Stanford AI Lab Showcases Research at ACL 2022
The Stanford Artificial Intelligence Laboratory (SAIL) has announced its contributions to the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022), held from May 22nd to May 27th. The lab is presenting a diverse portfolio of accepted papers covering significant advancements in natural language processing and machine learning. Key research topics include LinkBERT, a method for pretraining language models using document links; an analysis of BERT's sensitivity to word order in grammatical role classification; and investigations into cosine similarity measures for high-frequency words. Additionally, SAIL researchers are addressing challenges in abstractive summarization, specifically the trade-off between faithfulness and abstractiveness, as well as spurious correlations in reference-free evaluation metrics for text generation. Other notable works include TABi, a type-aware bi-encoder for open-domain entity retrieval, and a few-shot semantic parser for Wizard-of-Oz dialogues. This collection highlights Stanford's ongoing leadership in computational linguistics, offering resources such as paper links, videos, and code repositories for the broader academic community.
Wire timeline
Stanford AI Lab Showcases Research at ACL 2022
The Stanford Artificial Intelligence Laboratory (SAIL) has announced its contributions to the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022), held from May 22nd to May 27th. The lab is presenting a diverse portfolio of accepted papers covering significant advancements in natural language processing and machine learning. Key research topics include LinkBERT, a method for pretraining language models using document links; an analysis of BERT's sensitivity to word order in grammatical role classification; and investigations into cosine similarity measures for high-frequency words. Additionally, SAIL researchers are addressing challenges in abstractive summarization, specifically the trade-off between faithfulness and abstractiveness, as well as spurious correlations in reference-free evaluation metrics for text generation. Other notable works include TABi, a type-aware bi-encoder for open-domain entity retrieval, and a few-shot semantic parser for Wizard-of-Oz dialogues. This collection highlights Stanford's ongoing leadership in computational linguistics, offering resources such as paper links, videos, and code repositories for the broader academic community.
The Stanford AI Lab Blog