tomas et al

Multimedia Tools and Applications (2019) Volume 78:25539–25568

Abstract

  • SAM Overview: SAM (Socialising Around Media) is a social media platform aimed at enhancing viewers' experiences of video content in traditional settings (like living rooms).
  • Key Functionalities:
    • Semantic Analysis
    • Context Awareness
    • Dynamic Communities
  • Objective: Evaluate system functionalities through dataset and user evaluations to determine effectiveness and efficiency.

Keywords

  • Social TV
  • Second Screen
  • Semantic Analysis
  • Entity Linking
  • Sentiment Analysis
  • Context Awareness
  • Community Detection
  • Dynamic Communities

Introduction

  • Shift in Media Interaction: The evolution from passive to proactive media consumption due to the advent of consumer-centric Internet devices, especially smartphones.
  • Second Screen Usage: Definition as the use of a mobile device to enhance the viewing experience of primary content (e.g., TV shows).
  • Challenges: Lack of standard protocols for second screen applications, leading to a fragmented experience.
  • Goals of SAM:
    • Provide open and standardized technical means for characterising, discovering, and syndicating media assets
    • Aid in the creation of interactive, socially-oriented experiences for media consumption.

Benefits of SAM

  • For Businesses: Enables engagement through dynamic content syndication, real-time statistics tracking, and customer insights.
  • For Users: Offers enhanced interactivity with media, providing contextual information tailored to user interests.
  • Business Objective: Increase audience engagement through improved user experiences.

Three Key Functionalities of SAM

  1. Semantic Analysis:

    • Involves natural language processing technologies to enhance content understanding.
    • Primary features include:
      • Sentiment Analysis: Gauges user emotions and opinions through comments, aiding in clustering users not only demographically.
      • Entity Linking: Connects content with related entities and external knowledge bases (e.g., Wikipedia).

  2. Context Awareness:

    • Manages user data to create personalized recommendations based on interactions and preferences.
    • Aggregates contextual information effectively to enhance user experience.

  3. Dynamic Communities:

    • Analyses communication patterns to form user communities and manage memberships dynamically.

Evaluation Methodology

  • The study includes intrinsic evaluations for each functionality and an extrinsic evaluation of the whole platform with participation from around sixty users.

Related Work

Social Television and Second Screen

  • The role of digital technology and social media in facilitating social television, which leverages multi-screen interactions to boost viewer engagement.
  • Research References:
    • Courtois and D’heer’s study on multi-screen experiences.
    • Visual attention studies concerning second screen interactions.
  • SAM's unique features identified among competitors in the second screen market (comparison table included).

Individual Functionalities

  1. Entity Linking: Matches entity mentions in text to a knowledge base.

    • Example: Linking “Al Pacino” to its Wikipedia entry.
    • Comparisons made with traditional Named Entity Recognition (NER).

  2. Sentiment Analysis: Focuses on extracting subjective information from texts.

    • Both global and aspect-based sentiment analysis approaches are discussed.
    • Applications include determining user opinions on media assets.

  3. Context Awareness: Captures and utilizes user contextual data for personalized experiences.

    • Related studies highlight how user location and behavior enhance viewing experiences.
    • Novel graph analysis methods utilized in SAM’s context management.

  4. Dynamic Communities: Discusses the user-driven creation of online communities via enhanced interactions through SAM.

SAM Platform Architecture

  • SAM structured as a modular system with four layers:
    • Data Management: Handles media asset storage and retrieval, including cloud storage and content gateways.
    • Control: Manages core functionalities such as semantic services, community dynamics, and identity protection.
    • Communication: Interconnects various components through an interconnection bus.
    • Interaction: Presents front-end components for business and user interactions, encompassing the Marketplace, Linker, Analytics, and Dashboard.

Business Use Cases

  • SAM aims to streamline the content creation process for second screen experiences across various media types.
  • Content managers can utilize the Marketplace component for media asset management.
  • Analytics tools enable data visualization of user interactions and sentiments.

End User Use Cases

  • Users engage seamlessly with media content across first and second screens.
  • SAM's applications allow for interactive features complementing live TV content, enhancing the viewers' engagement through insights and additional context.

Improving the Second Screen Experience

Semantic Analysis

  • Key component for content enrichment through various techniques such as:
    • Sentiment analysis
    • Entity linking
    • Asset editing.
  • The role of ontology in enhancing the semantic representation of media assets.

Context Awareness

  • Contextual information is leveraged to prioritize video content and suggest relevant second screen assets.
  • Utilizes a Neo4j graph database for efficient data representation and user interactions.

Dynamic Communities

  • User communities dynamically formed based on interests and communication patterns.
  • Algorithms for community detection implemented within SAM to evaluate user interactions continuously.

Evaluation of SAM Functionalities

Semantic Analysis Evaluation Results

  • Precision rates and metrics for entity linking and sentiment analysis detailed, showcasing the robustness of SAM’s semantic functionalities.

Context Awareness Evaluation Results

  • Comparative analysis of SAM’s recommendation effectiveness against other machine learning based baselines.

Dynamic Communities Evaluation Results

  • Evaluating the effectiveness of the BigCLAM algorithm for community detection within SAM environments.

User Evaluation and Feedback

Evaluation Setup

  • Conducted among participants aged 13 to 17 in a classroom setting to measure acceptance and enjoyment.
  • Four evaluation rounds focusing on different functionalities of the SAM application.

Results Overview

  • Participants found SAM’s application useful for discovering topics related to video content.
  • The majority appreciated the dynamic communities, indicating a desire for continued use in school and at home.

Conclusion

  • SAM presents a viable system for creating interactive second screen experiences, emphasizing user engagement through semantic analysis, context awareness, and community dynamics.
  • Future research should focus on creating datasets for deeper exploration of complex workflows and longitudinal studies for predetermined areas of interest.

Acknowledgments

  • Funding sources include the European Commission, the Spanish Government, and the Generalitat Valenciana.