Abstract
This paper proposes a customized Social TV platform for users of Social Network Services (SNS) and IPTV. By moving beyond simple security and basic social features, the platform offers personalized TV channel recommendations based on user preferences. Utilizing Hadoop MapReduce, the system analyzes keywords from a user's SNS activity to recommend relevant television channels. The research compares information across five different TV programs in descending order of priority and provides a customized TV schedule. Through a smartphone application, users can access real-time information, interact with others, and manage TV channel orders, creating a synchronized social viewing experience.
Introduction: The Shift to Social TV
Traditional television viewing is transitioning from a passive, solitary activity into "Social TV"—a shared digital experience. However, current Social TV services often lack deep personalization, typically offering only basic real-time commenting or content restricted to a single broadcaster. With the proliferation of hundreds of channels via IP networks, content discovery has become a significant hurdle for users. This research addresses the challenge of content personalization by bridging the gap between a user's social interests and broadcast data to create a truly user-customized viewing experience.
Methodology: Personalized Recommendation Architecture
The proposed platform employs a full-stack architecture designed for high-volume data processing and real-time interaction:
- Data Analysis: The system leverages Hadoop MapReduce to process large-scale SNS data from Facebook and Twitter. It extracts frequently used keywords to identify core user interests.
- Recommendation Logic:
- Batch Analysis (Case E1): A daily process that identifies the intersection between a user’s social media keywords and broadcast channel metadata.
- Real-time Analysis (Case E2): A tracker that monitors channel clicks to dynamically adjust program priority.
- System Architecture: The backend utilizes a MySQL database for storage, Apache Tomcat as the web server, and the Spring Framework for application management.
- Client Interface: An Android-based mobile application serves as the primary controller, communicating with a Java-based Set-Top Box (STB) to manage the TV display and social feeds.
Implementation and Results
The research successfully integrated SNS authentication, IPTV control, and big data analysis into a unified mobile interface. A key feature is the "Social TV Schedule," which allows users to view a dynamic guide based on their friends' activities and recommendations.
The system effectively generates a "My TV Schedule," filtering upcoming programs based on keyword intersections analyzed via Hadoop. Users can change TV channels directly from their smartphones and receive real-time notifications for recommended programs. Furthermore, the platform facilitates a "Watch Together" flow, where users can invite friends via SNS, and the friend's STB can automatically synchronize to the same channel.
Conclusion and Future Work
While the platform demonstrates a robust architecture for personalized Social TV, certain limitations remain. The effectiveness of recommendations is tied to the user's social media activity level, and the current daily batch processing for keyword analysis may lag behind real-time trends. Future work will focus on refining the recommendation algorithms, addressing privacy concerns regarding SNS data analysis, and expanding compatibility with a wider range of Set-Top Box hardware.
Summary updated