Patent Application
20220400309
The present invention relates to an engagement estimation device, an engagement estimation method, and a program.
Communication services (voice communication, video communication, Web, IoT, and the like) for transferring video, sound (hereinafter including voice), and text between terminals or between a server and a terminal via the Internet are popular.
The Internet is a network in which communication quality is not necessarily guaranteed and hence, in the case where communication is performed with an audio medium or a visual medium, packet loss, packet transfer delay, and packet retransmission occur due to a reduction in bit rate and line congestion caused by narrowness of the bandwidth of a line between a viewer terminal and a network, and the quality of the audio medium or the visual medium perceived by a viewer is degraded.
Specifically, in adaptive bit rate video streaming in which a video is distributed while the bit rate of the audio or visual medium is changed according to the state of the throughput of the network, reductions in sound quality and image quality caused by a reduction in throughput occur, and a wait for a playback start and a playback stop caused by buffering processing which occurs due to the fact that a predetermined amount of data is not accumulated in a buffer of a receiving terminal also occur.
The reduction in bit rate, the wait for the playback start, and the playback stop influence not only quality experienced by a user but also engagement (viewing time, viewing permission suspension/discontinuation, viewing cancellation).
Accordingly, in order for a video distributor to determine whether the above-described video communication service is provided with excellent quality and engagement, it is important to be able to measure audiovisual quality experienced by the viewer and engagement during the provision of the service to monitor high levels of the audiovisual quality and the engagement provided to the viewer.
Consequently, a quality estimation technique and an engagement estimation technique capable of properly expressing the audiovisual quality experienced by the viewer and the engagement are necessary.
One of conventional objective quality assessment methods is ITU-T recommendation P. 1203 shown in NPL 1 or a technique shown in NPL 2. The present technique is a technique in which quality is estimated from packets received by a receiving terminal (a smartphone or an STB (Set-Top Box)) by using quality parameters such as a resolution, a frame rate, a bit rate, a playback start waiting time, and a playback stop time of a video.
However, in the case where the video distributor attempts to monitor the engagement (viewing time, viewing permission suspension/discontinuation, hereinafter viewing discontinuation will be described) by using information in a user terminal, an engagement estimation technique is necessary, and such a technique does not exist at the moment.
The present invention has been made in view of the above points, and an object thereof is to allow estimation of engagement from parameters which can be observed in a terminal.
In order to solve the above problem, an engagement estimation device includes an engagement estimation unit which calculates an estimated value of an index for assessing engagement of a viewer of a video at a certain elapsed time from a start of playback of the video distributed via a network, and the engagement estimation unit calculates the estimated value such that, in a case where the playback of the video is stopped at the certain elapsed time, the stop of the playback influences the estimated value more significantly than in a case where the playback of the video is not stopped at the certain elapsed time.
It is possible to allow the estimation of the engagement from parameters which can be observed in the terminal.
Hereinbelow, an embodiment of the present invention will be described based on the drawings.
A program for implementing processing in the engagement estimation device 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 in which the program is stored is set in the drive device 100, the program is installed in the auxiliary storage device 102 from the recording medium 101 via the drive device 100. Note that the installation of the program does not necessarily need to be performed from the recording medium 101, and the program may also be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program and also stores required files and data.
In the case where an instruction to activate the program is issued, the memory device 103 reads the program from the auxiliary storage device 102 and stores the program. The CPU 104 implements functions related to the engagement estimation device 10 according to the program stored in the memory device 103. The interface device 105 is used as an interface for connection to a network. The display device 106 displays a GUI (Graphical User Interface) by the program or the like. The input device 107 is constituted by a keyboard, and a mouse, a button, or a touch panel, and is used to input various operation instructions.
