sca2013 presentation: a web-based content analysis tool
DESCRIPTION
This is a presentation at SCA2013, Karlsruhe, Germany. This shows the design (architecture, database, sketch, wireframe, prototype) of a simple web-based tool that supports asynchronous collaboration among researchers when conducting content analysis on qualitative social media data.TRANSCRIPT
A Web-Based Tool for Collaborative Social Media Data Analysis
Xin Chen, Mihaela Vorvoreanu, Krishna Madhavan{chen654, mihaela, cm}@purdue.edu
Motivation
Social Media Discourse
Motivation
Hidden Insights On Human Behaviors & Social Phenomenon
Human generated textual data on social media are:
Qualitative Data
Large-scale Data
Motivation
Human generated textual data on social media are:
Qualitative Data
Large-scale Data
requires qualitative interpretation
Motivation
Human generated textual data on social media are:
Qualitative Data
Large-scale Data
requires qualitative interpretation
requires large-scale data mining techniques
Motivation
Goal
To build a tool that:
Goal
Acquire social media data.
To build a tool that:
Goal
Acquire social media data.
Integrate qualitative content analysis and data mining techniques to analyze textual data on social media.
To build a tool that:
Goal
Acquire social media data.
Integrate qualitative content analysis and data mining techniques to analyze textual data on social media.
Support asynchronous collaboration among researchers.
To build a tool that:
Social Media Analytics & Monitoring Tools
Existing Tools
Focus on marketing
Do not usually incorporate human input
Qualitative Analysis Tools
Existing Tools
Complicated to use
Expensive
Do not acquire social media data
Social Media Content
API or Web Crawler
Researchers
Computation Server
Web UI
Web Server
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Data Server
Social Media Content
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Data Server
Computation Server
Web UI
Web Server
Social Media Content
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Twitter search API
Data Server
Computation Server
Web UI
Web Server
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQLClassification & Detection Modeling
Inter-rater Agreement Computation
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Sample tweets for researchers to analyze
Send results back to data server
Communicate with computation server
Social Media Content
Data Server
API or Web Crawler
Researchers
SWAB (Social Web Analytics Buddy)
1
2
3
4
5
Computation Server
Web UI
Web Server
MySQL
Web UI DesignSketch
Web UI DesignWireframes: the “Overview” tab
Project title
Web UI Design
User account
Wireframes: the “Overview” tab
Web UI Design
Collaborator List
Wireframes: the “Overview” tab
Web UI Design
Datasource
Wireframes: the “Overview” tab
Web UI Design
Multiple datasets streamed using different criteria.
Wireframes: the “Overview” tab
Web UI Design
Export data and graphs.
Wireframes: the “Overview” tab
Web UI Design
Multiple visualizations and charts to provide data overview.
Wireframes: the “Overview” tab
Web UI DesignWireframes: the “Analyze” tab
Themes emerged from exploring the data.
Web UI DesignWireframes: the “Analyze” tab
Choose Sample size.
Web UI DesignWireframes: the “Analyze” tab
Analyze tweets and write comment.
Web UI DesignWireframes: the “Result” tab
All researchers’ results are aggregated in the background. Collaboration happens asynchronously. Reliability measures are computed.
Web UI DesignWireframes: the “Result” tab
Classification models can be trained based on the qualitative input.
Web UI DesignWireframes: the “Model Application” tab
Apply the trained model to a new dataset to detect similar data as in dataset1 from dataset2.
Web UI DesignWireframes: the “Model Application” tab
Choose how to explore the detected data from the new dataset: view list of tweets, user accounts, or geomap.
Simple Working Prototype of “Analyze” Tab Feature
!Demo
Future Work
Future Work
Design features to better support data exploration.
Future Work
Design features to better support data exploration.
Explore NoSQL database to handle large datasets.
Future Work
Design features to better support data exploration.
Explore NoSQL database to handle large datasets.
Implement more sophisticated data mining and visualization features.
Thank you!
Q & A