Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Visually Impaired Users on an Online Social Network


Published on

This is a large-scale study on the behavior of visually impaired users on online social networks, specifically Facebook.

Published in: Education
  • Be the first to comment

Visually Impaired Users on an Online Social Network

  1. 1. Visually Impaired Users on an Online Social Network Shaomei Wu Lada Adamic Facebook Inc. Menlo Park Presented by Sahithi Thandra CS886-Winter 2016 1
  2. 2. CS886-Winter 2016 2 ✦ Introduction ✦ Data ✦ Demographics ✦ Facebook activity ✦ Content Analysis ✦ Network Structure ✦ Conclusion ✦ Discussion Contents
  3. 3. CS886-Winter 2016 3 Introduction ✦ Vision Impairment is a prevalent health problem worldwide. ✦ 285 million are visually impaired globally. ✦ 6.6 million are visually impaired in the US. ✦ How do the visually impaired people use the Internet today ?
  4. 4. CS886-Winter 2016 4 Introduction ✦ Visually impaired people engage with online social networks through technologies such as: ✦ OCR Readers ✦ WAI-ARIA ✦ Screen Readers ✦ They access online networks through Desktop and Mobile.
  5. 5. CS886-Winter 2016 5 Questions ✦ Behaviour of visually impaired users differ significantly from behavioural patterns of other users? ✦ What kind of content do they produce and share on Social networking sites? ✦ How are visually impaired users’ social networks structured?
  6. 6. CS886-Winter 2016 6 ✦ Analysis focuses on Facebook mobile users through VoiceOver. ✦ Facebook’s mobile interface is more accessible and usable for blind users than Desktop version. Data Visually Impaired Users (VoiceOver Sample) 50K Random sample of Users (iOS Sample) 160K
  7. 7. CS886-Winter 2016 7 Demographics Rank VoiceOver Sample iOS Sample 1 US(32.5%) US(35.5%) 2 UK(7.2%) UK(7.3%) 3 France(4.9%) Japan(4.9%) 4 Germany(4.3%) Canada(3.8%) 5 Italy(4.1%) Germany(3.6%)
  8. 8. CS886-Winter 2016 8 Demographics ✦ Country: Sample of vision-impaired users are highly skewed towards people from developed western countries. ✦ Age: Both samples are skewed towards young adults. Average age for VoiceOver sample is 30.14 years & iOS sample is 30.43 years. ✦ Gender: Both genders are well represented in two samples. ✦ Users are mostly younger and are with relatively high income.
  9. 9. CS886-Winter 2016 9 ✦ How visually impaired users engage with these 4 types of activities: ✦ Status updates ✦ Photo Sharing ✦ Comments & ✦ Likes ✦ What do visually impaired users do on Facebook? ✦ How do other users interact with the visually impaired? Facebook Activity
  10. 10. CS886-Winter 2016 10 Key Findings Per user activity over three weeks, bootstrap averaged across user samples
  11. 11. CS886-Winter 2016 11 Question asking behaviour ✦ Overall, question asking behaviour is rare in both populations and asking questions about photos is uncommon. ✦ This result is partly explained by the smaller number of photos uploaded by VoiceOver users.
  12. 12. CS886-Winter 2016 12 Status Updates Two-point trends detection algorithm. Content Analysis
  13. 13. CS886-Winter 2016 13 Status Updates ✦ Top 10 words from the VoiceOver sample by both metrics are related to vision disability, while words from iOS sample are very general. ✦ Compared to other iPhone users, VoiceOver users on Facebook are much more likely to discuss vision impairment and accessibility issues found in both the physical world and on the Internet. ✦ This data helps to distinguish visually impaired users from other social media users. Content Analysis
  14. 14. CS886-Winter 2016 14 Photo Sharing ✦ Why do visually impaired users’ photos not receive as much feedback from others as their status updates do ? Content Analysis
  15. 15. CS886-Winter 2016 15 Photo Sharing ✦ Many photos uploaded by visually impaired users are automatically created by apps instead of users themselves. ✦ Sighted users may consider these images spammy and ignore them. Findings ✦ Visually impaired users openly talk about their experiences and issues with vision disability and web accessibility. Content Analysis
  16. 16. CS886-Winter 2016 16 ✦ Can the openness we observe in vision-impaired users be a result of their social networks being denser than average? ✦ What could be the reason behind visually impaired users receiving more comments and likes ? ✦ Is it because they have more friends (and thus a bigger audience) than an average user ? Network Structure
  17. 17. CS886-Winter 2016 17 Network Size ✦ VoiceOver median friend count of 208 is lower than the iOS median of 242. ✦ VoiceOver users have on average been on Facebook for 38 months compared to 46 months for the iOS sample. Network Structure
  18. 18. CS886-Winter 2016 18 Network Density ✦ The level of diversity in personal social networks is almost identical for users from the VoiceOver sample and users from iOS sample. ✦ Most visually impaired users have closely connected social networks with a few communities, but this is also true for other users on Facebook. ✦ Visually impaired users don’t have denser than average networks over all. Interconnectivity among visually impaired users. ✦ Visually impaired users are more likely to friend other visually impaired users. Network Structure
  19. 19. CS886-Winter 2016 19 ✦ Visually impaired users use Facebook just like normal Facebook users. ✦ They receive more feedback than average user. ✦ They can use Facebook as an utility to openly share their experience, voice their concerns and receive attention and support from others. ✦ Difference in network size between two groups diminished over time – progress towards increasingly equal and accessible online environment. Conclusion
  20. 20. CS886-Winter 2016 20 Limitations of the study ✦ Users who access Facebook with VoiceOver and iPhones. ✦ Users are from developed countries with high-economic status. ✦ History and severity of vision impairment of Users is not taken into account. ✦ Studied only Facebook users.
  21. 21. CS886-Winter 2016 21 Design Implications ✦ Auto-detection of users’ vision impairment could help online services to recognize this underserved market. ✦ Auto-detection of visually impaired users could also enable online services to better adapt to their needs. ✦ Currently, there is no a standard and effective method to identify visually impaired users. ✦ If identified, we can redirect visually impaired users to the accessible version of websites.