Intrinsic and Extrinsic Evaluations of Word Embeddings

Authors: Michael Zhai, Johnny Tan, Jinho Choi

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments show that the word embedding clusters give high correlations to the synonym and hyponym sets in Word Net, and give 0.88% and 0.17% absolute improvements in accuracy to named entity recognition and part-of-speech tagging, respectively.
Researcher Affiliation Academia Michael Zhai, Johnny Tan, Jinho D. Choi Department of Mathematics and Computer Science Emory University Atlanta, GA 30322 {michael.zhai,johnny.tan,jinho.choi}@emory.edu
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes All resources are available at http://github.com/emorynlp.
Open Datasets Yes From Word Net, sets of synonyms and hyponyms of the 100 most frequent nouns and verbs in the New York Times corpus1 are extracted and compared to the clusters generated from the word embeddings. and 1https://catalog.ldc.upenn.edu/LDC2008T19. Also: The English portion of Onto Notes 5 is used for experiments following the standard split suggested by Pradhan et al. (2013).
Dataset Splits Yes The English portion of Onto Notes 5 is used for experiments following the standard split suggested by Pradhan et al. (2013).
Hardware Specification No The paper does not specify any hardware details (e.g., CPU/GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions 'Ada Grad is used for training statistical models' but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes Ada Grad is used for training statistical models. All of the above experiments are using the maximum cluster size of 1,500. We also tested on the max cluster size of 15,000, which showed very similar results. additional experiments are conducted by concatenating the word and contextual vectors (w+c).