TL;DR
This paper introduces DCDChain, a novel blockchain-based architecture for credible digital copyright detection. It addresses the issue of untrustworthy detection parties by enabling media providers to verify detection results and using smart contracts for arbitration, ensuring both security and efficiency in copyright protection.
What Problem Does This Paper Address?
The digital age has made it incredibly easy to copy and distribute media, leading to widespread copyright infringement. While plagiarism detection tools exist, their reliability is often questioned. Detection agencies might be untrustworthy, or their tools could produce incorrect results, leaving copyright holders vulnerable. Existing blockchain solutions for copyright primarily focus on secure storage or trading of digital assets, but they haven't tackled the critical problem of ensuring the credibility of the copyright detection process itself. This gap leaves room for manipulation and doubt, undermining the effectiveness of copyright protection.
Methods
DCDChain proposes a "detect-verify-and-arbitrate" mechanism built upon blockchain technology. Here's how it works:
- Off-Chain Detection: A detection agency (DA) performs copyright detection on digital media locally. This is computationally intensive, so it's done off-chain.
- Blockchain Storage: The detection results, along with hash values of the media, are uploaded to the blockchain. This ensures tamper-resistance and public accessibility.
- Verification: Media providers (MPs) can access these blockchain records to verify the correctness of the detection results.
- Smart Contract Arbitration: If an MP disputes a detection result, they can appeal to a smart contract deployed on the blockchain. This smart contract then arbitrates the dispute by re-verifying the detection results on-chain.
- Hybrid Hashing: The system utilizes both secure hash algorithms (like SHA256) for unique media identification and locality-sensitive hashing (LSH) for efficient similarity comparisons, crucial for detecting partial piracy.
- IPFS Integration: For storing the actual media content, the InterPlanetary File System (IPFS) is used. This is a distributed file system that provides content addressing and tamper resistance, with only hash values of the media stored on the blockchain.
- Incentive Mechanism: Financial incentives (service fees and deposits) are incorporated to encourage honest behavior from both detection agencies and media providers.
Datasets
The paper uses four corpora for text similarity experiments:
- Clinical Document Architecture (CDA): Medical records.
- Douban: Public dialogue corpus.
- Sougou News: News articles from various categories.
- Sina Weibo: Microblogs.
These corpora were used to evaluate the relationship between Hamming distances (from LSH) and text similarities, and to construct regression models for predicting piracy.
Key Insights and Ideas
- Blockchain for Credibility, Not Just Storage: The core innovation is using blockchain not just for immutable storage but as a trust layer for the process of copyright detection.
- Decentralized Arbitration: Smart contracts provide an automated and transparent way to resolve disputes, removing reliance on a single trusted authority.
- Hybrid Hashing for Efficiency and Accuracy: Combining secure hashing with LSH allows for both precise identification and efficient detection of similar, potentially pirated, content.
- Incentivizing Honesty: The economic model, where deposits can be lost for dishonest behavior and fees are earned for correct detection, is crucial for system integrity.
- IPFS for Scalability: Using IPFS for media storage addresses the scalability limitations of storing large files directly on the blockchain.
Key Results
- Credible Detection: The detect-verify-and-arbitrate mechanism is shown to guarantee the credibility of copyright detection.
- Security and Efficiency: Security analysis and experimental simulations demonstrate that DCDChain is credible, secure, and efficient.
- Linear Relationship: Experiments showed a roughly linear relationship between Hamming distances and text similarities across various corpora, validating the use of LSH for predicting piracy.
- Smart Contract Performance: Execution times for smart contract functions (hashIDJudge and lshvJudge) were measured on local Ganache (around 60ms) and Rinkeby testnet (around 15s), indicating feasibility.
Limitations and Future Work
- LSH Algorithm Improvement: The authors acknowledge that future work should focus on designing more effective locality-sensitive hash algorithms tailored for different types of digital media.
- Accumulation of Legitimate Data: The security analysis assumes that there are already enough legitimate hash values stored on the blockchain, which is an "accumulation process." This implies that the system's robustness improves over time as more legitimate data is added.
Summary updated