，而且会对视频质量做一定过滤(比如去重和限制最少观看次等)；UGC内容生命周期比较短，所以推荐系统的设计比较强调时效性，把最新的视频推荐给用户，保持推荐的新鲜性；同时UGC网站的内容比较多样化并且用户一般没有很强的目的性，所以推荐要尽量多样化并且和用户最近的行为相关。
对于专业视频网站，内容一般有很好的结构化内容数据，基本上都是按照电视剧或者电影来统一组织视频，所以推荐一般都是基于剧集而不是单个视频；剧集按照现在是否在播放又分为首播剧(On-airshow)和重播剧(Library show)，相比较而言首播剧一般比较热门，用户获知的渠道很多并对其有明确的追看(Catch-up)需求，而过往剧集更适合推荐；视频的长度一般比较长而且剧集中有很多视频，用户的接受成本相对比较高，推荐的时机也是专业视频网站需要考虑的问题，比如周末或者节假日用户比较空闲而且一般没有首播剧，适合推荐一些较长的连续剧等。
当然除了这些不同点外，一般认为视频网站的推荐系统设计还应遵循一些基本的原则，比如系统能给出合理的推荐赢得用户的信任；系统能对用户的行为作出即时的反应；推荐的逻辑对用户透明；在适当时机鼓励用户积极参与和反馈；推荐的结果要提供足够的信息等。
从推荐产品形态上来看，目前在视频网站中使用的推荐有以下几类：相关推荐、个性化推荐以及个性化电视频道等。
相关推荐就是把用户正在观看或者浏览的视频的相关视频推荐给用户，也就是“喜欢这个视频的用户还喜欢”，结果展示出视频的重要信息(标题、缩略图、平均打分、类型、年代、时长、简介、导演、演员等)，并且提供打分或者不感兴趣的选项来收集用户反馈。
个性化推荐是根据用户所有的历史行为推断出用户的兴趣，并以此推荐用户最可能感兴趣的视频列表。与相关推荐相比，个性化推荐综合使用了用户的所有历史行为，包括打分、观看、订阅、搜索、标注标签、分享和评论等，而不仅仅是当前的单个浏览或观看行为，所以更准确反映出用户的兴趣。从历史行为中反映的用户兴趣也可能会多样化，所以一般个性化推荐展示结 果会按照相对独立的兴趣点(比如类型等)进行聚合。例如个性化推荐的展示形式：我们按照推荐视频的类型对结果进行聚合，用户也可以切换自己感兴趣的类型；结果有很详细的视频内容信息以及最有用的用户评论信息；推荐结果有解释；用户可以直接对推荐结果进行“已经看过”或者“感兴趣与否”的反馈，如果感兴趣还可以具体收藏推荐结果或者直接进入观看页面等。
用户可以创建自己的频道，或者推荐系统根据用户的历史行为自动创建符合用户兴趣的视频节目频道。用户可以不间断地观看频道里面的视频，播放的过程中系统不断收集用户对视频的反馈(喜欢、不喜欢、跳过、看完等)实时调整推荐列表，让用户看到越来越满意的结果。
个性化推荐系统具有良好的发展和应用前景。目前，几乎所有的大型电子商务系统，如Amazon、eBay等不同程度的使用了各种形式的推荐系统。国内方面，知名购物网站麦包包、凡客诚品、库巴网、红孩子等都率先选择了本土最先进的百分点推荐引擎系统构建个性化推荐服务系统。在日趋激烈的竞争环境下，个性化推荐系统能有效的保留客户，提高电子商务系统的服务能力。成功的推荐系统会带来巨大的效益。另一方面，各种提供个性化服务的Web站点也需要推荐系统的大力支持，国内推荐系统领航者百分点科技就Web站点个性化内容推荐方面也做出了贡献，在信息爆棚的今天，实施个性化阅读势在必行。
一般认为推荐系统的方法可以按照数据和模型两个维度进行分类。从使用的数据上来看，推荐系统可以分为协同过滤系统、内容过滤系统和社会化过滤系统等；从使用的模型来看可分为基于邻域的模型、矩阵分解模型和图模型等。
协同过滤是推荐系统中最著名的方法，它主要通过用户的历史行为分析出用户的兴趣并给用户做出推荐。协同过滤有很多算法，比较常见的有邻域算法(User CF和Item CF等)、矩阵分解算法(或Latent Factor Model，如RSVD和SVD++等)和图算法等。目前视频网站中比较常用的协同过滤算法是Item CF，它的基本假设就是用户会喜欢跟自己之前喜欢视频比较类似的视频。因此在给这个用户做推荐的时候，需要先从用户的历史行为中得到他喜欢的视频列表，然后从剩下的视频中找到和用户之前喜欢列表最相似的视频推荐给他。可见这个方法最核心的就是怎样合理计算两个视频的相似度，比 较常用的有余弦相似度或者皮尔逊相关系数等，实际使用的时候需要根据具体情况进行修正。一般认为Item CF算法比较简单、容易扩展，准确度比较高，能实时更新而且可以解释，可以处理显式(打分或者感兴趣)或者隐式反馈(其他如观看等行为)，所以在实际的视频推荐系统中如Netflix、Hulu和YouTube都使用了它。协同过滤方法的一个重要缺点是不能冷启动，也就是对于新加入的视频或者用户都不能做出推荐，一般需要混合其他推荐方法(比如内容过滤等)来处理这个问题。
