Solving Analogies on Words based on Minimal Complexity Transformation
Authors: Pierre-Alexandre Murena, Marie Al-Ghossein, Jean-Louis Dessalles, Antoine Cornuéjols
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We illustrate our method on a large-scale benchmark dataset and compare with state-of-the-art approaches to demonstrate the interest of using complexity to solve analogies on words. |
| Researcher Affiliation | Academia | 1Helsinki Institute for Information Technology HIIT, Department of Computer Science, Aalto University, Espoo, Finland 2LTCI, T el ecom Paris, Institut Polytechnique de Paris, Palaiseau, France 3UMR MIA-518, Agro Paris Tech INRA, Paris, France |
| Pseudocode | No | The paper describes its algorithm textually but does not provide structured pseudocode or an algorithm block. |
| Open Source Code | Yes | The source code for a Python interpreter is available on the authors webpage. |
| Open Datasets | Yes | We base our work on one of the largest datasets available for solving analogies on words that we denote by SIGMORPHON 16 and that is presented in [Lepage, 2017]. This dataset is from the Track 1 Task 1 of SIGMORPHON 2016 Shared Task 2. |
| Dataset Splits | No | The paper uses the SIGMORPHON 16 dataset but does not explicitly detail the train/validation/test splits (e.g., percentages or specific counts) used for their experiments. |
| Hardware Specification | Yes | For instance, and with a reasonable implementation in C++1 run on a machine with one processor Intel Core i7 2.9 GHz and 8G of RAM, it requires a couple of seconds to find a solution for the analogy rosa:rosam::vita:x and about 20 minutes for the analogy orang:orang-orang::burung:x . |
| Software Dependencies | No | The paper mentions implementation in Python and C++ but does not provide specific version numbers for these languages or any other software dependencies (e.g., libraries, frameworks). |
| Experiment Setup | Yes | In our experiments, we fixed it to |c| + |b| |a|, taking our inspiration from the works on proportional analogy [Lepage, 2017]. |