Presented by Jun-Ming Chen 6/18/2010. Outline Introduction Background research Methodology – Study 1: Exploratory questionnaire – Study 2: Playing with.

Social Tagging Revamped: Supporting the Users’ Need of

Self-promotion through Persuasive Techniques

Presented by Jun-Ming Chen 6/18/2010

Outline

• Introduction• Background research• Methodology

– Study 1: Exploratory questionnaire– Study 2: Playing with mutual modeling– Study 3: Self-promotion experiment

• Conclusion• Comments

2

3

Facebook statistics

• 100 Million photos uploaded to the site each day

• More than 5 billion pieces of content (web links, news stories, blog posts, notes, photo albums, etc.) shared each week

• Average user has 130 friends on the site

• Average user clicks the Like button on 9 pieces of content each month

• Average user writes 25 comments on Facebook content each month

4

5

Automatic Tagging

6

7

Problem space

8

Background research

• Why normal users tagged infrequently– Users could easily retrieve their pictures using their chronological order

– Solution: • Sharing pictures with remote peers (meanwhile, maintaining relationships)• Telling stories with these photos helps nurture the relationships

– Future issue• Study how to support sharing practices within social networks

9

Background research

• Social Tagging and GWAP (Games With A Purpose)– Background:

People tend to organize their pictures by date and events

– Users add tags to multimedia content• facilitate their retrieval

– These limitations of user-generated tags• A lack of standardization of the chosen vocabulary• A lack of precision in choosing the best words

that could describe the content

Alternative methods to generate descriptors of multimedia contentESP Game

– Shortcoming : the redundancy in the tag sets

10

Background research

• Information Overload– Bad money drives out good– Users might restrict their focus to

a specific group of trusted friends– Study 1

• Information Overload, is it really a problem ?

– Solution• Signaling

– The preference ranking– thumbs up/down (facebook)

– Signal the quality of their work through an objective and unambiguous signal– The two quality signaling mechanisms tested in study 3

11

social navigation

Methodology

• Study 1: Exploratory questionnaire– How do people cope with information overload in social networks (Facebook)?

• Common solutions: this problem is to have valid descriptors for the content; these can be obtained using techniques Still have drawbacks Study 2

• Study 2: Playing with mutual modeling– When describing the same multimedia content,

do members of the same social network generate terms that are more specific than those generated by people out side the social network?

12

Tagging 的活動的資訊取得• 藉由同儕間的互動• 透過共同的熟識的人所提供正確的資訊

Tagging 的活動的資訊取得• 藉由同儕間的互動• 透過共同的熟識的人所提供正確的資訊

Methodology

• Study 3: Self-promotion experiment– Finally, Akerlof suggested that a signaling functionality can be useful to people

to distinguish information sources [2].– Therefore, we thought that the descriptors obtained within the social network

could be used to design a signaling functionality for multimedia content shared among the peers (i.e., promote a peers profile in a social network)

– We also wondered whether such signaling functionality could be readily appropriated by the users.

– Can we design a quality control mechanism (involving persuasive techniques) such that social network peers would be willing to provide meaningful descriptors of multimedia content that would support personal promotion?

13[2] AKERLOF,G.A.The market for lemons: Quality uncertainty and the market mechanism. The Quartely Journal of Economics 84, 3 (August 1970), 488]500.

Study 1: Exploratory questionnaire

• How do people cope with information overload in social networks (Facebook)?• Common solutions:

this problem is to have valid descriptors for the content; these can be obtained using techniques Still have drawbacks Study 2

• 48 Facebook users / m: 36, f:12, median age: 26.5 years• 3 independent social networks

– graduate students, U.S. residents of different nationalities, different ages, and areas of study– both students and full-time employees of a larges of software company specialized in

customer services– an international group of researchers and administrative assistant within a big

telecommunications company in Spain

14

Do you feel overwhelmed by the amount of updates from your friends’ activity on Facebook?

15

What strategies do you use to keep up with what friends are doing on Facebook?

