FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2283472011> ?p ?o ?g. }

• W2283472011 abstract "Information retrieval algorithms attempt to match a user's description of their information need with relevant information in a collection of documents or other data. Applications include Web search engines, filtering and recommendation systems, computer-assisted language tutors, and many others. A key challenge of retrieval algorithms is to perform effective matching when many factors, such as the user's true information need, may be highly uncertain and can only be partially observed via a small number of keywords. This dissertation develops broadly applicable algorithms for measuring and exploiting such uncertainty in retrieval algorithms to make them more effective and reliable. Our contributions include new theoretical models, statistical methods, evaluation techniques, and retrieval algorithms. As an application, we focus on a long-studied approach to improving retrieval matching that adds related terms to a query — a process known as query expansion. Query expansion works well on average, but even state-of-the-art methods are still highly unreliable and can greatly hurt results for individual queries. We show how sensitivity information for an expansion algorithm can be obtained and used to improve its reliability without reducing overall effectiveness. Our approach proceeds in two steps. First, treating the base expansion method as a 'black box', we gather information about how the algorithm's output — a set of expansion terms — changes with perturbations of the initial query and top-ranked documents. This step also results in a set of plausible expansion model candidates. We then introduce a novel risk framework based on convex optimization that prunes and combines these candidates to produce a much more reliable version of the original baseline expansion algorithm. Highlights of our results include: • A new algorithmic framework for estimating more precise query and document models, based on treating queries and document sets as random variables instead of single observations. • The first significant application and analysis of convex optimization methods to query expansion problems in information retrieval. • A new family of statistical similarity measures we call perturbation kernels that are efficient to compute and give context-sensitive word clustering. • The introduction of risk-reward analysis to information retrieval, including tradeoff curves, analysis, and risk measures. • A new general form of query difficulty measure that reflects clustering in the collection as well as the relation between a query and the collection." @default.
• W2283472011 created "2016-06-24" @default.
• W2283472011 creator A5005551272 @default.
• W2283472011 creator A5009879041 @default.
• W2283472011 date "2008-01-01" @default.
• W2283472011 modified "2023-09-26" @default.
• W2283472011 title "Robust model estimation methods for information retrieval" @default.
• W2283472011 cites W1496921672 @default.
• W2283472011 cites W1512558044 @default.
• W2283472011 cites W1568385842 @default.
• W2283472011 cites W1573190978 @default.
• W2283472011 cites W1646006088 @default.
• W2283472011 cites W174734731 @default.
• W2283472011 cites W190008395 @default.
• W2283472011 cites W1928657940 @default.
• W2283472011 cites W1973288683 @default.
• W2283472011 cites W1974782086 @default.
• W2283472011 cites W1976526581 @default.
• W2283472011 cites W1977261934 @default.
• W2283472011 cites W1979459060 @default.
• W2283472011 cites W1984599386 @default.
• W2283472011 cites W1987996059 @default.
• W2283472011 cites W1989885825 @default.
• W2283472011 cites W1990550922 @default.
• W2283472011 cites W1999817920 @default.
• W2283472011 cites W2002306339 @default.
• W2283472011 cites W2009923109 @default.
• W2283472011 cites W2017186527 @default.
• W2283472011 cites W2020662977 @default.
• W2283472011 cites W2028709054 @default.
• W2283472011 cites W2028716813 @default.
• W2283472011 cites W2030524533 @default.
• W2283472011 cites W2030603245 @default.
• W2283472011 cites W2034117506 @default.
• W2283472011 cites W2034701578 @default.
• W2283472011 cites W2036181041 @default.
• W2283472011 cites W2037140704 @default.
• W2283472011 cites W2048045485 @default.
• W2283472011 cites W2049540727 @default.
• W2283472011 cites W2052088591 @default.
• W2283472011 cites W2057028302 @default.
• W2283472011 cites W2062229217 @default.
• W2283472011 cites W2062270497 @default.
• W2283472011 cites W2066766178 @default.
• W2283472011 cites W2068632118 @default.
• W2283472011 cites W2068905009 @default.
• W2283472011 cites W2079370602 @default.
• W2283472011 cites W2087663869 @default.
• W2283472011 cites W2093390569 @default.
• W2283472011 cites W2107802551 @default.
• W2283472011 cites W2107961375 @default.
• W2283472011 cites W2112973517 @default.
• W2283472011 cites W2114512077 @default.
• W2283472011 cites W2117897510 @default.
• W2283472011 cites W2120084270 @default.
• W2283472011 cites W2122841972 @default.
• W2283472011 cites W2124158580 @default.
• W2283472011 cites W2128894810 @default.
• W2283472011 cites W2131381130 @default.
• W2283472011 cites W2132005946 @default.
• W2283472011 cites W2136144468 @default.
• W2283472011 cites W2138505392 @default.
• W2283472011 cites W2140666719 @default.
• W2283472011 cites W2142380402 @default.
• W2283472011 cites W2147152072 @default.
• W2283472011 cites W2150632918 @default.
• W2283472011 cites W2151531457 @default.
• W2283472011 cites W2155681181 @default.
• W2283472011 cites W2165014986 @default.
• W2283472011 cites W2165299010 @default.
• W2283472011 cites W2172251087 @default.
• W2283472011 cites W2197919320 @default.
• W2283472011 cites W2296319761 @default.
• W2283472011 cites W2767905780 @default.
• W2283472011 cites W2912934387 @default.
• W2283472011 cites W2950225692 @default.
• W2283472011 cites W2093257937 @default.
• W2283472011 cites W2963770254 @default.
• W2283472011 hasPublicationYear "2008" @default.
• W2283472011 type Work @default.
• W2283472011 sameAs 2283472011 @default.
• W2283472011 citedByCount "9" @default.
• W2283472011 countsByYear W22834720112012 @default.
• W2283472011 countsByYear W22834720112019 @default.
• W2283472011 crossrefType "journal-article" @default.
• W2283472011 hasAuthorship W2283472011A5005551272 @default.
• W2283472011 hasAuthorship W2283472011A5009879041 @default.
• W2283472011 hasConcept C105795698 @default.
• W2283472011 hasConcept C124101348 @default.
• W2283472011 hasConcept C157692150 @default.
• W2283472011 hasConcept C165064840 @default.
• W2283472011 hasConcept C177264268 @default.
• W2283472011 hasConcept C199360897 @default.
• W2283472011 hasConcept C23123220 @default.
• W2283472011 hasConcept C26517878 @default.
• W2283472011 hasConcept C33923547 @default.
• W2283472011 hasConcept C38652104 @default.
• W2283472011 hasConcept C41008148 @default.
• W2283472011 hasConcept C4969071 @default.
• W2283472011 hasConcept C99016210 @default.

• next