C3MM: Clique-Closure based Hyperlink Prediction
Authors: Govind Sharma, Prasanna Patil, M. Narasimha Murty
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform experiments (both the hypothesis-test as well as the hyperlink prediction) on multiple real datasets, report results, and provide both quantitative and qualitative arguments favouring better performances w.r.t. the state-of-the-art. |
| Researcher Affiliation | Academia | Govind Sharma , Prasanna Patil and M. Narasimha Murty Indian Institute of Science, Bengaluru, India |
| Pseudocode | Yes | Algorithm 1 An algorithm to test CCH on a temporal hypergraph H = (V, F, T). |
| Open Source Code | Yes | 3Code available at https://github.com/govindjsk/c3mm |
| Open Datasets | Yes | We have performed our experiments on altogether ten datasets, of which four are temporal hypergraphs and we use the six non-temporal metabolite hypergraphs from Zhang et al. [Zhang et al., 2018]. ... For more information, we suggest the reader to refer to Benson et al. [Benson et al., 2018] for an extensive analysis of the four (and more) temporal datasets, and to Zhang et al. [Zhang et al., 2018] for the six metabolites datasets. |
| Dataset Splits | Yes | For our experiments, we first partition the set of hyperlinks into two parts, namely observed and unobserved hyperlinks (Fobs and Funobs). For temporal hypergraphs, the set of hyperlinks is partitioned chronologically whereas for non-temporal hypergraphs, it is done randomly. ... We sample 15 times as many non-hyperlinks as there are hyperlinks in the unobserved hypergraph for all of the temporal hypergraphs. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU/CPU models or memory used for the experiments. |
| Software Dependencies | No | The paper mentions software components like 'Symmetric NMF' and 'Factorization Machines' but does not provide specific version numbers for any of the software dependencies used in their experiments. |
| Experiment Setup | Yes | We fix the size of latent dimension for symmetric NMF in (6) to be k = 30 for the all the datasets, just as Zhang et al. [Zhang et al., 2018] do as a default choice for CMM. |