Transcript
Page 1: Vine Measurement Study

ELIS – Multimedia Lab

Baptist Vandersmissen, Fréderic Godin, Abhineshwar Tomar, Wesley De Neve, Rik Van de Walle

Multimedia Lab

Department of Electronics and Information Systems

Faculty of Engineering and Architecture

Ghent University – iMinds

Belgium

The Rise of Mobile and Social Short-Form Video: An In-depth Measurement Study of Vine

Page 2: Vine Measurement Study

2/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 3: Vine Measurement Study

3/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 4: Vine Measurement Study

4/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Introduction

Widespread use of smartphones Popular social video platforms

Expected that by 2017 more than two-third of mobile data traffic will be video

Huge amount of video data Internet revenue model based on ads

What content is popular? Automatic content understanding?

Page 5: Vine Measurement Study

5/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Introduction

annotated?

noisy?

people?

event-related?

First step is measurement study

Numerous challenges and ambiguities

Huge amount of video data

Page 6: Vine Measurement Study

6/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

State of the Art

1. Content and Audience Analysis

Answering the who, what and when questions

2. Popularity Analysis and Prediction Analysis of the popularity distribution regarding videos

3. Social Sensing Analysis of cultural behaviour, finding points of interests or movement patterns

Page 7: Vine Measurement Study

7/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 8: Vine Measurement Study

8/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

What is Vine?

Social & mobile video platform, established in January 2013. Create & distribute looping videos of up to 6 seconds. Acquired by Twitter and now plus 40 million users. Max video length resembles Twitter’s character limitation.

Page 9: Vine Measurement Study

9/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

What is Vine?

Page 10: Vine Measurement Study

10/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

What is Vine?

Page 11: Vine Measurement Study

11/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

What is Vine?

Page 12: Vine Measurement Study

12/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

What is Vine?

Page 13: Vine Measurement Study

13/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Why Vine?

New and first large scale platform focusing on mobile video Potential to become social news platform

Strong connection with Twitter Open nature and easy accessible through URLs in tweets

Page 14: Vine Measurement Study

14/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 15: Vine Measurement Study

15/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Data Collection

No official Vine API Use of Twitter as a gateway to access Vine videos Use of unofficial Vine API methods to extract metadata

Date fetched Date created

Tweet

Description

Location

Comments

Revines Likes

Explicit

User User

Hashtags Mentions

Followers

Following

Posts

Page 16: Vine Measurement Study

16/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Data Collection

Tracked keyword “vine” on Twitter Streaming API Collected 851,039 tweets containing a Vine URL Originating from 365,188 Twitter users

425,971 unique Vine videos fetched Created by 193,355 different Vine users

Page 17: Vine Measurement Study

17/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 18: Vine Measurement Study

18/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Metadata Characteristics

Vine videos can have a description containing #hashtags or @mentions

34% contains at least one hashtag (only 8% on Twitter) 9% contains at least one mention

Page 19: Vine Measurement Study

19/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Metadata Characteristics

Matching of hashtags with WordNet categories

11% of hashtags can be matched to a WordNet synset

Page 20: Vine Measurement Study

20/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Channel Characteristics

Vine videos can be added to a channel (i.e., a category)

Most popular categories are comedy and music Similar to YouTube’s popular categories which are entertainment and music

Channel Count (%)

comedy 60.96% music 8.93% wierd 5.27% dogs 4.46% cats 3.25%

family 2.74%

art & experimental 2.74% sports 1.88% food 1.61% special fx 1.52%

nature 1.47% urban 1.41% scary 1.27% beauty & fashion 1.09% news & politics 0.75% health & fitness 0.65%

Page 21: Vine Measurement Study

21/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Popularity Characteristics

Emperical analysis of (popular) content Popularity is measured based on revines, likes, comment count Popular content is not event- or news-related However Many “big” events are described by Vine videos These can for example be used to enhance textual news articles

Page 22: Vine Measurement Study

22/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Event-related Example

Page 23: Vine Measurement Study

23/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Event-related Example

Page 24: Vine Measurement Study

24/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Popularity Aspects

Analyze Vine video popularity based on content-agnostic factors: 1. Twitter shares: number of tweets sharing same Vine video

2. Twitter exposure: sum of followers of these Twitter users

3. Vine exposure: number of followers of Vine creator

Page 25: Vine Measurement Study

25/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Popularity Aspects

Hypothesis: high amount of shares on Twitter can be linked to more popular Vine video Correlations between factors measured based on Pearson method

Twitter shares Twitter exposure Vine exposure

Likes 0.29 0.07 0.60

Revines 0.32 0.08 0.46

Comments 0.31 0.07 0.39

Low correlation between amount of shares on Twitter and Vine popularity measures.

Page 26: Vine Measurement Study

26/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Popularity Aspects

Hypothesis: number of shares on Twitter shortly after the video’s creation can be linked to a higher video popularity

Figure depicts number of shares on Twitter after one hour and likes on Vine after one week

Page 27: Vine Measurement Study

27/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Attention Aspects

Analyze amount of user attention received by a Vine video Hypothesis: attention span is short and peaks shortly after its creation

Evolution of number of likes on Vine

Page 28: Vine Measurement Study

28/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Attention Aspects

Evolution of number of shares on Twitter

Analyze amount of user attention received by a Vine video Hypothesis: attention span is short and peaks shortly after its creation

Page 29: Vine Measurement Study

29/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 30: Vine Measurement Study

30/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Conclusions

First large-scale academic measurement study on Vine Collection of ~ 850,000 tweets containing ~425,000 unique Vine videos 34% of Vine videos in our dataset contain atleast one hashtag Content is highly personal, created for entertainment purposes Twitter cannot be used as a measure for the popularity of a Vine video User attention peaks shortly after the creation but continues even after weeks

Page 31: Vine Measurement Study

31/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

1. Introduction

2. Vine

3. Data Collection

4. Characteristics & Popularity Aspects

5. Conclusions

6. Future Research Directions

Outline

Page 32: Vine Measurement Study

32/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Future Research Directions

Investigate what portion of Vine videos is news- or event-related How can these be used to enhance news stories. Create a (geo-based) hashtag recommendation and categorization system Make use of content (visual) and context (textual) information Leverage user mentions to detect community formation By using mentions in description and faces in video

Page 33: Vine Measurement Study

33/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Future Research Directions

Create personalized television channels Based on hashtags and user preferences Compare short-form video usage on Vine with short-form video usage on Instagram and YouTube

Page 34: Vine Measurement Study

34/34

ELIS – Multimedia Lab

Short-form Video on Vine: Overview and Applications Baptist Vandersmissen

01/04/2014

Questions?


Top Related