The impact of visual attributes on online image diffusion

Abstract

Little is known on how visual content affects the popularity on social networks, despite images being now ubiquitous on the Web, and currently accounting for a considerable fraction of all content shared. Existing art on image sharing focuses mainly on non-visual attributes. In this work we take a complementary approach, and investigate resharing from a mainly visual perspective. Two sets of visual features are proposed, encoding both aesthetical properties (brightness, contrast, sharpness, etc.), and semantical content (concepts represented by the images). We collected data from a large image-sharing service (Pinterest) and evaluated the predictive power of different features on popularity (number of reshares). We found that visual properties have low predictive power compared that of social cues. However, after factoring-out social influence, visual features show considerable predictive power, especially for images with higher exposure, with over 3:1 accuracy odds when classifying highly exposed images between very popular and unpopular.

Who Will Share My Image?

“…Second, we see that social features alone produce better prediction results than image features alone. This corroborates many previous findings (e.g., [8,22,33]), which showed that social features are more important than image features for content popularity prediction. In our case, we are predicting the diffusion path, which is a much more difficult problem, and when no social features are used, it is impossible to say precisely how an image will be propagated through a social network.…”

“…A large number of recent efforts have explored ways to predict content popularity, including for images [7,8,10,14,22,33], videos [26], GitHub repositories [5], blogs [1], memes [36], and tweets [18,19,25,28], by combing content features with user social features. In contrast to these prior work, which mainly focus on predicting a popularity score (e.g., number of shares) of the content, we aim to predict the entire content diffusion path through the social network, which is a much more challenging task.…”

A picture is worth a thousand words: an empirical study on the influence of content visibility on diffusion processes within a virtual world

“…Various content and information can be the subject of diffusion within social networks, including images (Totti et al 2014), video (Boynton 2009;Nelson-Field, Riebe, and Newstead 2013), news (Yang and Counts 2010), rumours (Jin et al 2013), information about promotions and offers (Leskovec, Adamic, and Huberman 2007), virtual goods in virtual worlds, and special effects that can be applied to avatars, hair styles, or clothing (Bakshy, Karrer, and Adamic 2009;Huffaker et al 2011).…”

Dynamic Rankings for Seed Selection in Complex Networks: Balancing Costs and Coverage

“…Information spreading processes are a basement of viral marketing [1], rumours [2], social and political changes [3], innovation adoption [4] and other initiatives with emotional appeal [5]. Possible types of contents propagated within electronic systems include texts [6], concepts [7], video material [8], images [9] and marketing messages [10]. Research in this area is, among others, related to information flow within the networks [11], factors affecting performance of marketing campaigns [12] and the selection of highly influential nodes [13].…”