This thesis proposes a way to contribute to the data generation process for the Semantic Web. It picks up the idea of Extreme Tagging Systems (ETS), which are an extension to well-known Collaborative Tagging Systems (CTS) like Delicious and Flickr, where users can associate keywords (or "tags") with resources like websites or pictures. ETS allow for more flexibility because they let the users not only tag resources but also tags themselves and even let the users specify the relation between two tags.
A concept for an ETS implementation as a game is developed. This game is primarily meant to be fun for the users, by this trying to motivate them to continue to use the system. But at the same time through the game the users do work and help to solve problems that cannot yet be automated - in this case problems of Extreme Tagging. Such games that are both fun and useful are called Games With a Purpose (GWAPs).
The concept for an ETS as a GWAP is implemented as well as an ETS
as a website. The work then compares both implementations in respect to user motivation (fun), efficiency and productivity. It shows that the implementation as game outperforms the website in all examined aspects. For example the evaluation reveals that a user of the website is expected to generate 33 tags during her lifetime use of the system whereas a player of the game is expected to generate 53 tags. This thesis also describes several points that could further increase the advance of the game and tries to encourage further research and work being done in this direction as it seems to be a promising approach.
Table of Contents
1 Introduction
1.1 Structure of this Work
2 Background and Related Work
2.1 Background
2.1.1 Collaborative Tagging
2.1.2 Extreme Tagging
2.1.3 Games With a Purpose
2.1.4 Mathematical Model For GWAPs
2.2 Related Work
2.2.1 Semantics From User Data
2.2.2 Games With a Purpose
3 Problem
3.1 eXTS
3.1.1 The First Version of eXTS
3.1.2 The Second Version of eXTS
3.2 Theoretical Problems
3.3 Problems derived from eXTS
3.3.1 Setup of the Evaluation Run
3.3.2 Results of the Evaluation Run
3.3.3 Problems Derived From the Results of the Evaluation
3.4 Purpose
4 Proposed Solution
4.1 The Minigames
4.1.1 Racing Game
4.1.2 Falling Words Game
4.1.3 Door Opening Game
4.2 Order of Games
4.3 The Overall Flow of the Game
4.4 Further Incentives
5 Implementation
5.1 General Considerations
5.2 Data Representation
5.3 Frameworks
5.3.1 jMonkeyEngine 3.0
5.3.2 Golden T Game Engine
5.3.3 Game Gardens Framework
5.4 Implementation of the Racing Game
5.4.1 The Distributed Object
5.4.2 The Game Manager
5.4.3 The Game Controller and the View Classes
5.4.4 Computer Controlled Players
5.5 Implementation of the Falling Words Game and the Door Opening Game
6 Mathematical Modelling
6.1 A Formal Model for the Racing Game
6.1.1 The Game Problem Domain
6.1.2 The Game Rules
6.1.3 The Flow of the Game
7 Evaluation
7.1 Setup of the Evaluation Run
7.2 General Conclusions of the Evaluation
7.3 Analysis of the Productivity
7.4 Evaluation of the Questionnaire
7.4.1 General Aspects
7.4.2 Evaluation of the Fun Factor
7.4.3 Evaluation of the Ease of Use
7.4.4 Miscellaneous Results of the Questionnaire
8 Summary and Conclusion
8.1 Summary
8.2 Future Work
A Running the Code
A.1 Server Side
A.2 Client Side
A.3 Playing the Game
B Additional Figures
Objectives and Topics
This master's thesis aims to improve data generation for the Semantic Web by developing a "Game With a Purpose" (GWAP) as an alternative to traditional tagging systems. The research seeks to overcome the "chicken-and-egg" problem of the Semantic Web by leveraging human intelligence through gamification to collect high-quality semantic data, such as tags and entity relations, while maintaining user engagement.
- Extreme Tagging Systems (ETS) as an extension of Collaborative Tagging.
- The concept and design of Games With a Purpose (GWAPs) for semantic data creation.
- Implementation of a multi-stage, client-server game architecture.
