TL;DR
This paper introduces SteemOps, a comprehensive dataset of over 900 million operations from the Steemit blockchain, designed to bridge the gap between social media analysis and blockchain research. By extracting and categorizing key user activities, SteemOps enables deeper insights into the unique interplay of social interactions and decentralized governance on blockchain-based platforms.
What Problem Does This Paper Address?
Blockchain-based social media platforms like Steemit offer a novel paradigm where user interactions are recorded on an immutable ledger, creating a rich source of data. However, analyzing this data presents significant challenges. It requires specialized knowledge of both blockchain technology and software engineering, and extracting meaningful information from the sheer volume and variety of operations is a substantial undertaking. Existing research often focuses on either social network analysis or blockchain system analysis in isolation, missing the crucial intersection where user behavior directly influences decentralized mechanisms. This paper addresses the need for a structured, accessible dataset that facilitates cross-domain research by capturing and organizing these integrated operations.
Methods
The core of this paper is the creation of the SteemOps dataset. The authors collected data from over 38 million blocks on the Steem blockchain spanning a 45-month period (March 2016 to November 2019). They then developed a process to extract and categorize ten key types of user operations. These operations were grouped into three main sub-datasets:
- Social-Network Operation Dataset (SOD): Captures user interactions like posting, commenting, voting, and following.
- Witness-Election Operation Dataset (WOD): Records operations related to the election and management of block producers (witnesses) in Steemit's Delegated Proof of Stake (DPoS) consensus mechanism.
- Value-Transfer Operation Dataset (VOD): Encompasses transactions involving the transfer of cryptocurrencies (STEEM, SBD) and their conversion into Steem Power (VESTS).
The paper details the schema for each operation type, providing a clear understanding of the data's structure and content.
Datasets
The primary dataset introduced in this paper is SteemOps. It comprises over 900 million operations extracted from the Steem blockchain, covering a period from March 2016 to November 2019. This dataset is further divided into three sub-datasets:
- Social-Network Operation Dataset (SOD)
- Witness-Election Operation Dataset (WOD)
- Value-Transfer Operation Dataset (VOD)
SteemOps is presented as a valuable resource for researchers, offering a more structured and analyzed alternative to raw blockchain data.
Key Insights and Ideas
- The Interconnectedness of Social and Blockchain Operations: Steemit exemplifies how social interactions (posting, voting) are deeply intertwined with blockchain governance (witness elections) and economic activities (value transfers). This integration opens up new avenues for research that were previously difficult to explore.
- Social Operations Dominate Usage: Preliminary analysis reveals that social-network operations, particularly "vote" operations, are the most frequently performed, indicating that users primarily engage with Steemit for its social media functionalities.
- DPoS as a Scalability Solution: The paper highlights Steemit's use of Delegated Proof of Stake (DPoS) as a mechanism to achieve higher transaction throughput compared to Proof-of-Work (PoW) systems like Bitcoin, making it more suitable for social media applications.
- Potential for Cross-Domain Research: SteemOps is designed to facilitate research in areas such as decentralization analysis, cryptocurrency transfer patterns, performance benchmarking of blockchain systems, community and user behavior analysis, and bot detection.
Key Results
The paper's main result is the creation and release of the SteemOps dataset, containing over 900 million operations. Preliminary analysis shows:
- Social-network operations constitute the largest portion of user activity.
- "Vote" is the most common social-network operation.
- "Transfer" is the most frequent value-transfer operation.
- Witness-election operations are performed less frequently compared to social and value-transfer operations.
The paper also provides detailed schemas for ten key operation types, enabling researchers to understand and utilize the dataset effectively.
Limitations and Future Work
Limitations: * The dataset covers a specific time period (March 2016 - November 2019). While extensive, it does not include more recent activity on the Steem blockchain. * The paper acknowledges that the Steem blockchain generates over thirty different types of operations, and SteemOps focuses on ten key ones. This means some niche operations are not included.
Future Work: The authors suggest several future research directions enabled by SteemOps:
- Decentralization Analysis: Further analysis of decentralization from the perspectives of voters, authors, and proxies.
- Cryptocurrency Transfer Analysis: Utilizing the value-transfer data for tracing transactions, analyzing market manipulation, and understanding token ecosystems.
- Performance Benchmarking: Using SteemOps as a realistic workload for evaluating new blockchain systems, especially those employing DPoS.
- Social Network Analysis: Investigating community structures, user behavior, and the impact of blockchain on social dynamics.
- Curation Mechanism Analysis: Studying how stake-weighted voting influences content curation and reward distribution.
- Bot Detection: Leveraging memo information and network correlations to identify and understand bot activity on the platform.
Summary updated