内容过滤的基本思想是给用户推荐和他们之前喜欢的视频在内容上相似的其他视频。比如用户喜欢看《两杆大烟枪》，那么内容过滤系统就会推荐盖·里奇的其他类似内容作品如《偷拐抢骗》等；如果用户喜欢《火影忍者》，系统就会推荐《火影忍者疾风传》或者其他热血类的日本动画。因此内容过滤的核心是怎样计算两个视频之间的内容相似度。一般情况下，计算视频的内容相似度是从视频内容(比如标题、类型、地区、出品公司、年代、导演、演员、剧情简介、用户标签、评论等)中抽取出关键词，然后确定这些关键词的权重，这样得到了这个视频的向量模型，再计算两个视频向量模型的相似度。随着专家标注系统Pandora在音乐推荐领域的大获成功，目前在视频领域也出现了类似的网站比如Jinni，它定义了描述电影基因的900多个标签(类型、剧情、类别、年代、地点、心情、适合的观影人群、好评、风格、态度、画面等)，然后电影专家会给每部电影标注这些标签，从而可以得到每部电影的专家标注的向量空间并以此做出推荐。由于专家标注的工作量非常大又没有公认的收益，目前在实际中并没有大规模使用，一般视频网站还是使用比较传统的方法，结合视频的内容和用户的标签进行内容过滤。
社会化过滤的思想是用户的喜好可能会受他在社会网络中的好友影响。随着SNS网络的兴起，社会化网络的推荐越来越受到关注，比如视频搜索网站Clicker就利用Facebook的好友关系做出推荐；当然利用Facebook的另外一个好处是视频网站可以得到用户的更多信息，特别是一些站外的Like信息，这些也会帮助改善推荐质量。
视频推荐系统主要工作是从用户的历史行为中分析出用户的兴趣然后找出符合其兴趣的视频展示给用户。因此一个完整的推荐系统，至少包括日志系统、推荐引擎和展示界面设计等部分。
日志系统主要收集用户的行为和对推荐系统的反馈。推荐引擎也分离线和在线两部分：离线系统主要负责生成视频相关矩阵，存储在数据库中，供在线系统实时查询和调用；在线系统负责实时响应用户的请求，在线提取和分析用户行为并生成最终推荐结果。
推荐引擎离线部分利用收集到的用户行为日志：计算出一系列的关联矩阵(比如视频之间的两两相似度、电影主题和视频的之间关联等)；计算全局或者某些群体用户的对推荐系统反馈(比如用户行为的权重、推荐算法的权重等)。
在进行视频内容推荐时，面对的往往是在成千上万个小时的海量视频中大海捞针，传统上须要投入的人力和时间，简直让人不敢想象，也很不现实。因此，如何通过从海量视频中搜索特定目标并推荐给用户，已经成为当前视频内容推荐迫切须要解决的问题。
目前各种人工进行视频内容推荐筛选，无法对不同频道进行精细的视频内容推荐。
发明内容
本申请实施例提供一种视频内容推荐方法、设备和系统，用以解决现有技术中用户无法对不同频道进行精细的视频内容推荐的缺陷。
本申请的一方面提供一种视频内容推荐方法，包括：对用户历史观看数据进行分析，得到各种个性化喜好参数；根据各种个性化喜好参数和用户个体特性信息，对用户历史观看数据进行整理和交叉，得到按照个性化喜好参数和用户个体特性信息的分组观看数据；根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理；基于处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；按一定的比例将频道作为自变量得到对应的系数折算成推荐权重系数Wi；根据不同的权重系数Wi进行对应视频内容的推荐。
本申请的另一方面提供一种视频内容推荐设备，包括：处理器、发送器、接收器；
接收器，配置为接收用户观看数据；
处理器配置为对用户历史观看数据进行分析，得到各种个性化喜好参数；根据各种个性化喜好参数和用户个体特性信息，对用户历史观看数据进行整 理和交叉，得到按照个性化喜好参数和用户个体特性信息的分组观看数据；根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理；基于处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；按一定的比例将频道作为自变量得到对应的系数折算成推荐权重系数Wi；根据不同的权重系数Wi确定对应的推荐视频内容；
发送器，配置为发送推荐视频内容。
本申请的再另一方面提供一种视频内容推荐系统，包括网络传输系统和上述的视频内容推荐设备。
本申请实施例提供的一种视频内容推荐方法，设备和系统，可以对不同频道进行精细的视频内容推荐。
本申请实施例提供一种在其上记录有用于执行上述方法的程序的计算机可读记录介质。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请一实施例的一种数字有线电视广播系统结构示意图；
图2为本申请另一实施例的一种UMTS通信系统结构示意图；
图3为本申请另一实施例中的一种视频内容推荐设备的结构示意图；
图4为本申请另一实施例的一种视频内容推荐方法的流程示意图。
具体实施方式
为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本文中描述的视频内容推荐方法及设备可配置为各种视频系统，例如有线电视系统，视频网站，电子商务网站等等。
本文中描述的视频内容推荐方法及设备可以通过各种终端实现，例如电脑，电视剧和具有视频播放系统的无线终端。
其中，具有视频播放系统的无线终端，可以是指向用户提供语音和数据连通性的设备，具有无线连接功能的手持式设备、或连接到无线调制解调器的其他处理设备。具有视频播放系统的无线终端可以经无线接入网(例如，RAN，Radio Access Network)与一个或多个核心网进行通信，具有视频播放系统的无线终端可以是移动终端，如移动电话和具有移动终端的计算机，例如，可以是便携式、袖珍式、手持式、计算机内置的或者车载的具有视频播放系统的移动装置，它们与无线接入网交换语言和/或数据。例如，具有视频播放系统的个人通信业务(PCS，Personal Communication Service)电话、无绳电话、会话发起协议(SIP)话机、无线本地环路(WLL，Wireless Local Loop)站、个人数字助理(PDA，Personal Digital Assistant)等设备。具有视频播放系统的无线终端也可以称为系统、订户单元、订户站，移动站、移动台、远程站、接入点、远程终端、接入终端、用户终端、用户代理、用户设备或用户装备。
