Search (3 results, page 1 of 1)

0.02477492 = product of:
  0.07432476 = sum of:
    0.07432476 = weight(_text_:objects in 520) [ClassicSimilarity], result of:
      0.07432476 = score(doc=520,freq=2.0), product of:
        0.25313336 = queryWeight, product of:
          5.315071 = idf(docFreq=590, maxDocs=44218)
          0.047625583 = queryNorm
        0.29361898 = fieldWeight in 520, product of:
          1.4142135 = tf(freq=2.0), with freq of:
            2.0 = termFreq=2.0
          5.315071 = idf(docFreq=590, maxDocs=44218)
          0.0390625 = fieldNorm(doc=520)
  0.33333334 = coord(1/3)

Abstract

The rapid growth of the numbers of images and their users as a result of the reduction in cost and increase in efficiency of the creation, storage, manipulation, and transmission of images poses challenges to those who organize and provide access to images. One of these challenges is similarity matching, a key component of current content-based image retrieval systems. Similarity matching often is implemented through similarity measures based on geometric models of similarity whose metric axioms are not satisfied by human similarity judgment data. This study is significant in that it is among the first known to test Tversky's contrast model, which equates the degree of similarity of two stimuli to a linear combination of their common and distinctive features, in the context of image representation and retrieval. Data were collected from 150 participants who performed an image description and a similarity judgment task. Structural equation modeling, correlation, and regression analyses confirmed the relationships between perceived features and similarity of objects hypothesized by Tversky. The results hold implications for future research that will attempt to further test the contrast model and assist designers of image organization and retrieval systems by pointing toward alternative document representations and similarity measures that more closely match human similarity judgments.

0.02477492 = product of:
  0.07432476 = sum of:
    0.07432476 = weight(_text_:objects in 3250) [ClassicSimilarity], result of:
      0.07432476 = score(doc=3250,freq=2.0), product of:
        0.25313336 = queryWeight, product of:
          5.315071 = idf(docFreq=590, maxDocs=44218)
          0.047625583 = queryNorm
        0.29361898 = fieldWeight in 3250, product of:
          1.4142135 = tf(freq=2.0), with freq of:
            2.0 = termFreq=2.0
          5.315071 = idf(docFreq=590, maxDocs=44218)
          0.0390625 = fieldNorm(doc=3250)
  0.33333334 = coord(1/3)

Abstract

This paper reports on a research project that compared two different approaches for the indexing of ordinary images representing common objects: traditional indexing with controlled vocabulary and free indexing with uncontrolled vocabulary. We also compared image retrieval within two contexts: a monolingual context where the language of the query is the same as the indexing language and, secondly, a multilingual context where the language of the query is different from the indexing language. As a means of comparison in evaluating the performance of each indexing form, a simulation of the retrieval process involving 30 images was performed with 60 participants. A questionnaire was also submitted to participants in order to gather information with regard to the retrieval process and performance. The results of the retrieval simulation confirm that the retrieval is more effective and more satisfactory for the searcher when the images are indexed with the approach combining the controlled and uncontrolled vocabularies. The results also indicate that the indexing approach with controlled vocabulary is more efficient (queries needed to retrieve an image) than the uncontrolled vocabulary indexing approach. However, no significant differences in terms of temporal efficiency (time required to retrieve an image) was observed. Finally, the comparison of the two linguistic contexts reveal that the retrieval is more effective and more efficient (queries needed to retrieve an image) in the monolingual context rather than the multilingual context. Furthermore, image searchers are more satisfied when the retrieval is done in a monolingual context rather than a multilingual context.

0.0032263044 = product of:
  0.009678913 = sum of:
    0.009678913 = product of:
      0.019357827 = sum of:
        0.019357827 = weight(_text_:22 in 4054) [ClassicSimilarity], result of:
          0.019357827 = score(doc=4054,freq=2.0), product of:
            0.16677667 = queryWeight, product of:
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.047625583 = queryNorm
            0.116070345 = fieldWeight in 4054, product of:
              1.4142135 = tf(freq=2.0), with freq of:
                2.0 = termFreq=2.0
              3.5018296 = idf(docFreq=3622, maxDocs=44218)
              0.0234375 = fieldNorm(doc=4054)
      0.5 = coord(1/2)
  0.33333334 = coord(1/3)

Date

6. 1.2011 11:22:12