For example, when an instruction to start viewing of a video by the adaptive bit rate video streaming (hereinafter referred to as “target video”) is issued, the quality parameter acquisition unit 11 acquires, for a preset measurement section (e.g., a unit such as one second, one minute, five minutes, or one hour, or each unit time of a chunk/segment unit used in video distribution (the unit time is assumed to be one second in the present embodiment)), parameters indicative of playback states such as a chunk start, a playback start waiting state, and a playback stop state in addition to parameters such as, e.g., a resolution, a frame rate, and a bit rate in the measurement section (or at the time of start or end of the measurement section) (hereinafter these parameters are collectively referred to as “quality parameters”) from, e.g., an application or the like which plays back the video, and the quality parameter acquisition unit 11 associates the acquired quality parameters with an elapsed time from a playback start, and stores the quality parameters in the quality parameter storage unit 13 in chronological order. Note that the application operates in a terminal for viewing. Accordingly, each quality parameter is data which can be observed in the terminal.
The engagement estimation unit 12 calculates an estimated value (hereinafter referred to as “viewing discontinuation value QR(t)”) of an index for assessing the engagement at a certain elapsed time t from the start of the playback of the target video (t=0 is satisfied when the playback is started) based on a history or the like of the quality parameters stored in the quality parameter storage unit 13. The value of t may be input by, e.g., a user, or may also be a cyclic value of each measurement section. In the present embodiment, the number of viewers who discontinue viewing (hereinafter referred to as “viewing-discontinuation viewer”) is used as the index. Specifically, a viewing discontinuation rate at the elapsed time t is described as an example of the viewing discontinuation value QR(t). Note that the viewing discontinuation rate at the elapsed time t denotes the rate of viewers who discontinue viewing cumulatively before the elapsed time t, and has a value between 0 and 1.
Note that the engagement estimation device 10 is, e.g., a computer which is used for the viewing of the adaptive bit rate video streaming (hereinafter referred to as “viewer terminal”). Specifically, a smartphone, a tablet terminal, or PC (Personal Computer) may be used as the engagement estimation device 10.
In the present embodiment, the engagement estimation unit 12 estimates the viewing discontinuation value by a method in the case where a playback stop is not present at the elapsed time t, and estimates the viewing discontinuation value by another method in the case where the playback stop is present at the elapsed time t. Note that the case where the playback stop is not present at the elapsed time t denotes a state in which the playback is not stopped at the elapsed time t, and the presence or absence of the occurrence of the playback stop at a point of time before the elapsed time t does not matter. Similarly, the case where the playback stop is present denotes a state in which the playback is stopped at the elapsed time t, and the presence or absence of the occurrence of the playback stop at a point of time before the elapsed time t does not matter.
First, a description will be given of a calculation method of the viewing discontinuation value QR(t) in the case where the playback stop is not present at the elapsed time t. In this case, an estimation formula of QR(t) (hereinafter referred to as “non-playback stop model”) is as follows.
Herein, in the case where t=0 is satisfied, QR(t) is 0. In addition, tchunk_start denotes an elapsed time at the time of start of playback of a video chunk (a start time of a video chunk, i.e., tchunk_start<=t) which is played back at the elapsed time t (is being played back at the elapsed time t). MOSvideo (x) denotes video quality at an elapsed time x. MOSaudio denotes audio quality at the elapsed time t. Note that each of the video quality and the audio quality is a value which is output with the bit rate, the resolution, the frame rate, a playback start waiting time, and a playback stop time used as inputs by a quality estimation technique such as ITU-T recommendation P. 1203.
tquality_change denotes an elapsed time at a point of time when the video quality is changed latest before the elapsed time t (last before the elapsed time t). Note that, in the case where tquality_change which meets the condition is not present (i.e., in the case where the video quality is not changed before the elapsed time t), tquality_change=0 is satisfied and, there is no change of the video quality, and hence ΔMOS=0 is satisfied.
In addition, tex_stalling_end denotes an elapsed time at a point of time when the playback stop is ended latest before the elapsed time t (last before the elapsed time t). Note that, in the case where tex_stalling_end which meets the condition is not present (i.e., in the case where the playback stop is not ended before the elapsed time t), tex_stalling_end=0 is satisfied.
