1 |
Understanding the effects of negative (and positive) pointwise mutual information on word vectors
|
|
|
|
Abstract:
Despite the recent popularity of contextual word embeddings, static word embeddings still dominate lexical semantic tasks, making their study of continued relevance. A widely adopted family of such static word embeddings is derived by explicitly factorizing the Pointwise Mutual Information (PMI) weighting of the cooccurrence matrix. As unobserved cooccurrences lead PMI to negative infinity, a common workaround is to clip negative PMI at 0. However, it is unclear what information is lost by collapsing negative PMI values to 0. To answer this question, we isolate and study the effects of negative (and positive) PMI on the semantics and geometry of models adopting factorization of different PMI matrices. Word and sentence-level evaluations show that only accounting for positive PMI in the factorization strongly captures both semantics and syntax, whereas using only negative PMI captures little of semantics but a surprising amount of syntactic information. Results also reveal that incorporating negative PMI induces stronger rank invariance of vector norms and direction, as well as improved rare word representations.
|
|
URL: https://eprints.whiterose.ac.uk/184560/
|
|
BASE
|
|
Hide details
|
|
3 |
Assessing idiomaticity representations in vector models with a noun compound dataset labeled at type and token levels
|
|
|
|
BASE
|
|
Show details
|
|
5 |
AStitchInLanguageModels : dataset and methods for the exploration of idiomaticity in pre-trained language models
|
|
|
|
BASE
|
|
Show details
|
|
6 |
CogNLP-Sheffield at CMCL 2021 Shared Task: Blending cognitively inspired features with transformer-based language models for predicting eye tracking patterns
|
|
|
|
BASE
|
|
Show details
|
|
7 |
Investigating language impact in bilingual approaches for computational language documentation
|
|
|
|
BASE
|
|
Show details
|
|
8 |
Unsupervised compositionality prediction of nominal compounds
|
|
|
|
BASE
|
|
Show details
|
|
9 |
A dual-attention hierarchical recurrent neural network for dialogue act classification
|
|
|
|
BASE
|
|
Show details
|
|
10 |
When the whole is greater than the sum of its parts : multiword expressions and idiomaticity
|
|
|
|
BASE
|
|
Show details
|
|
11 |
Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)
|
|
|
|
BASE
|
|
Show details
|
|
13 |
Empirical evaluation of sequence-to-sequence models for word discovery in low-resource settings
|
|
|
|
BASE
|
|
Show details
|
|
15 |
Similarity Measures for the Detection of Clinical Conditions with Verbal Fluency Tasks
|
|
|
|
BASE
|
|
Show details
|
|
16 |
A corpus study of verbal multiword expressions in Brazilian Portuguese
|
|
|
|
BASE
|
|
Show details
|
|
17 |
Unwritten languages demand attention too! Word discovery with encoder-decoder models
|
|
|
|
BASE
|
|
Show details
|
|
18 |
Restricted recurrent neural tensor networks: Exploiting word frequency and compositionality
|
|
|
|
BASE
|
|
Show details
|
|
19 |
UFRGS&LIF at SemEval-2016 task 10: Rule-based MWE identification and predominant-supersense tagging
|
|
|
|
BASE
|
|
Show details
|
|
20 |
How naked is the naked truth? A multilingual lexicon of nominal compound compositionality
|
|
|
|
BASE
|
|
Show details
|
|
|
|