Auto-MLM: Improved Contrastive Learning for Self-supervised Multi-lingual Knowledge Retrieval

TL;DR

Auto-MLM introduces a joint training approach combining contrastive learning and masked language modeling to enhance self-supervised multi-lingual knowledge retrieval, outperforming previous state-of-the-art methods across multiple languages and datasets.

Contribution

It proposes a novel joint training method that integrates CL and Auto-MLM to better extract internal information for multi-lingual knowledge retrieval.

Findings

Outperforms all previous SOTA methods on AliExpress and LAZADA datasets.

Effective across 8 languages with consistent improvements.

Enhances internal information extraction from queries.

Abstract

Contrastive learning (CL) has become a ubiquitous approach for several natural language processing (NLP) downstream tasks, especially for question answering (QA). However, the major challenge, how to efficiently train the knowledge retrieval model in an unsupervised manner, is still unresolved. Recently the commonly used methods are composed of CL and masked language model (MLM). Unexpectedly, MLM ignores the sentence-level training, and CL also neglects extraction of the internal info from the query. To optimize the CL hardly obtain internal information from the original query, we introduce a joint training method by combining CL and Auto-MLM for self-supervised multi-lingual knowledge retrieval. First, we acquire the fixed dimensional sentence vector. Then, mask some words among the original sentences with random strategy. Finally, we generate a new token representation for predicting the masked tokens. Experimental results show that our proposed approach consistently outperforms all the previous SOTA methods on both AliExpress $\&$ LAZADA service corpus and openly available corpora in 8 languages.