Length is a Curse and a Blessing for Document-level Semantics
Submission Type: Regular Long Paper
Submission Track: Semantics: Lexical, Sentence level, Document Level, Textual Inference, etc.
Submission Track 2: Information Retrieval and Text Mining
Keywords: contrastive learning, sentence representation learning, document representation learning, semantics, document length
TL;DR: We present the overlooked length vulnerability of contrastive text encoders, and use length as a standalone solution to provide semantic signal for sentence and document-level representation learning.
Abstract: In recent years, contrastive learning (CL) has been extensively utilized to recover sentence and document-level encoding capability from pre-trained language models.
In this work, we question the length generalizability of CL-based models, i.e., their vulnerability towards length-induced semantic shift.
We verify not only that length vulnerability is a significant yet overlooked research gap, but we can devise unsupervised CL methods solely depending on the semantic signal provided by document length.
We first derive the theoretical foundations underlying length attacks, showing that elongating a document would intensify the high intra-document similarity that is already brought by CL. Moreover, we found that isotropy promised by CL is highly dependent on the length range of text exposed in training.
Inspired by these findings, we introduce a simple yet universal document representation learning framework,
**LA(SER)$^3$**: length-agnostic self-reference for semantically robust sentence representation learning, achieving state-of-the-art unsupervised performance on the standard information retrieval benchmark. [Our code is publicly available.](https://github.com/gowitheflow-1998/LA-SER-cubed)
Submission Number: 4865
Loading