Understanding the role of social context and user factors in video Quality of Experience
Introduction
Online video services show a continuous growth. By 2010, over 71% of internet users had watched videos online, and this number grew from 33% in 2006 (Moore, 2011). These figures are forecasted to further grow in the coming years (Cisco, 2012, Moore, 2011). With a constantly increasing volume of streamed video data, maintaining a satisfactory video service to users at all times is challenging for internet and multimedia providers. Due to different technological limitations (e.g., bandwidth and storage constraints, network malfunctioning), visible artifacts (e.g., blockiness or blur due to compression, freezes or jerkiness due to transmission errors) can be introduced to any stage of the video delivery cycle (Pérez et al., 2011, Wang et al., 2003). This, in turn, can severely degrade the user’s satisfaction, and evidence shows that users intend to pay less if a service cannot meet their expectations (Naumann et al., 2010, Yamori and Tanaka, 2004). As a consequence, online video providers are eager to find ways to measure and predict user’s satisfaction with videos in order to optimize their video delivery chains.
Quality of Experience (QoE) is a concept commonly used to describe user’s overall satisfaction (Le Callet, Möller, & Perkis, 2012), reflecting the degree of delight or annoyance of a user with a (multimedia) system, service or application. In the past decades, user’s satisfaction with videos has been estimated mainly from a technical perspective, i.e., based on either the information gathered from the network and service conditions or from image and video analysis (Serral-Gracià et al., 2010). From a network management perspective, the concept Quality of Service (QoS) has often been equated to QoE. Here, network parameters, such as packet loss or delay (Asghar, Le Faucheur, & Hood, 2009), as well as video QoS parameters, e.g., the so-called join time at the start of playing the video or the buffering time during the video (Dobrian et al., 2011), were monitored; their compliance to given standards was considered enough to guarantee sufficiently high QoE. The signal processing community has instead relied more often on the analysis of information extracted from the decoded image/video signal to estimate the visibility of artifacts in it (Hemami and Reibman, 2010, Lin and Jay Kuo, 2011). Artifact visibility was considered to be inversely related to perceptual quality, and therefore to user satisfaction (Chikkerur, Sundaram, Reisslein, & Karam, 2011). In both cases, user satisfaction was mainly associated to technical properties of the multimedia signal, service or system.
Lately, research has shown that this approach has limitations, and that other elements concur to guarantee user satisfaction when watching video (Le Callet et al., 2012, Zhu et al., 2014). For example, recent studies claimed that QoE should also be considered from a user perspective (De Pessemier, De Moor, Joseph, De Marez, & Martens, 2013): evidence has been provided that user’s interest (Kortum & Sullivan, 2010) and personality (Wechsung, Schulz, Engelbrecht, Niemann, & Möller, 2011) influence QoE too. Such findings reveal the complexity of QoE: it is a combination of many influencing factors, not limited to QoS parameters nor artifact visibility.
Influencing factors on QoE are often grouped into three categories, i.e., system, user and context factors (Le Callet et al., 2012). System factors concern the technical aspects of a multimedia system (e.g., network parameters, media genre and media configuration). User factors refer to individual characteristics of the user who is experiencing the video (e.g., demographics, personal interest or personality). Context factors refer to the characteristics of the environment within which the video experience is consumed (e.g., physical features of the environment, economical factors related to the video fruition, presence or absence of co-viewers). As mentioned earlier, most research in the field has focused on system factors, leaving the contribution of user and context factors largely unexplored. However, the rise of online video fruition has created a shift from a passive viewing experience to a more active, personalized and shared experience, changing the traditional television market considerably (Tercek, 2011). Compared to traditional TV users who just watch scheduled programs, internet users are free to choose the content they want, at any point in time and space they want, through a variety of devices (e.g., tablets, smartphone or computers). Thus, it is expected that personal characteristics as well as context of fruition will play an important role in such viewing experiences. Moreover, the rise of social media has led to a new type of social viewing experience, where preferences for video content are clearly reported on social media platforms (through comments and ratings), and are visible to the rest of the (vast) online community. The social context in which the video is experienced is therefore expected to play a key role in the eventual user satisfaction.