另外，本文中术语“系统”和“网络”在本文中常被可互换使用。本文中术语“和/或”，仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，一般表示前后关联对象是一种“或”的关系。
本申请实施例的视频内容推荐设备为一种视频内容提供设备或视频内容提供服务器，所示视频内容提供设备或视频内容提供服务器通过数字有线电视广播系统或无线网络通信系统给用户终端提供视频内容，用户终端可以是电脑，PDA，电视机，移动手机电视，车载移动电视等等。
模拟电视有NTSC、PAL和SECAM三种标准。目前，数字电视在美国、欧洲和日本各自形成三种不同的数字电视标准。美国的标准是先进电视制式委员会(ATSC，Advanced Television System Committee)标准；欧洲的标准是数字视频广播(DVB，Digital Video Broadcasting)标准；日本的标准是综合业务数字广播(ISDB，Integrated Services Digital Broadcasting)标准。我国也制定了相关的标准：中国移动多媒体广播(CMMB，China Mobile Multimedia Broadcasting)标准，数字多媒体移动广播(DMB，Digital Multimedia  Broadcasting)标准。DVB传输系统涉及卫星、有线电视、地面、SMATV、MMDS等所有传输媒体。它们对应的DVB标准为：数字卫星广播系统标准(DVB-S)、数字有线电视广播系统标准(DVB-C)、数字地面电视广播系统标准(DVB-T)。
例如，如图1所示，为本申请一实施例的一种数字有线电视广播系统结构示意图，数字有线电视广播系统包括视频内容提供服务器10、前端系统11、网络系统12和用户终端13，其中，视频内容提供服务器10配置为提供视频内容，前端系统11是整个数字有线电视系统的核心，网络系统12是系统的基础平台，用户终端13是实现最终的结果。
前端系统11是有线电视网络的信息源、交换中心，一般由数字卫星接收机、视频服务器、编解码器、复用器、QAM调制器、各种管理服务器以及控制网络部分等设备组成。数字电视前端系统11一般可分为四个主要部分：信号输入部分、信号处理部分、信号输出部分和系统管理部分，每一个部分都有其特定的功能，最终组成完整的数字电视前端系统。
输入部分，接收来自不同网络的许多节目，如卫星、开路接收等各种接入方式，也有的是本地的模拟电视节目经编码压缩以及视频服务器形成的，将接收的信号转换为统一的格式送入信号处理部分。
信号处理部分包括：解扰、复用、SI处理等，它是数字前端的核心。信号处理部分主要完成的是对所有节目进行解扰、截取、复用等处理。服务信息随时更新，以保证正确地引导机顶盒的正常工作，并且所有的应用数据均能正确地插入。另外，信号处理部分的管理，须采用集成的管理系统，在所有的前端处理部分，均以异步串行接口(ASI，Asynchronous Serial Interface)作为标准接口，这样以后就能容易增加任何厂商所提供的设备，具有良好的兼容性。
信号输出部分接收信号处理部分已经处理的信息后，把它变成传输网络所需的信号格式，典型的64QAM调制器用于有线电视网。在调制器的使用中，对输出电平和频率的设置调试非常重要。
系统管理部分的各种管理服务器主要完成一些用户信息管理和计费工作，以及影视材料的管理工作和安全保密等。控制网络部分主要完成各种服务器中的各种信息传递工作及后台的影视材料和数据的交换。
网络部分12包括各种光发射机，各种光发射机组成的拓朴结构一般有星型结构、树型结构和星树型混合结构、以及两级光链路级联的双星型结构。
用户终端13，可以由数字机顶盒(STB)和显示器组成，或者由网络终端和显示器组成，利用有线电视网络作为传输平台，使用户享受数字电视、数据广播等全方位的信息服务。
视频内容提供设备或视频内容提供服务器不仅可以和数字有线电视广播系统结合，还可以和各种通信系统结合，用于向用户提供数字电视、数据广播等全方位的信息服务。
各种通信系统，例如当前2G，3G通信系统和下一代通信系统，例如全球移动通信(GSM，Global System for Mobile communication)系统，码分多址(CDMA，Code Division Multiple Access)系统，时分多址(TDMA，Time Division Multiple Access)系统，宽带码分多址(WCDMA，Wideband Code Division Multiple Access Wireless)系统，频分多址(FDMA，Frequency Division Multiple Access)系统，正交频分多址(OFDMA，Orthogonal Frequency-Division Multiple Access)系统，通用分组无线业务(GPRS，General Packet Radio Service)系统，通用移动通信UMTS(Universal Mobile Telecommunications)系统，长期演进(LTE，Long Term Evolution)系统，以及其他此类通信系统。
