Saturday, April 19, 2014
   
Text Size
New ranking methods
Research papers on new search ranking methods and concepts

DocumentsDate added

Order by : Name | Date | Hits [ Descendent ]
"The main goal of this paper is to customize the Web for specific feature and/or community graph, finding out the confidence of each page in the graph in question from the past experience and calculate the page rank of the pages in the graph from confidence obtained and link structure. We view the Web in the Universe from the users query points of view and customize accordingly".
"In this paper, we propose a new method to discover collectionadapted ranking functions based on Genetic Programming (GP). Our Combined Component Approach (CCA) is based on the combination of several term-weighting components (i.e., term frequency, collection frequency, normalization) extracted from well-known ranking functions."
"We address the task of learning rankings of documents from search engine logs of user behavior. Previous work on this problem has relied on passively collected clickthrough data. In contrast, we show that an active exploration strategy can provide data that leads to much faster learning."
"The paper is concerned with applying learning to rank to document retrieval. Ranking SVM is a typical method of learning to rank. We point out that there are two factors one must consider when applying Ranking SVM, in general a “learning to rank” method, to document retrieval. First, correctly ranking documents on the top of the result list is crucial for an Information Retrieval system. One must conduct training in a way that such ranked results are accurate. Second, the number of relevant documents can vary from query to query."
"Maximizing only the relevance between queries and documents will not satisfy users if they want the top search results to present a wide coverage of topics by a few representative documents. In this paper, we propose two new metrics to evaluate the performance of information retrieval: diversity, which measures the topic coverage of a group of documents, and information richness, which measures the amount of information contained in a document. Then we present a novel ranking scheme, Affinity Rank, which utilizes these two metrics to improve search results. We demonstrate how Affinity Rank works by a toy data set, and verify our method by experiments on real-world data sets".
"Spam in the form of link spam and click spam has become a major obstacle in the effective functioning of ranking and reputation systems. Even in the absence of spam, difficulty in eliciting feedback and self-reinforcing nature of ranking systems are known problems. In this paper, we make a case for sharing with users the revenue generated by such systems as incentive to provide usefulfeedback and present an incentive based ranking scheme in a realistic model of user behavior which addresses the above problems".
"In this paper, we conduct a study on the approach of directly optimizing evaluation measures in learning to rank for Information Retrieval (IR). We focus on the methods that minimize loss functions upper bounding the basic loss function defined on the IR measures."
"Our purpose with this work is threefold: rst, in enumerating the various measures in an orthogonal framework we make it straightforward for other researchers to describe and discuss similarity measures; second, by experimenting with a wide range of the measures, we hope to observe which features yield good retrieval behaviour in a variety of retrieval environments; and third, by describing our results so far, to gather feedback on the issues we have uncovered. We demonstrate that it is surprisingly dicult to identify which techniques work best, and comment on the experimental methodology required to support any claims as to the superiority of one method over another."
Page 1 of 2
Please update your Flash Player to view content.
Restore Default Settings