Matches in SemOpenAlex for { <https://semopenalex.org/work/W3170720008> ?p ?o ?g. }
- W3170720008 abstract "Efficient inner product search on embedding vectors is often the vital stage for online ranking services, such as recommendation and information retrieval. Recommendation algorithms, e.g., matrix factorization, typically produce latent vectors to represent users or items. The recommendation services are conducted by retrieving the most relevant item vectors given the user vector, where the relevance is often defined by inner product. Therefore, developing efficient recommender systems often requires solving the so-called maximum inner product search (MIPS) problem. In the past decade, there have been many studies on efficient MIPS algorithms. This task is challenging in part because the inner product does not follow the triangle inequality of metric space. Compared with hash-based or quantization-based MIPS solutions, in recent years graph-based MIPS algorithms have demonstrated their strong empirical advantages in many real-world MIPS tasks. In this paper, we propose a new index graph construction method named norm adjusted proximity graph (NAPG), for efficient MIPS. With adjusting factors estimated on sampled data, NAPG is able to select more meaningful data points to connect with when constructing graph-based index for inner product search. Our extensive experiments on a variety of datasets verify that the improved graph-based index strategy provides another strong addition to the pool of efficient MIPS algorithms." @default.
- W3170720008 created "2021-06-22" @default.
- W3170720008 creator A5011202621 @default.
- W3170720008 creator A5041660006 @default.
- W3170720008 creator A5066690809 @default.
- W3170720008 creator A5068755696 @default.
- W3170720008 creator A5073070877 @default.
- W3170720008 creator A5075541712 @default.
- W3170720008 date "2021-08-14" @default.
- W3170720008 modified "2023-10-16" @default.
- W3170720008 title "Norm Adjusted Proximity Graph for Fast Inner Product Retrieval" @default.
- W3170720008 cites W1455310343 @default.
- W3170720008 cites W1537946535 @default.
- W3170720008 cites W1974627246 @default.
- W3170720008 cites W1977046819 @default.
- W3170720008 cites W1983645263 @default.
- W3170720008 cites W1994389483 @default.
- W3170720008 cites W2010416066 @default.
- W3170720008 cites W2024668293 @default.
- W3170720008 cites W2042281163 @default.
- W3170720008 cites W2060094367 @default.
- W3170720008 cites W2080320419 @default.
- W3170720008 cites W2086179657 @default.
- W3170720008 cites W2101409192 @default.
- W3170720008 cites W2102765684 @default.
- W3170720008 cites W2129841113 @default.
- W3170720008 cites W2147717514 @default.
- W3170720008 cites W2250445771 @default.
- W3170720008 cites W2362855512 @default.
- W3170720008 cites W2512971201 @default.
- W3170720008 cites W2612493630 @default.
- W3170720008 cites W2740920897 @default.
- W3170720008 cites W2744136723 @default.
- W3170720008 cites W2783625345 @default.
- W3170720008 cites W2801922433 @default.
- W3170720008 cites W2808891190 @default.
- W3170720008 cites W2886950694 @default.
- W3170720008 cites W2897002648 @default.
- W3170720008 cites W2950729111 @default.
- W3170720008 cites W2963284996 @default.
- W3170720008 cites W2963469388 @default.
- W3170720008 cites W2984115492 @default.
- W3170720008 cites W2997411837 @default.
- W3170720008 cites W2998704965 @default.
- W3170720008 cites W3028864969 @default.
- W3170720008 cites W3105698638 @default.
- W3170720008 cites W4249267926 @default.
- W3170720008 cites W4285750439 @default.
- W3170720008 cites W4290960530 @default.
- W3170720008 doi "https://doi.org/10.1145/3447548.3467412" @default.
- W3170720008 hasPublicationYear "2021" @default.
- W3170720008 type Work @default.
- W3170720008 sameAs 3170720008 @default.
- W3170720008 citedByCount "7" @default.
- W3170720008 countsByYear W31707200082021 @default.
- W3170720008 countsByYear W31707200082022 @default.
- W3170720008 countsByYear W31707200082023 @default.
- W3170720008 crossrefType "proceedings-article" @default.
- W3170720008 hasAuthorship W3170720008A5011202621 @default.
- W3170720008 hasAuthorship W3170720008A5041660006 @default.
- W3170720008 hasAuthorship W3170720008A5066690809 @default.
- W3170720008 hasAuthorship W3170720008A5068755696 @default.
- W3170720008 hasAuthorship W3170720008A5073070877 @default.
- W3170720008 hasAuthorship W3170720008A5075541712 @default.
- W3170720008 hasConcept C119857082 @default.
- W3170720008 hasConcept C124101348 @default.
- W3170720008 hasConcept C132525143 @default.
- W3170720008 hasConcept C23123220 @default.
- W3170720008 hasConcept C38652104 @default.
- W3170720008 hasConcept C41008148 @default.
- W3170720008 hasConcept C557471498 @default.
- W3170720008 hasConcept C75564084 @default.
- W3170720008 hasConcept C80444323 @default.
- W3170720008 hasConcept C99138194 @default.
- W3170720008 hasConceptScore W3170720008C119857082 @default.
- W3170720008 hasConceptScore W3170720008C124101348 @default.
- W3170720008 hasConceptScore W3170720008C132525143 @default.
- W3170720008 hasConceptScore W3170720008C23123220 @default.
- W3170720008 hasConceptScore W3170720008C38652104 @default.
- W3170720008 hasConceptScore W3170720008C41008148 @default.
- W3170720008 hasConceptScore W3170720008C557471498 @default.
- W3170720008 hasConceptScore W3170720008C75564084 @default.
- W3170720008 hasConceptScore W3170720008C80444323 @default.
- W3170720008 hasConceptScore W3170720008C99138194 @default.
- W3170720008 hasLocation W31707200081 @default.
- W3170720008 hasOpenAccess W3170720008 @default.
- W3170720008 hasPrimaryLocation W31707200081 @default.
- W3170720008 hasRelatedWork W1863924475 @default.
- W3170720008 hasRelatedWork W1965139815 @default.
- W3170720008 hasRelatedWork W1973113276 @default.
- W3170720008 hasRelatedWork W2000536584 @default.
- W3170720008 hasRelatedWork W2018047529 @default.
- W3170720008 hasRelatedWork W2026292488 @default.
- W3170720008 hasRelatedWork W2122974868 @default.
- W3170720008 hasRelatedWork W2129006149 @default.
- W3170720008 hasRelatedWork W2589051495 @default.
- W3170720008 hasRelatedWork W3021548270 @default.
- W3170720008 isParatext "false" @default.
- W3170720008 isRetracted "false" @default.
- W3170720008 magId "3170720008" @default.