- Comparative evaluation of game-based tagging versus traditional website-based tagging.
- Formal mathematical modeling of social games and their application to tagging tasks.
Excerpt from the Book
4.1.1 Racing Game
The racing game is a multiplayer game that requires at least two players playing it at the same time (at least one of the players must be a human player, the others can be computer controlled players, see section 5.4.4). It is a hybrid game regarding the strategy players follow to reach the winning conditions. This means it is both competitive and collaborative. Competitive here means that every player fights for her own progress in the game to be the first to reach the finishing line. Collaborative means that a player needs to cooperate with other players to win the game, too. The purpose of this game (for me) is to let the players generate new tags for given entities and implicitly strengthen existing entity-tag-pairs. I call an entity-tag-pair being “implicitly” strengthened if it is being generated by a player but was already present in the database before. This means it has a higher weight1 as more than one player produced this pair.
As I aim to collect objective information (because that is the kind of information that should be represented in an ontology) I choose the racing game to be a slightly modified output-agreement game as described in section 2.1.3.1 - modified in the way that it allows more than two players but requires only one. And according to [5] an output-agreement game should be used to collect objective information.
Summary of Chapters
1 Introduction: Discusses the "chicken-and-egg" problem of the Semantic Web and proposes Games With a Purpose as a solution for scalable semantic data creation.
2 Background and Related Work: Introduces collaborative tagging, Extreme Tagging Systems, and the theoretical foundation of Games With a Purpose.
3 Problem: Details the limitations of current eXTS prototypes and outlines the challenges identified through an initial evaluation, such as low participation and difficulty in defining relations.
4 Proposed Solution: Describes the design of a three-staged game approach (racing, falling words, and door opening) to address the identified data generation problems.
5 Implementation: Covers the technical realization of the racing game using Java and the Game Gardens framework, including system architecture and client-server communication.
6 Mathematical Modelling: Provides a formal framework for modeling the racing game based on the definitions for social games.
7 Evaluation: Presents a comparative study between the eXTS game and the website, demonstrating the superiority of the game-based approach in terms of productivity and user engagement.
8 Summary and Conclusion: Consolidates the findings of the thesis and provides recommendations for future research and improvements.
Keywords
Semantic Web, Games With a Purpose, GWAP, Extreme Tagging Systems, ETS, Collaborative Tagging, Gamification, Human Computation, Ontology Generation, User Motivation, Game Design, Data Quality, Evaluation, Multi-player Games, Human-Computer Interaction
Frequently Asked Questions
What is the core purpose of this research?
The thesis aims to address the data generation bottleneck of the Semantic Web by transforming tedious manual tagging tasks into an engaging "Game With a Purpose" (GWAP).
What are the central topics covered in this work?
The work covers Extreme Tagging Systems (ETS), the theory of Gamification, formal mathematical modeling of games, and the implementation and evaluation of a specific tagging game prototype.
What is the main research question or objective?
The objective is to determine if a GWAP-based tagging system can achieve higher user motivation, productivity, and data quality compared to a traditional tagging website.
Which scientific methods are employed?
The author uses a combination of literature analysis, software prototyping, mathematical modeling based on social game frameworks, and empirical evaluation (test runs and questionnaires).
What does the main body of the work focus on?
The main part focuses on the concept of the three-staged game, the technical implementation of the racing game, and the comparative performance evaluation against an existing eXTS website.
Which keywords characterize this work?
Key terms include Semantic Web, GWAP, Extreme Tagging, Gamification, and User Motivation.
How does the "racing game" encourage collaboration?
Players must enter tags that match those entered by other players or existing high-weight tags in the database to earn points and progress, effectively forcing agreement.
Why is the "door opening game" placed last?
It is placed last because it requires a more sophisticated understanding of relations; by placing it after the other games, players have built up a pool of trusted word-tag pairs to work with.
- Quote paper
- Dennis Hartrampf (Author), 2012, An Extreme Tagging System as a Game With a Purpose, Munich, GRIN Verlag, https://www.hausarbeiten.de/document/200644
