A Unified Bayesian Model of Scripts, Frames and Language
Authors: Francis Ferraro, Benjamin Van Durme
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirically, our approach yields improved scenario representations, reflected quantitatively in lower surprisal and more coherent latent scenarios. Experiments, Evaluation Criteria, Evaluation Results, Figure 4: The held-out averaged log-likelihood of our model versus the baseline., Figure 5: Topic coherence at M = 20. |
| Researcher Affiliation | Academia | Francis Ferraro1 Benjamin Van Durme1,2 1Center for Language and Speech Processing 2Human Language Technology Center of Excellence Johns Hopkins University |
| Pseudocode | No | The paper describes the inference process mathematically and textually but does not contain a structured pseudocode or algorithm block. |
| Open Source Code | Yes | In our publicly available C++ implementation,1 we use GSL. 1https://github.com/fmof/unified-probabilistic-frames |
| Open Datasets | Yes | we use 10K training and 1K held-out NYT articles sampled uniformly at random from all years of Concretely Annotated Gigaword (Ferraro et al. 2014). |
| Dataset Splits | No | The paper mentions '10K training and 1K held-out NYT articles' and '1K held-out documents' which implies a train/test split, but no explicit validation set or three-way split is detailed. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper mentions using 'GSL' in its C++ implementation but does not provide specific version numbers for GSL or the C++ compiler/environment. |
| Experiment Setup | No | The paper describes the model and inference approach but does not provide specific experimental setup details such as concrete hyperparameter values (e.g., learning rate, batch size, number of epochs) or specific training configurations beyond optimizing hyperparameters generally. |