Benitez, A.B., Beigi, M., Chang, S.F.: Using relevance feedback in content-based image metasearch. IEEE Internet Computing (1998) 59-69  Home/Search   Document Details and Download   Summary   Related Articles   Check

....class is a cue for similarities in their semantic content. While relevance feedback has achieved prevailing popularity in CBIR, less research has been focused on exploiting long term learning. Still, some form of long term learning has been incorporated into a number of CBIR systems. In MetaSeek [1], all user interactions were stored and used in later queries in selecting between a set of independent image search engines. The log les of the Viper CBIR system were used to adjust weights for di erent features in [2] A Bayesian framework for both short time and long time learning was ....

Benitez, A.B., Beigi, M., Chang, S.F.: Using relevance feedback in content-based image metasearch. IEEE Internet Computing (1998) 59-69

....not based on user interaction, to get from low level features to high level semantics. User based approaches, on the other hand, are adaptive to user behavior and try to construct individual profiles. An important component of most user based approaches is the technique of relevance feedback [2, 28], which has not been generally used on the web yet, especially for images. Some of the examples of system based approaches are [8, 20, 27, 31] 27] is the first paper that concerns retrieving images, in this case, graphic objects, based on user semantics. A methodology for composing features ....

A. B. Benitez, M. Beigi, and S. F. Chang, Using Relevance Feedback in ContentBased Image Metasearch, IEEE Internet Computing, Volume 2, Number 4, JulyAugust 1998, pp. 59-69.

....methods can be found in the literature [11, 55] yet most do not meet the efficiency requirements of Web multimedia searches, which expect both a prompt response and the search of a huge volume of Web multimedia data. A few content based image and video search engines are available on line [5, 22, 36, 49, 52]. Various indexing methods are applied to locate the desirable images or video. The major technologies include using camera object motion, colors, examples, locations, positional color texture, shapes, sketches, text, and texture as well as relevance feedback [18] However, a de facto Web image or ....

Ana B. Benitez, Mandis Beigi, and Shih-Fu Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, 2(4):59-69, July/August 1998.

....multimedia indexing methods can be found in the literature [10] yet most do not meet the efficiency requirements of Web multimedia searches, which expect both a prompt response and the search of a huge volume of Web multimedia data. A few image and video search engines are available on line [1, 4, 13, 30], but audio search engines have yet to be constructed since audio information retrieval [11] is considered to be one of the most difficult challenges for multimedia retrieval. 8. OTHERS Apart from the above major search techniques, some ad hoc methods worth mentioning include: ffl A natural ....

Ana B. Benitez, Mandis Beigi, and Shih-Fu Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, 2(4):59-69, July/August 1998.

....content based image retrieval. The best known system is Query By Image Content (QBIC) 2] developed at the IBM Almaden Research Center. Other notable systems include MIT s Photobook [3] and its more recent version, FourEyes [4] the search engine family of VisualSEEk [5] WebSEEk [6] and MetaSEEk [7], which all are developed at Columbia University, and Virage [8] a commercial content based search engine developed at Virage Technologies Inc. We introduce here a recently implemented image retrieval system called PicSOM. It uses a World Wide Web browser as the user interface and the Tree ....

A. B. Benitez, M. Beigi, and S.-F. Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, pages 59--69, July-August 1998.

....process takes place. Applications The database consists of images of ancient African artifacts from the Fowler Museum of Cultural History at UCLA. 22 MetaSEEk Developer Image and Advanced Television Lab, Columbia University, NY, USA. URL http: www.ctr.columbia.edu metaseek . References [BBC98] Features Content based retrieval can be done on the basis of color and texture. MetaSEEk has a color or texture matching and indexing of its own to cluster query images in a local performance database, used to select target serach engines. The actual image matching is forwarded to QBIC, VIR ....

A. B. Benitez, M. Beigi, and S.-F. Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, 2(4):59-69, July/August 1998. 44

....increases dramatically over searching only one engine. Lawrence and Giles found that combining the results of 11 major search engines increased the coverage to about 42 of the estimated size of the publicly indexable web. A consistent interface is necessary for a metasearch engine to be useful [4, 10]. Such an interface ensures that results from several places can be meaningfully combined, while insulating the user from the specifics of the underlying search engines. In this paper, we describe the architecture of the next generation of Inquirus, the metasearch tool at NEC Research Institute. ....

....use of decision theory as a means of making result ordering decisions. In addition to properly ordering the results, choosing where to look affects the overall search precision. To increase the precision of the results, some metasearch engines such as ProFusion [5] SavvySearch [8] or MetaSEEK [4] do not always send the user s query to the same search engines. ProFusion considers the performance, the predicted subject of the query and the user s explicit search engine preferences. On request, ProFusion downloads individual pages to check for broken links and duplicates. MetaSEEK considers ....

