English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recently, Dense Retrieval (DR) has become a promising solution to documentretrieval, where document representations are used to perform effective andefficient semantic search. However, DR remains challenging on long documents,due to the quadratic complexity of its Transformer-based encoder and the finitecapacity of a low-dimension embedding. Current DR models use suboptimalstrategies such as truncating or splitting-and-pooling to long documentsleading to poor utilization of whole document information. In this work, totackle this problem, we propose Segment representation learning for longdocuments Dense Retrieval (SeDR). In SeDR, Segment-Interaction Transformer isproposed to encode long documents into document-aware and segment-sensitiverepresentations, while it holds the complexity of splitting-and-pooling andoutperforms other segment-interaction patterns on DR. Since GPU memoryrequirements for long document encoding causes insufficient negatives for DRtraining, Late-Cache Negative is further proposed to provide additional cachenegatives for optimizing representation learning. Experiments on MS MARCO andTREC-DL datasets show that SeDR achieves superior performance among DR models,and confirm the effectiveness of SeDR on long document retrieval.

SeDR: Segment Representation Learning for Long Documents Dense Retrieval