21 |
Evaluating Multilingual Text Encoders for Unsupervised Cross-Lingual Retrieval ...
|
|
|
|
Abstract:
Pretrained multilingual text encoders based on neural Transformer architectures, such as multilingual BERT (mBERT) and XLM, have achieved strong performance on a myriad of language understanding tasks. Consequently, they have been adopted as a go-to paradigm for multilingual and cross-lingual representation learning and transfer, rendering cross-lingual word embeddings (CLWEs) effectively obsolete. However, questions remain to which extent this finding generalizes 1) to unsupervised settings and 2) for ad-hoc cross-lingual IR (CLIR) tasks. Therefore, in this work we present a systematic empirical study focused on the suitability of the state-of-the-art multilingual encoders for cross-lingual document and sentence retrieval tasks across a large number of language pairs. In contrast to supervised language understanding, our results indicate that for unsupervised document-level CLIR -- a setup with no relevance judgments for IR-specific fine-tuning -- pretrained encoders fail to significantly outperform models ... : accepted at ECIR'21 (preprint) ...
|
|
Keyword:
Computation and Language cs.CL; FOS Computer and information sciences; H.3.3; I.2.7; Information Retrieval cs.IR
|
|
URL: https://dx.doi.org/10.48550/arxiv.2101.08370 https://arxiv.org/abs/2101.08370
|
|
BASE
|
|
Hide details
|
|
22 |
RedditBias: A Real-World Resource for Bias Evaluation and Debiasing of Conversational Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
23 |
Parameter space factorization for zero-shot learning across tasks and languages ...
|
|
|
|
BASE
|
|
Show details
|
|
25 |
MirrorWiC: On Eliciting Word-in-Context Representations from Pretrained Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
26 |
UNKs Everywhere: Adapting Multilingual Language Models to New Scripts ...
|
|
|
|
BASE
|
|
Show details
|
|
27 |
How Good is Your Tokenizer? On the Monolingual Performance of Multilingual Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
28 |
Learning Domain-Specialised Representations for Cross-Lingual Biomedical Entity Linking ...
|
|
|
|
BASE
|
|
Show details
|
|
29 |
MirrorWiC: On Eliciting Word-in-Context Representations from Pretrained Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
30 |
Multilingual and Cross-Lingual Intent Detection from Spoken Data ...
|
|
|
|
BASE
|
|
Show details
|
|
31 |
Semantic Data Set Construction from Human Clustering and Spatial Arrangement ...
|
|
|
|
BASE
|
|
Show details
|
|
32 |
AM2iCo: Evaluating Word Meaning in Context across Low-Resource Languages with Adversarial Examples ...
|
|
|
|
BASE
|
|
Show details
|
|
33 |
Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders ...
|
|
|
|
BASE
|
|
Show details
|
|
34 |
Parameter space factorization for zero-shot learning across tasks and languages
|
|
|
|
In: Transactions of the Association for Computational Linguistics, 9 (2021)
|
|
BASE
|
|
Show details
|
|
35 |
AM2iCo: Evaluating Word Meaning in Context across Low-Resource Languages with Adversarial Examples ...
|
|
|
|
BASE
|
|
Show details
|
|
36 |
Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders ...
|
|
|
|
BASE
|
|
Show details
|
|
37 |
LexFit: Lexical Fine-Tuning of Pretrained Language Models ...
|
|
|
|
BASE
|
|
Show details
|
|
38 |
Verb Knowledge Injection for Multilingual Event Processing ...
|
|
|
|
BASE
|
|
Show details
|
|
39 |
A Closer Look at Few-Shot Crosslingual Transfer: The Choice of Shots Matters ...
|
|
|
|
BASE
|
|
Show details
|
|
40 |
Is supervised syntactic parsing beneficial for language understanding tasks? An empirical investigation
|
|
|
|
BASE
|
|
Show details
|
|
|
|