China
Africa
Explore chapters and articles related to this topic
Research Trends on the Web
Published in Akshi Kumar, Web Technology, 2018
Previous work in the area of CIR has focused on three main themes: user profile modeling, query expansion, and relevance. User profile modeling: Focuses on exploiting the sources of evidence that more precisely include approaches to build the user profile that allows learning the user’s context by implicitly inferring the information from the user’s behavior and from external or local context sources.Query expansion: The query expansion approaches attempt to expand the original search query by adding further, new, or related terms. These additional terms are inserted into an existing query, either by the user (interactive query expansion, IQE) or by the retrieval system (automatic query expansion, AQE), with the intent to increase the accuracy of the search.Relevance feedback: The notion of relevance feedback (RF) is to take into account the results that are initially returned in response to the input query and provide a means for automatically reformulating a query to more accurately reflect a user’s interests.
Image Retrieval
Published in Ling Guan, Yifeng He, Sun-Yuan Kung, Multimedia Image and Video Processing, 2012
where Q and Q′ are the original query and the updated query, respectively, and DR′ and DN′ are the sets of the positive and negative images labeled by the user, and NR′ and NN′ are the set sizes. α, β, and γ are weights. This technique was used in the Multimedia Analysis and Retrieval System (MARS) [73] to replace the document vector with visual feature vectors. Experiments show that retrieval performance can be improved by using these relevance feedback approaches.
Human-Computer Interaction and the Web
Published in Julie A. Jacko, The Human–Computer Interaction Handbook, 2012
Helen Ashman, Declan Dagger, Tim Brailsford, James Goulding, Declan O’Sullivan, Jan-Felix Schmakeit, Vincent Wade
This technique is particularly successful, and it has been repeatedly demonstrated that RF can improve the performance of a search engine at comparatively little cost to the user or the system (Ruthven and Laimas 2003; Harman 1992b; although see Spink, Jansen, and Ozmultu (2000), for contrast). For example, Koenemann (1996) noted that the... availability and use of relevance feedback increased retrieval effectiveness; and increased opportunity for user interaction with and control of relevance feedback made the interactions more efficient and usable while maintaining or increasing effectiveness.
Block-based pseudo-relevance feedback for image retrieval
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2022
Relevance feedback (RF) can be used to decrease the semantic gap. Relevance feedback is a supervised active learning process intended to improve system performance by refining the results of original queries (Salton, 1971). The main idea, which is the so-called user-in-the-loop version, is to ask users to provide positive (relevant) and/or negative (irrelevant) examples as feedback on the sets initially retrieved. In principle, relevance feedback is based on learning a set of ‘optimal’ feature weightings for a query or moving the query point towards the relevant objects/images (Tian, 2018; Zhou & Huang, 2003). In the literature, different RF techniques have been proposed for various domain problems, such as 3-D object retrieval (Leng et al., 2015), region-based image retrieval (Papadopoulos et al., 2014), sketch-based image retrieval (Qian et al., 2016), medical image retrieval (Banerjee et al., 2018), trademark image retrieval (Pinjarkar et al., 2020), and document retrieval (Raiber & Kurland, 2019; L. Wang et al., 2020).
Content-based image retrieval: A review of recent trends
Published in Cogent Engineering, 2021
Ibtihaal M. Hameed, Sadiq H. Abdulhussain, Basheera M. Mahmmod
HOG does not offer any spatial information about neighboring pixels. It offers only the orientation of the pixel under study. Co-occurrence histogram of oriented gradient (CoHOG) (Watanabe et al., 2010) overcomes this limitation. Baig et al. (Baig et al., 2020) introduced a CBIR system to reduce the semantic gap based on the merging of the benefits of CoHOG and SURF to cope with each other’s limitation. CoHOG depends on local discriminant analysis to reduce the dimension of each feature vector of CoHOG and SURF to decrease the computational cost. Relevance feedback is used to enhance the specificity (precision) and sensitivity (recall) with the help of the user by indicating the relevant and irrelevant images retrieved from the search of the database. The proposed system’s performance was assessed using Corel 1k, Corel 1.5k, Scene 15, and Caltech 256, and it recorded positive results; however, it has not been tested on large datasets.