例如，以UMTS网络为例，如图2所示，为本申请另一实施例的一种UMTS通信系统结构示意图。UMTS通信系统包括：相互通信的接入网和核心网，其中，接入网包括多个基站21和多个无线网络控制器22，核心网分为电路域(CS domain)和分组域(PS domain)，CS域主要是话音业务，由相互连接的移动交换中心(MSC，Mobile Switching Center)服务器和媒体网关(MGW，Media Gateway)组成，其中MSC服务器包括相互连接的访问移动交换中心(VMSC，Visited Mobile-services Switching Centre)23和网关移动交换中心(GMSC，Gateway Mobile Switching Center)24。PS域主要是移动数据业务，主要由相互连接的服务GPRS支持节点(SGSN，Serving GPRS Support Node)26和网关GPRS支持节点(GGSN，Gateway GPRS Support Node)27组成。MGW25还连接(PSTN，Public Switched Telephone Network)28等。GGSN27通过因特网29连接视频内容提供服务器20。
视频内容提供服务器20用于提供视频内容，通过UMTS通信系统给用 户终端提供视频内容。
视频内容提供服务器10和20具有相同的结构，视频内容提供服务器10和20还可以作为一种视频内容推荐设备，例如，如图3所示，为本申请另一实施例中的一种视频内容提供设备的结构示意图，其具体结构和工作过程如下。
视频内容推荐设备包括处理器(processor)301、发送器(transmitter)302、接收器303、通信接口(Communications Interface)304、存储器(memory)305和通信总线306；其中，处理器301、发送器302、接收器303、通信接口304和存储器305通过通信总线306完成相互间的通信。
处理器301可能是一个中央处理器CPU，或者是特定集成电路ASIC(Application Specific Integrated Circuit)，或者是被配置成实施本申请实施例的一个或多个集成电路。
存储器305配置为存放程序代码，程序代码包括计算机操作指令。存储器305可能包含高速RAM存储器，也可能还包括非易失性存储器(non-volatile memory)，例如至少一个磁盘存储器。
通信接口304，配置为实现这些装置之间的连接通信。
接收器303，配置为接收用户观看数据。
处理器301配置为执行程序代码，配置为对用户历史观看数据进行分析，得到各种个性化喜好参数；根据各种个性化喜好参数和用户个体特性信息，对用户历史观看数据进行整理和交叉，得到按照个性化喜好参数和用户个体特性信息的分组观看数据；根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理；基于处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；按一定的比例将频道作为自变量得到对应的系数折算成推荐权重系数Wi；根据不同的权重系数Wi确定对应的推荐视频内容。
发送器302，配置为发送推荐视频内容。
处理器301配置为根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理具体包括：处理器301配置为将表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵置于混 合效应模型公式(1)，并根据混合效应模型公式(1)对分组观看数据进行运算。
在本申请的另一实施例中，所有权重系数Wi序列之和是100％。
混合效应模型，简称混合模型，亦称混合方差分量模型、方差分析模型Ⅲ，既包含固定效应(Fixed Effect)模型，又包含随机效应(Random Effect)的统计模型。
固定效应类似于标准回归系数，直接有历史数据回归估计得到。
随机效应不是直接估计(尽管它可能取自事后估计)，而是从它们的方差和协方差估计值中总结而来。随机效应以随机截距或者随机系数的形式呈现，数据的组织结构可能包括嵌套分组的多重水平。这样，在文献中，混合效应模型还被称为多水平模型和分层模型。用于拟合反应分布之变异的混合效应模型的混合效应命令以符合正态分布的随机效应为条件。
混合效应模型的复杂性和全面性能够有效捕捉数据的稳定特征(固定效应)和随机变化的特征(机效应)。
处理器301采用如下的混合效应模型公式(1)进行数据运算。
Yi＝Xiβ+Zibi+εi           公式(1)
