FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4309698332> ?p ?o ?g. }

• W4309698332 endingPage "39" @default.
• W4309698332 startingPage "1" @default.
• W4309698332 abstract "Pseudo-relevance feedback mechanisms, from Rocchio to the relevance models, have shown the usefulness of expanding and reweighting the users’ initial queries using information occurring in an initial set of retrieved documents, known as the pseudo-relevant set. Recently, dense retrieval – through the use of neural contextual language models such as BERT for analysing the documents’ and queries’ contents and computing their relevance scores – has shown a promising performance on several information retrieval tasks still relying on the traditional inverted index for identifying documents relevant to a query. Two different dense retrieval families have emerged: the use of single embedded representations for each passage and query, e.g., using BERT’s [CLS] token, or via multiple representations, e.g., using an embedding for each token of the query and document (exemplified by ColBERT). In this work, we conduct the first study into the potential for multiple representation dense retrieval to be enhanced using pseudo-relevance feedback and present our proposed approach ColBERT-PRF. In particular, based on the pseudo-relevant set of documents identified using a first-pass dense retrieval, ColBERT-PRF extracts the representative feedback embeddings from the document embeddings of the pseudo-relevant set. Among the representative feedback embeddings, the embeddings that most highly discriminate among documents are employed as the expansion embeddings, which are then added to the original query representation. We show that these additional expansion embeddings both enhance the effectiveness of a reranking of the initial query results as well as an additional dense retrieval operation. Indeed, experiments on the MSMARCO passage ranking dataset show that MAP can be improved by up to 26% on the TREC 2019 query set and 10% on the TREC 2020 query set by the application of our proposed ColBERT-PRF method on a ColBERT dense retrieval approach.We further validate the effectiveness of our proposed pseudo-relevance feedback technique for a dense retrieval model on MSMARCO document ranking and TREC Robust04 document ranking tasks. For instance, ColBERT-PRF exhibits up to 21% and 14% improvement in MAP over the ColBERT E2E model on the MSMARCO document ranking TREC 2019 and TREC 2020 query sets, respectively. Additionally, we study the effectiveness of variants of the ColBERT-PRF model with different weighting methods. Finally, we show that ColBERT-PRF can be made more efficient, attaining up to 4.54× speedup over the default ColBERT-PRF model, and with little impact on effectiveness, through the application of approximate scoring and different clustering methods." @default.
• W4309698332 created "2022-11-29" @default.
• W4309698332 creator A5018894843 @default.
• W4309698332 creator A5057643560 @default.
• W4309698332 creator A5058010200 @default.
• W4309698332 creator A5079046603 @default.
• W4309698332 date "2023-01-16" @default.
• W4309698332 modified "2023-10-16" @default.
• W4309698332 title "ColBERT-PRF: Semantic Pseudo-Relevance Feedback for Dense Passage and Document Retrieval" @default.
• W4309698332 cites W2105157020 @default.
• W4309698332 cites W2515351093 @default.
• W4309698332 cites W3001344098 @default.
• W4309698332 cites W3037084154 @default.
• W4309698332 cites W3045745713 @default.
• W4309698332 cites W3157758108 @default.
• W4309698332 cites W3197464301 @default.
• W4309698332 cites W4289516287 @default.
• W4309698332 doi "https://doi.org/10.1145/3572405" @default.
• W4309698332 hasPublicationYear "2023" @default.
• W4309698332 type Work @default.
• W4309698332 citedByCount "5" @default.
• W4309698332 countsByYear W43096983322023 @default.
• W4309698332 crossrefType "journal-article" @default.
• W4309698332 hasAuthorship W4309698332A5018894843 @default.
• W4309698332 hasAuthorship W4309698332A5057643560 @default.
• W4309698332 hasAuthorship W4309698332A5058010200 @default.
• W4309698332 hasAuthorship W4309698332A5079046603 @default.
• W4309698332 hasBestOaLocation W43096983322 @default.
• W4309698332 hasConcept C115961682 @default.
• W4309698332 hasConcept C154945302 @default.
• W4309698332 hasConcept C158154518 @default.
• W4309698332 hasConcept C161156560 @default.
• W4309698332 hasConcept C1667742 @default.
• W4309698332 hasConcept C177264268 @default.
• W4309698332 hasConcept C17744445 @default.
• W4309698332 hasConcept C199360897 @default.
• W4309698332 hasConcept C199539241 @default.
• W4309698332 hasConcept C23123220 @default.
• W4309698332 hasConcept C2776359362 @default.
• W4309698332 hasConcept C2779532271 @default.
• W4309698332 hasConcept C38652104 @default.
• W4309698332 hasConcept C41008148 @default.
• W4309698332 hasConcept C41608201 @default.
• W4309698332 hasConcept C48145219 @default.
• W4309698332 hasConcept C94625758 @default.
• W4309698332 hasConcept C99016210 @default.
• W4309698332 hasConceptScore W4309698332C115961682 @default.
• W4309698332 hasConceptScore W4309698332C154945302 @default.
• W4309698332 hasConceptScore W4309698332C158154518 @default.
• W4309698332 hasConceptScore W4309698332C161156560 @default.
• W4309698332 hasConceptScore W4309698332C1667742 @default.
• W4309698332 hasConceptScore W4309698332C177264268 @default.
• W4309698332 hasConceptScore W4309698332C17744445 @default.
• W4309698332 hasConceptScore W4309698332C199360897 @default.
• W4309698332 hasConceptScore W4309698332C199539241 @default.
• W4309698332 hasConceptScore W4309698332C23123220 @default.
• W4309698332 hasConceptScore W4309698332C2776359362 @default.
• W4309698332 hasConceptScore W4309698332C2779532271 @default.
• W4309698332 hasConceptScore W4309698332C38652104 @default.
• W4309698332 hasConceptScore W4309698332C41008148 @default.
• W4309698332 hasConceptScore W4309698332C41608201 @default.
• W4309698332 hasConceptScore W4309698332C48145219 @default.
• W4309698332 hasConceptScore W4309698332C94625758 @default.
• W4309698332 hasConceptScore W4309698332C99016210 @default.
• W4309698332 hasFunder F4320322725 @default.
• W4309698332 hasFunder F4320334627 @default.
• W4309698332 hasIssue "1" @default.
• W4309698332 hasLocation W43096983321 @default.
• W4309698332 hasLocation W43096983322 @default.
• W4309698332 hasOpenAccess W4309698332 @default.
• W4309698332 hasPrimaryLocation W43096983321 @default.
• W4309698332 hasRelatedWork W1573151109 @default.
• W4309698332 hasRelatedWork W1887676092 @default.
• W4309698332 hasRelatedWork W2008117759 @default.
• W4309698332 hasRelatedWork W2036223361 @default.
• W4309698332 hasRelatedWork W2039356789 @default.
• W4309698332 hasRelatedWork W2100046112 @default.
• W4309698332 hasRelatedWork W2740747097 @default.
• W4309698332 hasRelatedWork W3197464301 @default.
• W4309698332 hasRelatedWork W4252055628 @default.
• W4309698332 hasRelatedWork W2183374463 @default.
• W4309698332 hasVolume "17" @default.
• W4309698332 isParatext "false" @default.
• W4309698332 isRetracted "false" @default.
• W4309698332 workType "article" @default.