Collaborative tagging as a tripartite network

R. Lambiotte,M. Ausloos

DOI: https://doi.org/10.1007/11758532_152

2005-12-29

Abstract:We describe online collaborative communities by tripartite networks, the nodes being persons, items and tags. We introduce projection methods in order to uncover the structures of the networks, i.e. communities of users, genre families... To do so, we focus on the correlations between the nodes, depending on their profiles, and use percolation techniques that consist in removing less correlated links and observing the shaping of disconnected islands. The structuring of the network is visualised by using a tree representation. The notion of diversity in the system is also discussed.

Data Structures and Algorithms,Digital Libraries

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to extract useful information structures from online collaborative communities, especially to identify collective behaviors by analyzing the complex relationships among users, items and tags, and to construct a hierarchical network structure. Specifically, the paper focuses on:

1. **Understanding the structure of collaborative websites**:
- The paper describes the online collaborative community as a tripartite network, with nodes being people (users), items (such as music, articles, etc.), and tags. These tags are descriptive keywords given to items by users.

2. **Introducing projection methods to reveal network structures**:
- In order to reveal the structures in the network, such as user communities or item categories, the author introduces projection methods. These methods analyze the correlations between nodes, remove links with low relevance, and observe the formed disconnected islands, thereby revealing the hierarchical structure of the network.

3. **Using percolation techniques**:
- Percolation techniques are used to remove unimportant links and retain highly relevant parts, thus helping to discover isolated subsets in the network. This helps to understand the interest and behavior patterns of different user groups.

4. **Defining diversity and hierarchical structures**:
- The author discusses the concept of diversity in the system and proposes a method to measure the diversity of user interests. Through this method, the distribution of user interests can be better understood and compared with traditional classification methods such as taxonomy.

5. **Application areas**:
- The paper mainly analyzes the data of music - sharing websites (such as audioscrobbler.com and musicmobs.com) and scientific literature - sharing websites (such as citeulike.com). Through these data, the author shows how to extract meaningful structural information from user behaviors.

In summary, this paper aims to reveal collective behaviors in online collaborative communities by analyzing the complex relationships among users, items and tags, and to provide a new method to automatically construct the hierarchical structure of content classification. This not only helps to understand user behavior patterns, but also can provide theoretical support for personalized recommendation systems.