其中，Xi＝ZiKi是已知的(ni×p)协方差矩阵。任何相关的参数都会根据实际情况定义。其中，β表示固定效应，bi表示随机效应，其中，β和bi为频道作为自变量得到对应的系数；其中Yi表示因变量矩阵，表示分组观看数据，例如对于某个用户看过的视频组(要细分到不同类型属性视频)；Xi表示自变量矩阵，表示用户个体特性信息，例如一个用户的用户个体特性信息(例如，年龄，性别，收入等)；εi表示误差项矩阵，是混合效应模型自带生成，不需要人工定义；Zi表示另一系列属性区别于Xi的自变量矩阵，例如用户历史观看数据；Ki表示一个权重系数，经过一系列事前设定满足Xi＝ZiKi；ni表示n个样本里第i个；p表示实际的矩阵运算结果反应的参数，不需要人工定义；i表示第i个样本中的序数i，为正整数，i＝1，2，3，…，i。
其中，混合效应模型公式(1)还需要满足如下要求：
bi～N(0,D)
εi～N(0,∑i)
cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0
其中，bi～N(0,D)表示b服从标准正态分布，其中N(0，D)表示标准正态分布。
εi～N(0,∑i)表示ε服从标准正态分布，其中εi～N(0,∑i)表示对应的标准正太分布，∑i表示加和运算。
cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0表示协方差矩阵，cov表示协方差。
在本申请的另一方面，混合效应模型公式(1)的模型结果还参考因变量Yi的密度方程，定义为公式(2)：
f(yi)＝∫f(yi|bi)f(bi)dbi          公式(2)
其中，f(yi)表示密度方程表达式符号，yi表示因变量Yi里的元素，f(yi|bi)表示用b来表达f(y)的密度方程，f(bi)表示b的密度方程，d表示微分符号。
本申请的另一实施例还提供一种视频内容推荐方法，由视频内容提供服务器(也即视频内容推荐设备)执行，如图4所示，为本申请另一实施例的一种视频内容推荐方法的流程示意图。
步骤401，对用户历史观看数据进行分析，得到各种个性化喜好参数。
例如，视频内容提供服务器对用户历史观看数据进行分析，得到用户偏好的视频所属类型，以及各种类型所喜欢观看的时间分布等。
步骤402，根据各种个性化喜好参数和用户个体特性信息，对用户历史观看数据进行整理和交叉，得到按照个性化喜好参数和用户个体特性信息的分组观看数据。
例如，视频内容提供服务器根据得到的各种个性化喜好参数，结合用户个体特性信息对对用户历史观看数据进行整理和交叉得到对应分组，例如，用户个体特性信息包括用户年龄、收入、社交平台活动的历史等数据，例如观看时间以及搭配的视频类型/长短，不同年龄收入甚至教育程度用户偏向的关键元素数据(例如演员，语种，视频流版本等)。
步骤403，根据表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵对分组观看数据进行处理。
例如，将表示历史观看数据的自变量矩阵、表示用户个体特性信息的自变量矩阵和表示分组观看数据的因变量矩阵置于混合效应模型公式(1)，并 根据混合效应模型公式(1)对分组观看数据进行运算，其中，混合效应模型公式(1)参考前文。
步骤404，基于处理结果，将各个视频内容对应的频道作为自变量得到对应的系数。
例如，基于混合效应模型公式(1)的运算结果，将各个视频内容对应的频道作为自变量得到对应的系数，频道作为自变量得到对应的系数为混合效应模型公式(1)中的β和bi，其中，β表示固定效应，bi表示随机效应。
步骤405，按一定的比例将频道作为自变量得到对应的系数折算成推荐权重系数Wi。
权重系数Wi是经过一些列计算方法合成的有关于Yi＝Xiβ+Zibi+εi里面β和bi的数学组合，所有权重系数Wi序列之和是100％。
步骤406，根据不同的权重系数Wi进行对应视频内容的推荐。
在上述视频内容推荐方法的运用中，步骤402-405的操作细节和制定规则都是独立。根据不同的用户特征和内容库内容进行符合行业规则的分组。最后统一按照自定的映射规则换算成加和为100％的权重系数Wi序列，按照权重数值Wi序行实时的视频内容组合推荐。
本申请还提供一种在其上记录有用于执行上述方法的程序的计算机可读记录介质。
所述计算机可读记录介质包括用于以计算机(例如计算机)可读的形式存储或传送信息的任何机制。例如，机器可读介质包括只读存储器(ROM)、随机存取存储器(RAM)、磁盘存储介质、光存储介质、闪速存储介质、电、光、声或其他形式的传播信号(例如，载波、红外信号、数字信号等)等。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。上述描述的系统，装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请所提供的几个实施例中，应该理解到，所揭露的系统，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示 意性的，例如，所述模块或单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
以上所描述的装置实施例仅仅是示意性的，所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本申请各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
所述集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (11)

