ROOT13: Spotting Hypernyms, Co-Hyponyms and Randoms

Authors: Enrico Santus, Alessandro Lenci, Tin-Shing Chiu, Qin Lu, Chu-Ren Huang

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate it with a 10-fold cross validation on 9,600 pairs, equally distributed among the three classes and involving several Parts-Of Speech (i.e. adjectives, nouns and verbs). When all the classes are present, ROOT13 achieves an F1 score of 88.3%, against a baseline of 57.6% (vector cosine).
Researcher Affiliation Academia * The Hong Kong Polytechnic University, Hong Kong esantus@gmail.com, cstschiu@comp.polyu.edu.hk, {qin.lu, churen.huang}@polyu.edu.hk University of Pisa, Italy alessandro.lenci@ling.unipi.it
Pseudocode No No pseudocode or clearly labeled algorithm blocks were found in the paper.
Open Source Code Yes More info: https://github.com/esantus/.
Open Datasets Yes We have used 9,600 pairs, randomly extracted from three datasets: Lenci/Benotto (Santus et al., 2014b), BLESS (Baroni and Lenci, 2011) and EVALution (Santus et al., 2015).
Dataset Splits Yes We evaluate it with a 10-fold cross validation on 9,600 pairs, equally distributed among the three classes and involving several Parts-Of Speech (i.e. adjectives, nouns and verbs).
Hardware Specification No No specific hardware details (like CPU/GPU models, memory, or specific computer specifications) used for running the experiments were provided in the paper.
Software Dependencies No The paper mentions 'Weka' for the Random Forest algorithm but does not specify its version number or any other software dependencies with version details.
Experiment Setup Yes ROOT13 uses the Random Forest algorithm implemented in Weka (Breiman, 2001), with the default settings. ... All the features are normalized in the range 0-1: ... a window-based Vector Space Model (VSM), built on a combination of uk Wa C and Wa Ckypedia corpora (around 2.7 billion words) and recording word co-occurrences within the 5 nearest content words to the left and right of each target.