Search

CN-122027853-A - Video playing method and device, electronic equipment and computer program product

CN122027853ACN 122027853 ACN122027853 ACN 122027853ACN-122027853-A

Abstract

The disclosure relates to a video playing method and device, electronic equipment and a computer program product, and relates to the technical field of streaming media. The method comprises the steps of obtaining a video watching data set, determining video playing quality and watching exit probability under different video playing parameter values according to data characteristics corresponding to the video watching data, constructing a video evaluation model based on mapping relations between the data characteristics, the video playing quality and the watching exit probability, determining estimated playing quality and estimated exit probability of a target video through the video evaluation model, predicting to obtain user experience scores according to the estimated playing quality and the estimated exit probability, performing playing decision operation according to the user experience scores, and determining target playing parameter values of the target video. According to the method and the device, the mapping relation between the data characteristics of the video watching data, the video playing quality and the watching exit probability is analyzed, and the differential algorithm strategy dynamic adjustment is realized aiming at different users so as to optimize the subjective experience of the users.

Inventors

  • XU ZHIMIN
  • ZHOU CHAO

Assignees

  • 北京达佳互联信息技术有限公司

Dates

Publication Date
20260512
Application Date
20260130

Claims (13)

  1. 1. A video playing method, comprising: Acquiring a video watching data set, wherein the video watching data set comprises video watching data of a user in at least one video playing scene; determining video playing quality and viewing exit probability under different video playing parameter values according to the data characteristics corresponding to the video viewing data; Constructing a video evaluation model based on a first mapping relation between the data features and the video playing quality and a second mapping relation between the data features and the watching exit probability; Determining estimated playing quality and estimated exit probability of a target video through the video evaluation model, and predicting to obtain user experience scores according to the estimated playing quality and the estimated exit probability; And performing play decision operation according to the user experience score, and determining a target play parameter value of the target video.
  2. 2. The method according to claim 1, wherein determining video playback quality and a viewing exit probability at different video playback parameter values according to the data characteristics corresponding to the video viewing data comprises: determining data characteristics corresponding to each video watching data, wherein the data characteristics comprise playing quality characteristics, network characteristics, user characteristics and video content characteristics; Determining video playing quality corresponding to each video playing parameter value according to the playing quality characteristics and the network characteristics, wherein the video playing quality comprises first screen duration distribution, pause duration distribution and pause rate distribution; And determining the watching exit probability corresponding to each video playing parameter value according to the user characteristics, the video content characteristics and the video playing quality.
  3. 3. The method of claim 2, wherein the viewing exit probability comprises a pre-play exit probability distribution, and wherein determining the viewing exit probability for each of the video playback parameter values based on the user characteristics, the video content characteristics, and the video playback quality comprises: Determining the total number of streaming times of video playing, the first screen duration distribution and the play starting waiting duration distribution; And determining pre-playing exit probability corresponding to each of a plurality of pre-playing waiting time periods and unbiased first screen time period distribution of the video according to the total number of pull streams, the first screen time period distribution and the start waiting time period distribution.
  4. 4. The method of claim 2, wherein the viewing exit probability comprises a distribution of exit probabilities in the katon, and wherein determining the viewing exit probability for each of the video playback parameter values based on the user characteristics, the video content characteristics, and the video playback quality comprises: Determining the total number of the cartoon and the last cartoon of video playing, and the cartoon waiting time length distribution; And determining the stopping probability in the stopping corresponding to a plurality of different stopping waiting time lengths and the unbiased stopping time length distribution of the video according to the total number of stopping and pulling, the last stopping time length distribution and the stopping waiting time length distribution.
  5. 5. The method of claim 2, wherein the viewing exit probability comprises an in-play exit probability distribution, and wherein the determining the viewing exit probability for each of the video playback parameter values based on the user characteristic, the video content characteristic, and the video playback quality comprises: determining the total number of times of stream pulling of a first screen of video playing, the number of times of stream pulling out of normal playing and the number of times of stream pulling out of blocking; and determining the probability of exiting in playing and the unbiased cartoon probability distribution corresponding to a plurality of different video watching time periods according to the total number of the pulling streams of the first screen, the number of the normal playing exiting pulling streams and the number of the cartoon exiting pulling streams.