  1. 一种视频内容推荐方法，其特征在于，包括：
    对用户历史观看数据进行分析，得到各种个性化喜好参数；
    根据所述各种个性化喜好参数和用户个体特性信息，对所述用户历史观看数据进行整理和交叉，得到按照所述个性化喜好参数和所述用户个体特性信息的分组观看数据；根据表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵对所述分组观看数据进行处理；
    基于所述处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；
    按一定的比例将所述频道作为自变量得到对应的所述系数折算成推荐权重系数Wi；
    根据不同的权重系数Wi进行对应视频内容的推荐。
  2. 根据权利要求1所述的方法，其特征在于，所述根据表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵对所述分组观看数据进行处理具体包括：
    将表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵置于混合效应模型公式(1)，并根据所述混合效应模型公式(1)对所述分组观看数据进行运算；
    其中，所述混合效应模型公式(1)为：
    Yi＝Xiβ+Zibi+εi
    其中，β表示固定效应，bi表示随机效应，其中，所述β和bi为所述频道作为自变量得到对应的系数；其中Yi表示因变量矩阵，表示分组观看数据；Xi表示自变量矩阵，表示所述用户个体特性信息；εi表示误差项矩阵，是混合效应模型自带生成；Zi表示另一系列属性区别于Xi的自变量矩阵，表示所述历史观看数据；Ki表示一个权重系数，经过一系列事前设定满足Xi＝ZiKi，Xi＝ZiKi是已知的(ni×p)协方差矩阵；ni表示n个样本里第i个；p表示实际的矩阵运算结果反应的参数；i表示第i个样本中的序数i，为正整数，i＝1，2，3，…，i。
  3. 根据权利要求2所述的方法，其特征在于，所述混合效应模型公式(1) 还需要满足如下要求：
    bi～N(0,D)
    εi～N(0,Σi)
    cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0
    其中，bi～N(0,D)表示b服从标准正态分布，其中N(0，D)表示标准正态分布；
    εi～N(0,Σi)表示ε服从标准正态分布，其中εi～N(0,Σi)表示对应的标准正太分布，Σi表示加和运算；
    cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0表示协方差矩阵，cov表示协方差。
  4. 根据权利要求2或3所述的方法，其特征在于，所述混合效应模型公式(1)的模型结果还参考因变量Yi的密度方程：
    f(yi)＝∫f(yi|bi)f(bi)dbi
    其中，f(yi)表示密度方程表达式符号，yi表示因变量Yi里的元素，f(yi|bi)表示用b来表达f(y)的密度方程，f(bi)表示b的密度方程，d表示微分符号。
  5. 根据权利要求1-3任意一项所述的方法，其特征在于，所有所述权重系数Wi序列之和是100％。
  6. 一种视频内容推荐设备，其特征在于，包括：处理器、发送器、接收器；
    所述接收器，配置为接收用户观看数据；
    所述处理器配置为对用户历史观看数据进行分析，得到各种个性化喜好参数；根据所述各种个性化喜好参数和用户个体特性信息，对所述用户历史观看数据进行整理和交叉，得到按照所述个性化喜好参数和所述用户个体特性信息的分组观看数据；根据表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵对所述分组观看数据进行处理；基于所述处理结果，将各个视频内容对应的频道作为自变量得到对应的系数；按一定的比例将所述频道作为自变量得到对应的所述系数折算成推荐权重系数Wi；根据不同的权重系数Wi确定对应的推荐视频内容；
    所述发送器，配置为发送所述推荐视频内容。
  7. 根据权利要求6所述的设备，其特征在于，所述处理器配置为根据表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵对所述分组观看数据进行处理具体包括：
    所述处理器配置为将表示所述历史观看数据的自变量矩阵、表示所述用户个体特性信息的自变量矩阵和表示所述分组观看数据的因变量矩阵置于混合效应模型公式(1)，并根据所述混合效应模型公式(1)对所述分组观看数据进行运算；
    其中，所述混合效应模型公式(1)为：
    Yi＝Xiβ+Zibi+εi
    其中，β表示固定效应，bi表示随机效应，其中，所述β和bi为所述频道作为自变量得到对应的系数；其中Yi表示因变量矩阵，表示分组观看数据；Xi表示自变量矩阵，表示所述用户个体特性信息；εi表示误差项矩阵，是混合效应模型自带生成；Zi表示另一系列属性区别于Xi的自变量矩阵，表示所述历史观看数据；Ki表示一个权重系数，经过一系列事前设定满足Xi＝ZiKi，Xi＝ZiKi是已知的(ni×p)协方差矩阵；ni表示n个样本里第i个；p表示实际的矩阵运算结果反应的参数；i表示第i个样本中的序数i，为正整数，i＝1，2，3，…，i。
  8. 根据权利要求7所述的设备，其特征在于，所述处理器还配置为根据所述混合效应模型公式(1)的如下要求进行运算：
    bi～N(0,D)
    εi～N(0,Σi)
    cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0
    其中，bi～N(0,D)表示b服从标准正态分布，其中N(0，D)表示标准正态分布；
    εi～N(0,Σi)表示ε服从标准正态分布，其中εi～N(0,Σi)表示对应的标准正太分布，Σi表示加和运算；
    cov(b1,b2,...,bi；ε1,ε2,...,εN)＝0表示协方差矩阵，cov表示协方差。
  9. 根据权利要求7或8所述的设备，其特征在于，所述处理器还配置为运算所述混合效应模型公式(1)的模型结果还参考因变量Yi的密度方程：
    f(yi)＝∫f(yi|bi)f(bi)dbi
    其中，f(yi)表示密度方程表达式符号，yi表示因变量Yi里的元素，f(yi|bi)表示用b来表达f(y)的密度方程，f(bi)表示b的密度方程，d表示微分符号。
  10. 一种视频内容推荐系统，其特征在于，包括网络传输系统和根据权利要求6-9任意一项所述的视频内容推荐设备。
  11. 一种在其上记录有用于执行权利要求1所述方法的程序的计算机可读记录介质。
PCT/CN2016/089055 2015-12-23 2016-07-07 一种视频内容推荐方法、设备和系统 WO2017107453A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/250,629 US20170188102A1 (en) 2015-12-23 2016-08-29 Method and electronic device for video content recommendation

