Matches in SemOpenAlex for { <https://semopenalex.org/work/W2233758814> ?p ?o ?g. }
- W2233758814 abstract "In this master thesis we describe a new automatic aspect extraction algorithm by incorporating relevance information to the dynamics of the Probabilistic Latent Semantic Analysis. An utility-biased likelihood statistical framework is described to formalize the incorporation of prior relevance information to the dynamics of the algorithm intrinsically. Moreover, a general abstract algorithm is presented to incorporate any arbitrary new feature variables to the analysis. A tempering procedure is inferred for this general algorithm as an entropic regularization of the utility-biased likelihood functional and a geometric interpretation of the algorithm is described, showing the intrinsic changes in the information space of the problem produced when different sources of prior utility estimations are provided over the same data. The general algorithm is applied to several information retrieval, recommendation and personalization tasks. Moreover, a set of post-processing aspect filters is presented. Some characteristics of the aspect distributions such as sparsity or low entropy are identified to enhance the overall diversity attained by the diversification algorithm. Proposed filters assure that the final aspect space has those properties, thus leading to better diversity levels. An experimental setup over TREC web track 09-12 data shows that the algorithm surpasses classic pLSA as an aspect extraction tool for the search diversification. Additional theoretical applications of the general procedure to information retrieval, recommendation and personalization tasks are given, leading to new relevanceaware models incorporating several variables to the latent semantic analysis. Finally the problem of optimizing the aspect space size for diversification is addressed. Analytical formulas for the dependency of diversity metrics on the choice of an automatically extracted aspect space are given under a simplified generative model for the relation between system aspects and evaluation true aspects. An experimental analysis of this dependence is performed over TREC web track data using pLSA as aspect extraction algorithm. ACM Computing Classification System. H.3.3 [Search and Retrieval]: retrieval models, relevance feedback I.2.7 [Natural Language Processing]: Language models." @default.
- W2233758814 created "2016-06-24" @default.
- W2233758814 creator A5043278211 @default.
- W2233758814 date "2015-09-28" @default.
- W2233758814 modified "2023-09-25" @default.
- W2233758814 title "Automatic aspect extraction in information retrieval diversity" @default.
- W2233758814 cites W1483313504 @default.
- W2233758814 cites W1540041485 @default.
- W2233758814 cites W1573082642 @default.
- W2233758814 cites W1612003148 @default.
- W2233758814 cites W1685426458 @default.
- W2233758814 cites W1732828232 @default.
- W2233758814 cites W1880262756 @default.
- W2233758814 cites W1968413755 @default.
- W2233758814 cites W1969077952 @default.
- W2233758814 cites W1973429478 @default.
- W2233758814 cites W1983578042 @default.
- W2233758814 cites W1986333823 @default.
- W2233758814 cites W1987133051 @default.
- W2233758814 cites W1993320088 @default.
- W2233758814 cites W1994808901 @default.
- W2233758814 cites W2005245664 @default.
- W2233758814 cites W2007463795 @default.
- W2233758814 cites W2018605660 @default.
- W2233758814 cites W2023188792 @default.
- W2233758814 cites W2026430213 @default.
- W2233758814 cites W2026893190 @default.
- W2233758814 cites W2032558547 @default.
- W2233758814 cites W2033403400 @default.
- W2233758814 cites W2038062826 @default.
- W2233758814 cites W2049455633 @default.
- W2233758814 cites W2049633694 @default.
- W2233758814 cites W2053943100 @default.
- W2233758814 cites W2056029990 @default.
- W2233758814 cites W2060796982 @default.
- W2233758814 cites W2063397738 @default.
- W2233758814 cites W2077096052 @default.
- W2233758814 cites W2077139171 @default.
- W2233758814 cites W2080674479 @default.
- W2233758814 cites W2090963365 @default.
- W2233758814 cites W2091891040 @default.
- W2233758814 cites W2093495945 @default.
- W2233758814 cites W2100419635 @default.
- W2233758814 cites W2104373244 @default.
- W2233758814 cites W2104983284 @default.
- W2233758814 cites W2107126505 @default.
- W2233758814 cites W2107743791 @default.
- W2233758814 cites W2113640060 @default.
- W2233758814 cites W2124379907 @default.
- W2233758814 cites W2127300005 @default.
- W2233758814 cites W2132314908 @default.
- W2233758814 cites W2134731454 @default.
- W2233758814 cites W2137227096 @default.
- W2233758814 cites W2140551944 @default.
- W2233758814 cites W2141429664 @default.
- W2233758814 cites W2146832283 @default.
- W2233758814 cites W2147152072 @default.
- W2233758814 cites W2148212498 @default.
- W2233758814 cites W2149414429 @default.
- W2233758814 cites W2150139843 @default.
- W2233758814 cites W2155194165 @default.
- W2233758814 cites W2156612134 @default.
- W2233758814 cites W2160709761 @default.
- W2233758814 cites W2163200373 @default.
- W2233758814 cites W2163382007 @default.
- W2233758814 cites W2168084958 @default.
- W2233758814 cites W2169901824 @default.
- W2233758814 cites W2174706414 @default.
- W2233758814 cites W2197919320 @default.
- W2233758814 cites W2263798363 @default.
- W2233758814 cites W2316514487 @default.
- W2233758814 cites W2567948266 @default.
- W2233758814 cites W2596677668 @default.
- W2233758814 cites W2911761077 @default.
- W2233758814 cites W3127518054 @default.
- W2233758814 cites W76530103 @default.
- W2233758814 cites W94637519 @default.
- W2233758814 cites W2082868123 @default.
- W2233758814 hasPublicationYear "2015" @default.
- W2233758814 type Work @default.
- W2233758814 sameAs 2233758814 @default.
- W2233758814 citedByCount "0" @default.
- W2233758814 crossrefType "dissertation" @default.
- W2233758814 hasAuthorship W2233758814A5043278211 @default.
- W2233758814 hasConcept C11413529 @default.
- W2233758814 hasConcept C119857082 @default.
- W2233758814 hasConcept C124101348 @default.
- W2233758814 hasConcept C154945302 @default.
- W2233758814 hasConcept C23123220 @default.
- W2233758814 hasConcept C41008148 @default.
- W2233758814 hasConceptScore W2233758814C11413529 @default.
- W2233758814 hasConceptScore W2233758814C119857082 @default.
- W2233758814 hasConceptScore W2233758814C124101348 @default.
- W2233758814 hasConceptScore W2233758814C154945302 @default.
- W2233758814 hasConceptScore W2233758814C23123220 @default.
- W2233758814 hasConceptScore W2233758814C41008148 @default.
- W2233758814 hasLocation W22337588141 @default.
- W2233758814 hasOpenAccess W2233758814 @default.
- W2233758814 hasPrimaryLocation W22337588141 @default.
- W2233758814 hasRelatedWork W1019876969 @default.