  6. 6. The method of claim 1, wherein the data characteristics include a play quality characteristic, a network characteristic, a user characteristic, and a video content characteristic, wherein the constructing a video assessment model based on a first mapping relationship between the data characteristics and the video play quality, and a second mapping relationship between the data characteristics and the viewing exit probability, comprises: Determining a first mapping relation among the playing quality characteristics, the network characteristics and the video playing quality; Constructing an initial play quality estimation model based on the first mapping relation, wherein the initial play quality estimation model comprises a first screen duration distribution estimation model, a pause duration distribution estimation model and a pause rate distribution estimation model; Determining a second mapping relationship among the user features, the video content features and the viewing exit probability; constructing an initial exit probability estimation model based on the second mapping relation, wherein the initial exit probability estimation model comprises a pre-playing exit rate estimation model, a blocking exit rate estimation model and a playing exit rate estimation model; And carrying out model training on the initial playing quality estimation model and the initial exit probability estimation model through the video watching data set to obtain a video estimation model, wherein the video estimation model comprises a playing quality estimation model and an exit probability estimation model.
  7. 7. The method of claim 1, wherein determining the predicted playback quality and predicted exit probability of the target video by the video assessment model comprises: Determining the estimated playing quality through the video evaluation model, wherein the estimated playing quality comprises a first screen duration distribution, a chucktime duration distribution and chucktime duration prediction probabilities corresponding to different watching and playing durations; And determining the predicted exit probability through the video evaluation model, wherein the predicted exit probability comprises a pre-playing exit prediction probability, a in-katon exit prediction probability and a in-playing exit prediction probability.
  8. 8. The method of claim 7, wherein predicting the user experience score based on the predicted play quality and the predicted exit probability comprises: determining scene streaming times of the target video in different scenes; Determining the total stopping probability of the target video according to the scene pulling times, the stopping time length distribution and the stopping prediction probability in stopping; Determining the distribution of the number of pull streams which are withdrawn in the clip corresponding to the target video and the distribution of the number of pull streams which are withdrawn in the normal play according to the total withdrawal probability of the clip, the clip prediction probability corresponding to the same-view play time length and the withdrawal prediction probability in the play; determining the number of times of pulling under a plurality of viewing time periods according to the number of times of pulling withdrawn from the clip and the number of times of pulling withdrawn from the normal play; And determining the user experience score according to the number of times of stream pulling under the plurality of viewing duration.
  9. 9. The method of claim 8, wherein the determining the number of scene pulls for the target video in different scenes comprises: determining the total number of the pull streams of the target corresponding to the target video; determining the first scene pull-stream times corresponding to the target video, wherein the first scene pull-stream times are the pull-stream times of the target video which exits before the first screen appears; Determining the second scene streaming times according to the churning time length distribution, wherein the second scene streaming times are streaming times when the target video appears on the first screen and is churning; And determining third scene pull times according to the target total pull times and the first scene pull times, wherein the third scene pull times are the pull total times of the first screen of the target video.
  10. 10. The method of claim 1, wherein the determining the target play parameter value of the target video according to the play decision operation performed by the user experience score comprises: obtaining a pre-configured candidate play decision mode, wherein the candidate play decision mode comprises a definition priority mode, a fluency priority mode and a delay priority mode; determining a target play decision mode based on the user experience score and the candidate play decision mode; And determining the target playing parameter value according to the target playing decision mode.
  11. 11. A video playback device, comprising: the system comprises a data set acquisition module, a video display module and a video display module, wherein the data set acquisition module is used for acquiring a video viewing data set, and the video viewing data set comprises video viewing data of a user in at least one video playing scene; The viewing and broadcasting index determining module is used for determining video playing quality and viewing and exiting probability under different video playing parameter values according to the data characteristics corresponding to the video viewing data; the evaluation model construction module is used for constructing a video evaluation model based on a first mapping relation between the data characteristics and the video playing quality and a second mapping relation between the data characteristics and the watching exit probability; The experience score prediction module is used for determining the estimated playing quality and the estimated exit probability of the target video through the video evaluation model, and predicting to obtain the user experience score according to the estimated playing quality and the estimated exit probability; And the play decision module is used for performing play decision operation according to the user experience score and determining a target play parameter value of the target video.
  12. 12. An electronic device, comprising: A processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the video playback method of any one of claims 1 to 10.
  13. 13. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the video playback method of any one of claims 1 to 10.

