Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387561301> ?p ?o ?g. }
Showing items 1 to 53 of
53
with 100 items per page.
- W4387561301 abstract "Prompt-based learning's efficacy across numerous natural language processing tasks has led to its integration into dense passage retrieval. Prior research has mainly focused on enhancing the semantic understanding of pre-trained language models by optimizing a single vector as a continuous prompt. This approach, however, leads to a semantic space collapse; identical semantic information seeps into all representations, causing their distributions to converge in a restricted region. This hinders differentiation between relevant and irrelevant passages during dense retrieval. To tackle this issue, we present Topic-DPR, a dense passage retrieval model that uses topic-based prompts. Unlike the single prompt method, multiple topic-based prompts are established over a probabilistic simplex and optimized simultaneously through contrastive learning. This encourages representations to align with their topic distributions, improving space uniformity. Furthermore, we introduce a novel positive and negative sampling strategy, leveraging semi-structured data to boost dense retrieval efficiency. Experimental results from two datasets affirm that our method surpasses previous state-of-the-art retrieval techniques." @default.
- W4387561301 created "2023-10-12" @default.
- W4387561301 creator A5008424883 @default.
- W4387561301 creator A5035579335 @default.
- W4387561301 creator A5079920323 @default.
- W4387561301 date "2023-10-10" @default.
- W4387561301 modified "2023-10-13" @default.
- W4387561301 title "Topic-DPR: Topic-based Prompts for Dense Passage Retrieval" @default.
- W4387561301 doi "https://doi.org/10.48550/arxiv.2310.06626" @default.
- W4387561301 hasPublicationYear "2023" @default.
- W4387561301 type Work @default.
- W4387561301 citedByCount "0" @default.
- W4387561301 crossrefType "posted-content" @default.
- W4387561301 hasAuthorship W4387561301A5008424883 @default.
- W4387561301 hasAuthorship W4387561301A5035579335 @default.
- W4387561301 hasAuthorship W4387561301A5079920323 @default.
- W4387561301 hasBestOaLocation W43875613011 @default.
- W4387561301 hasConcept C106131492 @default.
- W4387561301 hasConcept C111919701 @default.
- W4387561301 hasConcept C140779682 @default.
- W4387561301 hasConcept C154945302 @default.
- W4387561301 hasConcept C204321447 @default.
- W4387561301 hasConcept C23123220 @default.
- W4387561301 hasConcept C2778572836 @default.
- W4387561301 hasConcept C31972630 @default.
- W4387561301 hasConcept C41008148 @default.
- W4387561301 hasConcept C49937458 @default.
- W4387561301 hasConceptScore W4387561301C106131492 @default.
- W4387561301 hasConceptScore W4387561301C111919701 @default.
- W4387561301 hasConceptScore W4387561301C140779682 @default.
- W4387561301 hasConceptScore W4387561301C154945302 @default.
- W4387561301 hasConceptScore W4387561301C204321447 @default.
- W4387561301 hasConceptScore W4387561301C23123220 @default.
- W4387561301 hasConceptScore W4387561301C2778572836 @default.
- W4387561301 hasConceptScore W4387561301C31972630 @default.
- W4387561301 hasConceptScore W4387561301C41008148 @default.
- W4387561301 hasConceptScore W4387561301C49937458 @default.
- W4387561301 hasLocation W43875613011 @default.
- W4387561301 hasOpenAccess W4387561301 @default.
- W4387561301 hasPrimaryLocation W43875613011 @default.
- W4387561301 hasRelatedWork W2030292806 @default.
- W4387561301 hasRelatedWork W2147155098 @default.
- W4387561301 hasRelatedWork W2215759665 @default.
- W4387561301 hasRelatedWork W2494523064 @default.
- W4387561301 hasRelatedWork W2588219639 @default.
- W4387561301 hasRelatedWork W2938171715 @default.
- W4387561301 hasRelatedWork W2943623134 @default.
- W4387561301 hasRelatedWork W2960358116 @default.
- W4387561301 hasRelatedWork W3041172967 @default.
- W4387561301 hasRelatedWork W4287727129 @default.
- W4387561301 isParatext "false" @default.
- W4387561301 isRetracted "false" @default.
- W4387561301 workType "article" @default.