scispace - formally typeset
Search or ask a question
Topic

Ranking (information retrieval)

About: Ranking (information retrieval) is a research topic. Over the lifetime, 21109 publications have been published within this topic receiving 435130 citations.


Papers
More filters
Journal ArticleDOI
TL;DR: This article describes an effort to study whether the order of document presentation to judges influences the relevance scores assigned to those documents.
Abstract: Studies concerned with the evaluation of information systems have typically relied on judgments of relevance as the fundamental measure in determining system performance. In most cases, subjects are asked to assign a relevance score using some category rating scale (1–4, 1–11, or simply relevant/non-relevant) to each document in a set retrieved in response to some information need or query. While the extensive studies of relevance conducted in the 1960s indicated that relevance judgments are influenced by a range of variables, little attention has been paid to the possible effects of the order in which the stimuli are presented to judges. This effect of “stimulus order” has been found to exist in measuring variables in other fields (Stevens 1975, Gescheider 1985). Questioning possible “presentable order effects” is particularly appropriate in that systems are being developed and evaluated in information science which present documents in some systematic way (e.g., with the documents considered by the system to be most relevant presented first). This article describes an effort to study whether the order of document presentation to judges influences the relevance scores assigned to those documents. A query and set of documents with relevance judgments were available from a previous study. Subjects were randomly assigned one of two orders (one ranked high to low, the other low to high) of fifteen document descriptions. They were then asked to assign a score to each document description to match their judgment of relevance in relation to the stated information query. Both a category rating (1–7) and open-ended, magnitude estimation scaling procedure were tested, and it was found that the judgments were influenced by the order of document presentation. © 1988 John Wiley & Sons, Inc.

118 citations

Proceedings ArticleDOI
20 Jun 2011
TL;DR: This paper proposes a cascaded architecture which using the ranking SVM generates an ordered set of proposals for windows containing object instances, and demonstrates that this approach is robust, achieving higher overlap-recall values using fewer output proposals than the state-of-the-art.
Abstract: Object recognition has made great strides recently. However, the best methods, such as those based on kernel-SVMs are highly computationally intensive. The problem of how to accelerate the evaluation process without decreasing accuracy is thus of current interest. In this paper, we deal with this problem by using the idea of ranking. We propose a cascaded architecture which using the ranking SVM generates an ordered set of proposals for windows containing object instances. The top ranking windows may then be fed to a more complex detector. Our experiments demonstrate that our approach is robust, achieving higher overlap-recall values using fewer output proposals than the state-of-the-art. Our use of simple gradient features and linear convolution indicates that our method is also faster than the state-of-the-art.

118 citations

Proceedings ArticleDOI
Deok-Hwan Kim1, Chin-Wan Chung1
09 Jun 2003
TL;DR: This paper proposes a new content-based image retrieval method using adaptive classification and cluster-merging to find multiple clusters of a complex image query when the measures of a retrieval method are invariant under linear transformations, and the method can achieve the same retrieval quality regardless of the shapes of clusters of the query.
Abstract: The learning-enhanced relevance feedback has been one of the most active research areas in content-based image retrieval in recent years. However, few methods using the relevance feedback are currently available to process relatively complex queries on large image databases. In the case of complex image queries, the feature space and the distance function of the user's perception are usually different from those of the system. This difference leads to the representation of a query with multiple clusters (i.e., regions) in the feature space. Therefore, it is necessary to handle disjunctive queries in the feature space.In this paper, we propose a new content-based image retrieval method using adaptive classification and cluster-merging to find multiple clusters of a complex image query. When the measures of a retrieval method are invariant under linear transformations, the method can achieve the same retrieval quality regardless of the shapes of clusters of a query. Our method achieves the same high retrieval quality regardless of the shapes of clusters of a query since it uses such measures. Extensive experiments show that the result of our method converges to the user's true information need fast, and the retrieval quality of our method is about 22% in recall and 20% in precision better than that of the query expansion approach, and about 34% in recall and about 33% in precision better than that of the query point movement approach, in MARS.

118 citations

Proceedings Article
22 May 2006
TL;DR: This work proposes a suitable extension of label ranking that incorporates the calibrated scenario, and suggests a conceptually novel technique for extending the common learning by pairwise comparison approach to the multilabel scenario, a setting previously not being amenable to the pairwise decomposition technique.
Abstract: Label ranking studies the problem of learning a mapping from instances to rankings over a predefined set of labels. Hitherto existing approaches to label ranking implicitly operate on an underlying (utility) scale which is not calibrated in the sense that it lacks a natural zero point. We propose a suitable extension of label ranking that incorporates the calibrated scenario and substantially extends the expressive power of these approaches. In particular, our extension suggests a conceptually novel technique for extending the common learning by pairwise comparison approach to the multilabel scenario, a setting previously not being amenable to the pairwise decomposition technique. We present empirical results in the area of text categorization and gene analysis, underscoring the merits of the calibrated model in comparison to state-of-the-art multilabel learning methods.

118 citations

Patent
11 Mar 2004
TL;DR: In this article, a co-occurrence ranker was used to identify topics of document data, and a phrase selector for selecting the highest ranking phrases as the topic or topics of the document data; and an output device for outputting data relating to the selected topics.
Abstract: Apparatus for identifying topics of document data has: a word ranker (171) for ranking words that are present in or representative of the content of the document data; a co-occurrence ranker (172) for ranking co-occurrences of words that are present in or representative of the content of the document data; a phrase ranker (170) for ranking phrases in the document data; a word selector (174) for selecting the highest ranking words; a co-occurrence identifier (176) for identifying which of the highest ranking co-occurrences contain at least one of the highest ranking words; a phrase identifier (177) for identifying the phrases containing at least one word from the identified co-occurrences; a phrase selector (178) for selecting the highest ranking one or ones of the identified phrases as the topic or topics of the document data; and an output device (40) for outputting data relating to the selected topics.

118 citations


Network Information
Related Topics (5)
Web page
50.3K papers, 975.1K citations
83% related
Ontology (information science)
57K papers, 869.1K citations
82% related
Graph (abstract data type)
69.9K papers, 1.2M citations
82% related
Feature learning
15.5K papers, 684.7K citations
81% related
Supervised learning
20.8K papers, 710.5K citations
81% related
Performance
Metrics
No. of papers in the topic in previous years
YearPapers
20241
20233,112
20226,541
20211,105
20201,082
20191,168