16

Summary Study 1

• These results confirm that information overload is a problem is social networking sites

• Most of respondents felt overwhelmed by the amount of updates in FB and they adopted few strategies to overcome it

17

Study 2: Playing with mutual modeling

• Study 2: Playing with mutual modeling– When describing the same multimedia content,

do members of the same social network generate terms that are more specific than those generated by people out side the social network?

18

Tagging 的活動的資訊取得• 藉由同儕間互惠的互動• 透過共同的熟識的人所提供正確的資訊

Tagging 的活動的資訊取得• 藉由同儕間互惠的互動• 透過共同的熟識的人所提供正確的資訊

19

20

The participants were asked to submit a photo from their personal repository that they thought would be of interest to the other participants taking place in the experiment

The participants were asked to submit a photo from their personal repository that they thought would be of interest to the other participants taking place in the experiment

21

22

23

• Most of the specific single-word keywords were contained in the commentaries

• Peers belonging to the same social network generate terms that are more specific than those generated by people outside the social network

Summary Study 2

• People tend to like photo comments from their peers, mostly when they include jokes

• Relationships between peers are different and affect comment appreciation

• Descriptors of peers were more specific

• This study suggested the idea that commentaries could be used as source of metadata instead of single-word tags

24

Study 3: Self-promotion experiment

• Study 3: Self-promotion experiment

– Can we design a quality control mechanism (involving persuasive techniques) such that social network peers would be willing to provide meaningful descriptors of multimedia content that would support personal promotion?

25

使用說服力的佐證技術 對於一些議題，成員從相互比較的過程，更加肯定原有的認知 經由互動來吸引人們去激勵或影響他們的行為

Elicit knowledge (in the form of metadata or tags)from peers in the social network

使用說服力的佐證技術 對於一些議題，成員從相互比較的過程，更加肯定原有的認知 經由互動來吸引人們去激勵或影響他們的行為

Elicit knowledge (in the form of metadata or tags)from peers in the social network

26

27

Experimental Design

• Preparation phase– Four subjects (photo providers) from each social network were asked to send us

50 photos from one of their online photo albums and to choose the one photo that they thought could better promote their album

• Experimental phase– 3 phrase

28

Experimental phase1. First, the remaining peers of each social network (SN1: 17, SN2: 14, SN3:

20 participants) were asked to access four web pages similar to the one and to choose in each page the feed that they would choose in order to further explore the photo album

29

SN1: 17SN2: 14

SN3: 20

Experimental phase2. Each of the four web pages presented four different feeds promoting the

same photo album from one of the four photo providers from the preparation phase

3. Users’ interaction was logged and once they confirmed their choice for a particular feed, they also provided us with their main reasons for selecting that feed presentation mode for each webpage

30

31

significant difference

Comments that have been generated by means ofAn entertaining structure seem to attract more

attention than those usually posted in social networks.

Comments that have been generated by means ofAn entertaining structure seem to attract more

attention than those usually posted in social networks.

32

• Effective signaling should allow consumers to sample the quality of the content without the need of inspecting the content

• The number of “thumbs up” that appear in the bottom of a post allow efficient social navigation

• Effective signaling should allow consumers to sample the quality of the content without the need of inspecting the content

• The number of “thumbs up” that appear in the bottom of a post allow efficient social navigation

33

34

Conclusion

• Information overload is a real problem

• This study found that these descriptors might be used to signal the quality of multimedia content, thus supporting self-promotion

• How by leveraging the knowledge of peers in social networks– take advantage of the specific properties of social networks – complex and difficult to obtain with other techniques

• Thus, a future improvement of the social tagging mechanism needs to consider the social proximity among peers, e.g., by applying weights

35

Comments

• Goal: Wise designerBehaviors– Social Games With A Purpose (reduce the barriers & raise the incentives)

– Individual’s thoughts or actions are affected by other people, entertainment and exploitation of personal benefits (mutual help, social influence)

– Tags Are Not Just for Retrieval• Study 2: Single word statements (hard time)• Study 3: comments – the promotion of personal content

Technology– Exploit Innovation or mixed tech. to capture background knowledge and to– leave meaningful comments / tags

36