Description

Video playing method and device, electronic equipment and computer program product Technical Field The present disclosure relates to the field of streaming media technologies, and in particular, to a video playing method, a video playing device, an electronic device, and a computer program product. Background The network video refers to an online video playing service provided by a video website, and mainly utilizes video files in a streaming media format, which can comprise video on demand, live video and the like. Live broadcast is a high-interactivity video entertainment mode, especially refers to network live broadcast, and the network live broadcast covers the fields of live broadcast in a show, live broadcast in a game, live broadcast in a general life and the like. For serving users in millions or even tens of millions of heterogeneous networks, most content providers provide adaptive multi-rate algorithms to maximize the clarity and fluency of the user by dynamically deciding the definition range that the user views under different network conditions. However, under the condition of limited network bandwidth resources, the definition and the smoothness are required to be reasonably balanced, and the effect of optimizing both the definition and the smoothness cannot be achieved. It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the present disclosure and thus may include information that does not constitute prior art known to those of ordinary skill in the art. Disclosure of Invention The disclosure aims to provide a video playing method, a video playing device, an electronic device, a computer readable storage medium and a computer program product, so as to overcome the problems that the related adaptive multi-code rate technology does not consider subjective watching experience of users and cannot dynamically adjust differentiated strategies for different users to meet different user demands to at least a certain extent. Other features and advantages of the present disclosure will be apparent from the following detailed description, or may be learned in part by the practice of the invention. According to a first aspect of the present disclosure, a video playing method is provided, which includes obtaining a video viewing dataset, wherein the video viewing dataset includes video viewing data of a user in at least one video playing scene, determining video playing quality and viewing exit probability under different video playing parameter values according to data features corresponding to the video viewing data, constructing a video evaluation model based on a first mapping relation between the data features and the video playing quality and a second mapping relation between the data features and the viewing exit probability, determining estimated playing quality and predicted exit probability of a target video through the video evaluation model, predicting to obtain user experience scores according to the estimated playing quality and the predicted exit probability, and performing playing decision operation according to the user experience scores to determine target playing parameter values of the target video. In an exemplary embodiment of the disclosure, the determining video playing quality and viewing exit probability under different video playing parameter values according to the data features corresponding to the video viewing data includes determining data features corresponding to each video viewing data, wherein the data features include playing quality features, network features, user features and video content features, determining video playing quality corresponding to each video playing parameter value according to the playing quality features and the network features, wherein the video playing quality includes first screen duration distribution, pause duration distribution and pause rate distribution, and determining viewing exit probability corresponding to each video playing parameter value according to the user features, the video content features and the video playing quality. In an exemplary embodiment of the present disclosure, the determining the viewing exit probability corresponding to each video playing parameter value according to the user feature, the video content feature and the video playing quality includes determining a total number of pull streams of video playing, a first screen duration distribution and the play waiting duration distribution, and determining a play exit probability corresponding to each of a plurality of play waiting durations according to the total number of pull streams, the first screen duration distribution and the play waiting duration distribution. In an exemplary embodiment of the present disclosure, the determining the viewing exit probability corresponding to each video playing parameter value according to the user feature, the video content feature an