Quantum and classical models compared on dark-web link prediction
Original title: Quantum-Classical Link Prediction on Dark Web Hyperlink Graphs
A four-qubit classifier stayed near chance under degree control, while VGAE reached mean ROC-AUC about 0.91.
The paper studies HTML and hyperlinks from an offline Tor crawl of 383 onion hosts and 2,067 directed cross-host links, combining semantic features with degree-controlled quantum-classical link prediction. Two directory groups account for 94.8% of observed host links. Directed preferential attachment scores ROC-AUC 0.976 against random non-edges but about 0.550 under source-conditioned target-degree matching. On disjoint pairs, VGAE reaches 0.906±0.016, while the best exact four-qubit variational classifier mean is 0.549±0.023, close to classical controls on the same inputs.
Unconfirmed: Metrics are the authors' reported results and are not independently verified here.
- The crawl covers 383 onion hosts and 2,067 directed cross-host links.
- Two directory groups emit 94.8% of observed host links.
- GAE and VGAE score 0.899±0.016 and 0.906±0.016 on disjoint pairs.
- The VQC is tested with two encodings across 20 splits and 1,024-shot simulation.
- RBF SVC and a 16-parameter MLP score 0.551±0.016 and 0.573±0.029.
From the original article
1 min read at arxiv.orgWe analyze HTML content and hyperlinks from an offline Tor crawl containing 383 onion hosts and 2,067 directed cross-host links. The study combines semantic analysis with degree-controlled quantum--classical link prediction. Two directory groups emit 94.
By Vojtěch Novák · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.