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510980566.XA CN105915949A (zh) 2015-12-23 2015-12-23 一种视频内容推荐方法、设备和系统
CN201510980566.X 2015-12-23

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/250,629 Continuation US20170188102A1 (en) 2015-12-23 2016-08-29 Method and electronic device for video content recommendation

Publications (1)

Publication Number Publication Date
WO2017107453A1 true WO2017107453A1 (zh) 2017-06-29

Family

ID=56744326

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/089055 WO2017107453A1 (zh) 2015-12-23 2016-07-07 一种视频内容推荐方法、设备和系统

Country Status (2)

Country Link
CN (1) CN105915949A (zh)
WO (1) WO2017107453A1 (zh)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109819284A (zh) * 2019-02-18 2019-05-28 平安科技（深圳）有限公司 一种短视频推荐方法、装置、计算机设备及存储介质
CN109977271A (zh) * 2019-04-29 2019-07-05 华北理工大学 一种大数据处理系统及其处理方法
CN110941740A (zh) * 2019-11-08 2020-03-31 腾讯科技（深圳）有限公司 视频推荐方法及计算机可读存储介质
WO2020145702A1 (en) * 2019-01-09 2020-07-16 Samsung Electronics Co., Ltd. Temporal-based recommendations for personalized user contexts and viewing preferences
WO2020145615A1 (en) * 2019-01-07 2020-07-16 Samsung Electronics Co., Ltd. Method of providing recommendation list and display device using the same
CN112291624A (zh) * 2020-09-23 2021-01-29 咪咕视讯科技有限公司 视频处理方法、装置、电子设备及存储介质
CN112650884A (zh) * 2020-12-21 2021-04-13 四川长虹电器股份有限公司 基于深度学习的视频推送方法
CN113015010A (zh) * 2021-03-16 2021-06-22 杭州网易云音乐科技有限公司 推送参数确定方法、装置、设备及计算机可读存储介质
CN114168791A (zh) * 2021-11-24 2022-03-11 卓尔智联(武汉)研究院有限公司 视频推荐方法、装置、电子设备和存储介质
CN114615524A (zh) * 2022-02-18 2022-06-10 聚好看科技股份有限公司 服务器、媒资推荐网络的训练方法及媒资推荐方法
CN114866818A (zh) * 2022-06-17 2022-08-05 深圳壹账通智能科技有限公司 视频推荐方法、装置、计算机设备及存储介质
CN115086697A (zh) * 2022-06-08 2022-09-20 广州方硅信息技术有限公司 直播封面互动方法、系统、装置及电子设备
CN115379117A (zh) * 2022-08-22 2022-11-22 天翼数字生活科技有限公司 一种旧视频片源修复筛选方法、装置、终端及介质
CN116156263A (zh) * 2023-03-06 2023-05-23 四川长虹电器股份有限公司 一种用户实时追剧处理方法

Families Citing this family (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106484777B (zh) * 2016-09-12 2020-09-08 腾讯科技（深圳）有限公司 一种多媒体数据处理方法以及装置
CN106454431B (zh) * 2016-10-14 2017-09-05 合肥工业大学 电视节目推荐方法和系统
CN106851418B (zh) * 2017-01-24 2020-12-25 阿里巴巴（中国）有限公司 视频推荐方法及装置
CN107463853B (zh) * 2017-08-04 2020-01-21 晶赞广告（上海）有限公司 受众标签分析的方法及系统
CN107562848B (zh) * 2017-08-28 2020-07-14 广州优视网络科技有限公司 一种视频推荐方法和装置
CN107818183B (zh) * 2017-12-05 2021-07-13 云南大学 一种基于特征相似性度量的三阶段组合党建视频推荐方法
CN108419134B (zh) * 2018-02-05 2020-02-18 华南理工大学 基于个体历史与群体当前行为融合的频道推荐方法
CN108170868A (zh) * 2018-02-09 2018-06-15 宁夏灵智科技有限公司 视频推荐方法和装置
WO2019232440A1 (en) * 2018-06-01 2019-12-05 Systems And Software Enterprises, Llc Systems and methods for recommendation system based on implicit feedback
CN109257648B (zh) * 2018-09-30 2021-03-16 武汉斗鱼网络科技有限公司 一种直播间相似度修正方法、装置、终端及存储介质
CN109325146B (zh) * 2018-11-12 2024-05-07 平安科技（深圳）有限公司 一种视频推荐方法、装置、存储介质和服务器
CN109299327A (zh) * 2018-11-16 2019-02-01 广州市百果园信息技术有限公司 视频推荐方法、装置、设备及存储介质
CN110866180B (zh) * 2019-10-12 2022-07-29 平安国际智慧城市科技股份有限公司 资源推荐方法、服务器及存储介质
CN112714359B (zh) * 2019-10-24 2022-09-09 北京达佳互联信息技术有限公司 视频推荐方法、装置、计算机设备及存储介质
CN111737576B (zh) * 2020-06-22 2023-09-19 中国银行股份有限公司 应用功能个性化推荐方法和装置
CN112153424B (zh) * 2020-09-23 2024-01-26 腾讯科技（深圳）有限公司 内容推送方法、装置、电子设备和存储介质
CN112132622B (zh) * 2020-09-25 2021-07-16 北京达佳互联信息技术有限公司 数据预估方法及装置
CN112333547A (zh) * 2020-11-20 2021-02-05 广州欢网科技有限责任公司 电视端短视频用户喜爱偏好推荐方法、装置和智能电视

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2634707A1 (en) * 2012-02-29 2013-09-04 British Telecommunications Public Limited Company Recommender control system, apparatus and method
WO2013128155A1 (en) * 2012-02-29 2013-09-06 British Telecommunications Public Limited Company Recommender control system, apparatus, method and related aspects
CN103491441A (zh) * 2013-09-09 2014-01-01 东软集团股份有限公司 直播电视节目推荐方法及系统
CN103957434A (zh) * 2014-04-03 2014-07-30 三星电子（中国）研发中心 推荐节目的方法和设备

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140123161A1 (en) * 2012-10-24 2014-05-01 Bart P.E. van Coppenolle Video presentation interface with enhanced navigation features
CN102946566B (zh) * 2012-10-24 2015-07-01 北京奇虎科技有限公司 基于历史信息的视频推荐方法和装置
CN104572797A (zh) * 2014-05-12 2015-04-29 深圳市智搜信息技术有限公司 基于主题模型的个性化服务推荐系统和方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2634707A1 (en) * 2012-02-29 2013-09-04 British Telecommunications Public Limited Company Recommender control system, apparatus and method
WO2013128155A1 (en) * 2012-02-29 2013-09-06 British Telecommunications Public Limited Company Recommender control system, apparatus, method and related aspects
CN103491441A (zh) * 2013-09-09 2014-01-01 东软集团股份有限公司 直播电视节目推荐方法及系统
CN103957434A (zh) * 2014-04-03 2014-07-30 三星电子（中国）研发中心 推荐节目的方法和设备

