Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Prediction and Clustering in Signed Networks: A Local to Global Perspective
Authors: Kai-Yang Chiang, Cho-Jui Hsieh, Nagarajan Natarajan, Inderjit S. Dhillon, Ambuj Tewari
JMLR 2014 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We now present experimental results for sign prediction and clustering using our proposed methods. For sign prediction, we show that local methods, such as MOI and HOC (see Section 3), yield better predictive accuracy when longer cycles are considered. In addition, if we consider the global low-rank structure of the network, prediction via matrix factorization further outperforms local methods in terms of both accuracy and running time. For clustering, we show that clustering via low rank model gives us significantly better results than clustering via signed Laplacian. These results suggest the usefulness of the global perspective on social balance. |
| Researcher Affiliation | Academia | Kai-Yang Chiang EMAIL Cho-Jui Hsieh EMAIL Nagarajan Natarajan EMAIL Inderjit S. Dhillon EMAIL Department of Computer Science University of Texas at Austin Austin, TX 78701, USA Ambuj Tewari EMAIL Department of Statistics, and Department of Electrical Engineering and Computer Science University of Michigan Ann Arbor, MI 48109, USA |
| Pseudocode | Yes | Algorithm 1: Sign Prediction via Singular Value Projection (SVP) ... Algorithm 2: Clustering with Matrix Completion |
| Open Source Code | No | The paper describes algorithms and methods but does not provide an explicit statement about releasing code or a direct link to a code repository for the work described. |
| Open Datasets | Yes | We also consider three real-life signed networks: Epinions, Slashdot, Wikipedia. Epinions is a consumer review network in which users can either trust or distrust other consumer s reviews. Slashdot is a discussion web site in which users can recognize others as friends or foes. Wikipedia is a who-votes-for-whom network in which users can vote for or against others to be administrators in Wikipedia. These three data sets have previously been used as benchmarks for sign prediction (Leskovec et al., 2010a; Chiang et al., 2011). |
| Dataset Splits | Yes | For real-life data sets, we resort to 10-fold cross-validation: we (randomly) created ten disjoint test folds each consisting of 10% of the total number of edges in the network. For each test fold, the remaining 90% of the edges serve as the training set. |
| Hardware Specification | No | The paper discusses running times and experimental comparisons but does not provide specific details about the hardware used for running the experiments (e.g., CPU/GPU models, memory specifications). |
| Software Dependencies | No | The paper describes various algorithms like Alternating Least Squares (ALS) and Stochastic Gradient Descent (SGD) but does not provide specific version numbers for any software libraries, frameworks, or dependencies used in their implementation. |
| Experiment Setup | No | The paper mentions parameters like rank k, regularization λ, and step size η within the algorithm descriptions. However, it does not provide concrete values for these hyperparameters or other system-level training configurations (e.g., learning rates, batch sizes, number of epochs) in the experimental setup details for either synthetic or real-life datasets. |