As the optimization of online video watching requires a more in-depth understanding of the impact of user and context factors on QoE, we here want to contribute to the generation of this knowledge by considering the impact of social context in particular. Interestingly, very little is known about how social context (1) relates to QoE and (2) combines with system and user factors to determine the final user satisfaction with the viewing experience. We specifically focus on what we define as “direct” social context, that is, the presence or absence of co-viewers in the physical proximity of the user. We report the outcomes of an empirical study looking into the role played by direct social context in determining QoE when given system factors (i.e., video genre and bitrate) are in place. Furthermore, we analyze the interactions of direct social context with user influencing factors such as demographics, interest in the video genre and immersive tendency. We measure six different aspects of the viewing experience, namely perceived video quality, enjoyment, endurability, satisfaction, involvement and information assimilation. The outcomes should support building an accurate objective model for QoE on the longer term.
The paper continues by presenting the related work in Section 2, which we reviewed to define the hypotheses for the empirical study as described in Section 3. We then outline our experimental methodology in Section 4, followed by the analysis of the results in Section 5. We discuss our findings in Section 6, leading to the most important conclusions in Section 7.
Section snippets
Related work
In the past decades, the effectiveness of multimedia services has been linked to the notion of Quality of Service (QoS), defined as the “totality of characteristics of a telecommunication service that bears on its ability to satisfy stated and implied needs of the user of the service” (ITU-T, 1994). QoS is mainly operationalized in terms of system and network performance-related measures (e.g., packet loss ratio, jitter or delay). This approach has started showing its limitations, and was found
Research questions and hypotheses
Based on the literature overview given in Section 2, we formulate three research questions:
- 1.
What is the effect of direct social context on QoE?
- 2.
How is the impact of system factors on QoE affected by the direct social context?
- 3.
How is the impact of user factors on QoE affected by the direct social context?
To answer these research questions, QoE is measured along the six attributes, mentioned above: perceived visual quality, enjoyment, satisfaction, endurability, involvement and information
Experimental design
To test our hypotheses, we created two real-life viewing situations with varying direct social context. In the first situation, single users (hereafter indicated with S, shown in Fig. 1a) watched the videos alone (i.e., absence of direct social context). In the second one, a group of three friends (hereafter indicated with G, shown in Fig. 1b) watched the videos together. Participants who were involved in one social situation (e.g., single) were not presented with the other situation (e.g.,
Data preparation
Before discussing our results in more detail, we performed a number of bias checks on the distribution of our participants over the two social contexts, i.e., participation in the single vs. group viewing situation. Note that for some variables such as interest, immersive tendency and some demographic data, values of one participant contributing to the group viewing situation were missing. Thus, where applicable, the results of only 59 instead of 60 participants are reported. In addition, since
Discussion
Quality of Experience is a very complex concept and its proper quantification still has several challenges ahead. Based on existing literature, we proposed to measure various aspects of Quality of Experience, including perceived visual quality (along two separate dimensions of (1) artifact visibility and (2) overall quality), enjoyment, satisfaction, endurability, involvement and information assimilation. Measurement scales for perceived visual quality are well established; conversely, no
Conclusions
In this paper, we investigated a set of influencing factors on user’s QoE with videos. Our results showed that co-viewing videos with friends increased the user’s level of enjoyment and enhanced the endurability of the experience, indicating that social context should be further investigated in relation to QoE and considered also in automated measurements. The presence of co-viewers did not change participant’s ability to detect visual artifacts, yet the presence of visible artifacts did not
Acknowledgements
This work is supported in part by the scholarship from China Scholarship Council (CSC) under the Grant CSC No. 201206090028. This work is also partially supported by the NWO Veni Grant 639.021.230.
References (70)
- et al.
No-reference image and video quality estimation: Applications and human-motivated design
Signal Processing: Image Communication
(2010) - et al.
Perceptual visual quality metrics: A survey
Journal of Visual Communication and Image Representation
(2011) - et al.
The relationship between individual characteristics and experienced presence
Computers in Human Behavior
(2013) - et al.
Multimodal interaction: A suitable strategy for including older users?
Interacting with Computers
(2010) - et al.
User experience on mobile video appreciation: How to engross users and to enhance their enjoyment in watching mobile video clips
Technological Forecasting and Social Change
(2012) - et al.
Brain activity correlates of quality of experience
(2014) - et al.
Preserving video quality in IPTV networks
IEEE Transactions on Broadcasting
(2009) - et al.
A quest for an internet video quality-of-experience metric
(2012) Presence and image quality: The case of high-definition television
Media Psychology
(2005)- et al.
Emotion and motivation II: Sex differences in picture processing
Emotion
(2001)