Cited By (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020145615A1 (en) * 2019-01-07 2020-07-16 Samsung Electronics Co., Ltd. Method of providing recommendation list and display device using the same
WO2020145702A1 (en) * 2019-01-09 2020-07-16 Samsung Electronics Co., Ltd. Temporal-based recommendations for personalized user contexts and viewing preferences
US11157964B2 (en) 2019-01-09 2021-10-26 Samsung Electronics Company, Ltd. Temporal-based recommendations for personalized user contexts and viewing preferences
CN109819284B (zh) * 2019-02-18 2022-11-15 平安科技（深圳）有限公司 一种短视频推荐方法、装置、计算机设备及存储介质
CN109819284A (zh) * 2019-02-18 2019-05-28 平安科技（深圳）有限公司 一种短视频推荐方法、装置、计算机设备及存储介质
CN109977271A (zh) * 2019-04-29 2019-07-05 华北理工大学 一种大数据处理系统及其处理方法
CN109977271B (zh) * 2019-04-29 2022-12-20 重庆憨牛技术创新服务有限公司 一种大数据处理系统及其处理方法
CN110941740A (zh) * 2019-11-08 2020-03-31 腾讯科技（深圳）有限公司 视频推荐方法及计算机可读存储介质
CN110941740B (zh) * 2019-11-08 2023-07-14 深圳市雅阅科技有限公司 视频推荐方法及计算机可读存储介质
CN112291624A (zh) * 2020-09-23 2021-01-29 咪咕视讯科技有限公司 视频处理方法、装置、电子设备及存储介质
CN112291624B (zh) * 2020-09-23 2023-10-17 咪咕视讯科技有限公司 视频处理方法、装置、电子设备及存储介质
CN112650884B (zh) * 2020-12-21 2022-04-01 四川长虹电器股份有限公司 基于深度学习的视频推送方法
CN112650884A (zh) * 2020-12-21 2021-04-13 四川长虹电器股份有限公司 基于深度学习的视频推送方法
CN113015010A (zh) * 2021-03-16 2021-06-22 杭州网易云音乐科技有限公司 推送参数确定方法、装置、设备及计算机可读存储介质
CN114168791A (zh) * 2021-11-24 2022-03-11 卓尔智联(武汉)研究院有限公司 视频推荐方法、装置、电子设备和存储介质
CN114615524A (zh) * 2022-02-18 2022-06-10 聚好看科技股份有限公司 服务器、媒资推荐网络的训练方法及媒资推荐方法
CN114615524B (zh) * 2022-02-18 2023-10-24 聚好看科技股份有限公司 服务器、媒资推荐网络的训练方法及媒资推荐方法
CN115086697A (zh) * 2022-06-08 2022-09-20 广州方硅信息技术有限公司 直播封面互动方法、系统、装置及电子设备
CN114866818A (zh) * 2022-06-17 2022-08-05 深圳壹账通智能科技有限公司 视频推荐方法、装置、计算机设备及存储介质
CN114866818B (zh) * 2022-06-17 2024-04-26 深圳壹账通智能科技有限公司 视频推荐方法、装置、计算机设备及存储介质
CN115379117A (zh) * 2022-08-22 2022-11-22 天翼数字生活科技有限公司 一种旧视频片源修复筛选方法、装置、终端及介质
CN115379117B (zh) * 2022-08-22 2023-05-23 天翼数字生活科技有限公司 一种旧视频片源修复筛选方法、装置、终端及介质
CN116156263A (zh) * 2023-03-06 2023-05-23 四川长虹电器股份有限公司 一种用户实时追剧处理方法

Also Published As

Publication number Publication date
CN105915949A (zh) 2016-08-31

Similar Documents

Publication Publication Date Title
WO2017107453A1 (zh) 一种视频内容推荐方法、设备和系统
US20170188102A1 (en) Method and electronic device for video content recommendation
US11588767B2 (en) System and interface that facilitate selecting videos to share in a messaging application
US10262045B2 (en) Application representation for application editions
JP6855595B2 (ja) ライブストリームコンテンツを推奨するための機械学習の使用
US11741178B2 (en) Managing access rights to content using social media
US20120317085A1 (en) Systems and methods for transmitting content metadata from multiple data records
US10390085B2 (en) Video channel categorization schema
US11558480B2 (en) Tracking content use via social media
US20170094364A1 (en) Social news aggregation and distribution
US9560159B1 (en) Recommending media content to a user based on information associated with a referral source
US9305215B2 (en) Apparatus, method and computer readable recording medium for analyzing video using image captured from video
KR20120020949A (ko) 방송 선호 정보를 제공하기 위한 방법 및 장치
EP3543945A1 (en) Method and device for generating aggregated content, and medium
Tsunoda et al. Automatic metadata expansion and indirect collaborative filtering for TV program recommendation system
US20150294331A1 (en) Peer-to-peer data collector and analyzer
US20160021051A1 (en) Providing Access To Content Via Social Media
WO2018001223A1 (zh) 播放列表推荐方法及装置
CN114173200B (zh) 基于广电专网的视频管理推送方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16877265

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16877265

Country of ref document: EP

Kind code of ref document: A1