In addition, max(a, b) means one of values of a and b which is larger than the other. Further, c1, c2, c3, c4, c5, and c6 are coefficients. Note that λ is supposed to be a value of more than 0 and, in the case where λ is not more than 0, processing in the case where λ is not more than 0 such as replacing λ with a present minimum value or performing calculation with e{circumflex over ( )}(−(t−tchunk_start)/λ) of a calculation formula of QR(t) set to 0 is necessary. Note that
A description will be given of characteristics of QR(t) in the case where the playback stop is not present at the elapsed time t. As basic characteristics, when the elapsed time t is increased, QR(t) is also increased. In addition, when MOSvideo(t), MOSaudio, and Qalr have large values, the increase of QR(t) becomes gentle. In addition, in the case where ΔMOS<0 is satisfied, as its absolute value is increased, the increase of QR(t) is intensified. In the case where ΔMOS>0 is satisfied, as its absolute value is increased, the increase of QR(t) becomes gentle.
Next, a description will be given of a calculation method of the viewing discontinuation value QR(t) in the case where the playback stop is present at the elapsed time t. In this case, an estimation formula of QR(t) (hereinafter referred to as “playback stop model”) is as follows.
Herein, tstalling_start denotes an elapsed time at the time of start of the playback stop (a start time of the playback stop). Δtstalling denotes a playback stop continuation time. MOSvideo denotes the video quality at the elapsed time t. Note that the video quality is a value which is output with the bit rate, resolution, frame rate, playback start waiting time, and playback stop time used as inputs by the quality estimation technique such as ITU-T recommendation P. 1203. Note that, in the case where the playback stop is present, the influence of the audio quality on the viewing discontinuation value is small, and hence the audio quality is excluded from the estimation formula.
Further, tquality_change denotes an elapsed time at a point of time when the video quality is changed latest before the elapsed time t (last before the elapsed time t). Note that, in the case where tquality_change which meets the condition is not present (i.e., in the case where the video quality is not changed before the elapsed time t), tquality_change=0 is satisfied, and QR(tquality_change)=QR(0)=0 is satisfied.
In addition, tex_stalling_end denotes an elapsed time at a point of time when the playback stop is ended latest before the elapsed time t (last before the elapsed time t). Note that, in the case where tex_stalling_end which meets the condition is not present (i.e., in the case where the playback stop is not ended before the elapsed time t), tex_stalling_end=0 is satisfied. Further, s1, s2, s3, s4, s5, s6, and s7 are coefficients. Note that
A description will be given of characteristics of QR(t) in the case where the playback stop is present at the elapsed time t. When Δtstalling is increased, QR(t) is also increased. In addition, when the MOSvideo is increased, QR(t) is reduced in a relationship between MOSvideo and QR(t).
Thus, in each of the non-playback stop model and the playback stop model, the viewing discontinuation value Q(t) at a point of time (measurement section) previous to the elapsed time t is cumulatively (or recursively) calculated, and the viewing discontinuation value Q(t) at the elapsed time t is calculated. In addition, as is clear from the above description, the viewing discontinuation value Q(t) is calculated such that the stop of the playback influences the viewing discontinuation value Q(t) more significantly in the playback stop model than in the non-playback stop model.
Hereinbelow, a description will be given of a processing procedure executed by the engagement estimation device 10.
In Step S101, the quality parameter acquisition unit 11 acquires the quality parameters in each measurement section from the start of the playback of an assessment-target video (target video) to the end of the playback, and stores the acquired quality parameters in the quality parameter storage unit 13.
Thereafter, at any timing (e.g., timing when the estimation of the viewing discontinuation value QR(t) is requested by a user), the engagement estimation unit 12 calculates the viewing discontinuation value QR(t) related to the target video at the elapsed time t based on the history or the like of the quality parameters stored in the quality parameter storage unit 13 (S102). The elapsed time t may be set at the end of each measurement section from the start of the playback of the target video to the end of the playback, or may also be any one value. In the case of the former, the viewing discontinuation value QR(t) is obtained for each measurement section (every second). In the case of the latter, the viewing discontinuation value QR(t) of any one elapsed time t is obtained.
Note that S101 and S102 may also be performed in synchronization with each other for each measurement section. In this case, the viewing discontinuation value QR(t) is obtained for each measurement section.
Then, a description will be given of the detail of Step S102.
In Step S201, the engagement estimation unit 12 initializes various variables used in the processing procedure in
Then, the engagement estimation unit 12 compares values of t and x (S202). t is a numerical value indicative of the elapsed time of a calculation target of the viewing discontinuation value QR(t). In the present embodiment, the unit of the measurement section is one second, and hence the unit of each of t and x is a second. In the case where x is less than t (Yes in S202), the engagement estimation unit 12 adds 1 to x (S203). Then, the engagement estimation unit 12 determines whether or not the playback of the target video is stopped at an elapsed time x by referring to the quality parameter storage unit 13 (S204).
In the case where the playback of the target video is stopped at the elapsed time x (Yes in S204), the engagement estimation unit 12 determines whether or not the playback stop is started in a period from x−1 to x by referring to the quality parameter storage unit 13 (S205). Herein, the period from x−1 to x denotes a period before x (including x) after x−1, and the same applies to the following description. In the case where the playback stop is started in the period from x−1 to x (Yes in S205), the engagement estimation unit 12 assigns x−1 to tstalling_start (S206).
In the case of No in Step S205, or subsequently to S206, the engagement estimation unit 12 calculates a viewing discontinuation value QR(x) at the elapsed time x by using the playback stop model (S207), and returns to Step S202.
On the other hand, in the case where the playback of the target video is not stopped at the elapsed time x (i.e., in the case where the target video is being played back) (No in S204), the engagement estimation unit 12 determines whether or not a start time point (start time) of any chunk of the target video is included in the period from x−1 to x by referring to the quality parameter storage unit 13 (S208). In the case where the start time point of any chunk is included in the period from x−1 to x (Yes in S208), the engagement estimation unit 12 assigns x−1 to tchunk_start (S209). Then, the engagement estimation unit 12 determines whether or not MOSvideo is changed in the period from x−1 to x (whether or not MOSvideo (x−1) is different from MOSvideo(x)) by referring to the quality parameter storage unit 13 (S210). Note that MOSvideo at x−1 can be calculated based on the quality parameter related to x−1 stored in the quality parameter storage unit 13. Similarly, MOSvideo at x can be calculated based on the quality parameter related to x stored in the quality parameter storage unit 13.
In the case where MOSvideo is changed in the period from x−1 to x (Yes in S210), the engagement estimation unit 12 assigns a value obtained by subtracting MOSvideo(x−1) from MOSvideo(x) to ΔMOS (S211). On the other hand, in the case where MOSvideo is not changed in the period from x−1 to x (No in S210), the engagement estimation unit 12 determines whether or not the playback is resumed (has recovered from the playback stop) in the period from x−1 to x by referring to the quality parameter storage unit 13 (S212).
In the case where the playback is resumed in the period from x−1 to x (Yes in S212), or subsequently to Step S211, the engagement estimation unit 12 assigns QR(x−1) to Qalt (S213).
Subsequently to Step S213 or the case of No in Step S208, the engagement estimation unit 12 calculates the viewing discontinuation value QR(x) at the elapsed time x by using the non-playback stop model (S214), and returns to Step S202.
When the above processing procedure is repeated and the value of the variable x reaches t (No in S202), the engagement estimation unit 12 outputs the calculation result in Step S207 or S214 which is executed last as the viewing discontinuation value QR(t) (S215).
Thus, the viewing discontinuation value QR(t) is calculated cumulatively by using the previous viewing discontinuation value QR.
As described above, according to the present embodiment, it is possible to allow the estimation of the engagement from the parameters which can be observed in the terminal.
Consequently, according to the present embodiment, by monitoring the engagement value of the communication service which is actually viewed by the viewer, it is possible to easily determine whether or not the service which is being provided maintains engagement of a certain level or higher for the viewer, and it becomes possible to understand and manage the actual state of the engagement of the service which is being provided in real time.
In addition, the engagement is influenced not only by the real-time quality parameter (the playback stop or the like) but also by the previous quality parameter and, in the present embodiment, it is possible to estimate the engagement in consideration of the previous quality parameter.
While the embodiment of the present invention has been described in detail thus far, the present invention is not limited to the specific embodiment, and various modifications and changes can be made within the scope of the gist of the present invention described in the scope of claims.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/JP2019/044168 | 11/11/2019 | WO |