[Article contains additional citation context not shown here]

Ana B. Benitez, Mandis Beigi, and Shih-Fu Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, 2(4):58--69, 1998.

....to search indexed web pages by entering a keyword query. The basic architecture of a regular Internet search engine is shown in Figure 2. Research has shown that searching more than one search engine can significantly enhance coverage, locating significantly more results than one engine alone [3, 9, 14]. In addition, there are several specialized content web tools, such as ABCNews (www.abcnews.com) for example, which specializes in News, or Yahoo (www.yahoo.com) which is a manually built hierarchy by subject. Figure 2: The architecture of a standard search engine with no feedback A user might ....

....engine or might not have the patience to submit their query Figure 3: The architecture of a standard metasearch engine to more than one. As a result metasearch engines were created. Some metasearch engines include: Metacrawler (www.metacrawler.com) SavvySearch [9] savvysearch.com) MetaSEEK [3] (http: www.ctr.columbia.edu MetaSEEk ) and ProFusion [6] profusion.com) Figure 3 shows the architecture of a typical metasearch engine. Compared with a regular search engine, a metasearch engine does not have a local database and relies on other sources (other search engines) for its data. ....

[Article contains additional citation context not shown here]

Ana B. Benitez, Mandis Beigi, and Shih-Fu Chang. Using relevance feedback in content-based image metasearch. IEEE Internet Computing, 2(4):58--69, 1998.

....these systems (see Figure 4) 5.2 Metasearch engine The second demo we are developing is a next generation metasearch engine using the proposed image DS. Metasearch engines act as gateways linking users automatically and transparently to multiple search engines. Our metasearch engine, MetaSEEk [1], explores the issues involved in querying large, distributed, on line visual information systems [2] In this section, we will describe how the proposed image DS will be used to enhance the metasearch systems. MetaSEEk is designed to intelligently select and interface with multiple on line image ....

A. B. Benitez, M. Beigi, and S.-F. Chang, "Using Relevance Feedback in Content-Based Image Metasearch", IEEE Internet Computing, Vol. 2, No. 4, pp. 59-69, Jul./Aug. 1998; web site http://www.ctr.columbia.edu/metaseek/.

....(if that capability is available, as we propose in [1] and access the search engines description schemes. However, the only way for the metasearcher to adapt to the unknown and dynamic content distribution of a search engine would be to keep some kind of performance scores based on user feedback [2]. This approach requires time and effort from part of the user. If MPEG 7 includes the description of collection or archives of multimedia documents and search engines make them available to the metasearcher, a more efficient solution would be reached in the lines of [3] 3 DESIGN PRINCIPLES AND ....

A. B. Benitez, M. Beigi, and S.-F. Chang, "Using Relevance Feedback in Content-Based Image Metasearch", IEEE Internet Computing, Vol. 2, No. 4, pp. 59-69, Jul./Aug. 1998.

....event ref= 2 Approaching to bridge event node event hierarchy 4.3 Metasearch engine Metasearch engines act as gateways linking users automatically and transparently to multiple search engines. Most of the current metasearch engines work with text. Our metasearch engine, MetaSEEk [2], explores the issues involved in querying large, distributed, on line visual information systems [4] In this section, we will describe the impact that an interoperable content description for multimedia data such as MPEG 7 can have in metasearch engines. MetaSEEk is designed to intelligently ....

A. B. Benitez, M. Beigi, and S.-F. Chang, "Using Relevance Feedback in Content-Based Image Metasearch", IEEE Internet Computing, Vol. 2, No. 4, pp. 59-69, Jul./Aug. 1998; web site http://www.ctr.columbia.edu/metaseek/.

.... Approaching to bridge event node event hierarchy 4.3. Metasearch engine Metasearch engines act as gateways linking users automatically and transparently to multiple search engines. Most of the current metasearch engines work with text. Our metasearch engine, MetaSEEk [2], explores the issues involved in querying large, distributed, on line visual information systems [4] In this section, we will describe the impact that an interoperable content description for multimedia data such as MPEG 7 can have in metasearch engines. MetaSEEk is designed to intelligently ....

A. B. Benitez, M. Beigi, and S.-F. Chang, "Using Relevance Feedback in Content-Based Image Metasearch", IEEE Internet Computing, Vol. 2, No. 4, pp. 59-69, Jul./Aug. 1998; web site http://www.ctr.columbia.edu/metaseek/.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC