Abstract
Collaborative image tagging systems, such as Flickr, are very attractive for supporting keywordbased image retrieval, but some user-provided tags of collaboratively-tagged social images might be imprecise. Some people may use general or high-level words (i.e., abs tags) to tag their images for saving time and effort, but such general or high-level tags are too -abs to describe the visual content of social images precisely. As a result, users may not be able to find what they need when they use the specific for query specification. To tackle the problem of -abs tags, an ontology with threelevel semantics is constructed for detecting the candidates of -abs tags from large-scale social images. Then the image context (nearest neighbors) and tag context (most relevant tags) of social images with -abs tags are used to ultimately confirm whether these candidates are -abs or not and identify the specific tags which can further depict the images with -abs tags. In addition, all the relevant tags, which correspond with intermediate nodes between the -abs tags and specific tags on our concept ontology, are added to enrich the tags of social images so that users can have more choices to select various for query specification. We have tested our proposed algorithms on two types of data sets (revised standard datasets and selfconstructed dataset) and compared our approach with other approaches.
Original language | English |
---|---|
Pages (from-to) | 5-18 |
Number of pages | 14 |
Journal | Journal of Signal Processing Systems |
Volume | 74 |
Issue number | 1 |
DOIs | |
State | Published - Jan 2014 |
Keywords
- Abstract tag Z. Xia
- Concept ontology
- Image context
- J. Peng
- Tag context
- Tag enrichment
- Tag refinement
- X. Feng