Concept-Based Retrieval of Art Documents

This paper presents our work on the retrieval of art documents for color artistry concepts. First we show that the query-byexample paradigm popularly used in content-based retrieval can support only limited queryability. The paper then proposes a concept-based retrieval engine based on the generative grammar of elecepts methodology. In the latter, the language by which color artistry concepts are communicated in art documents is used to operate the retrieval processes. The concept language is explicated into a lexicon of elecepts and the associated generative grammar. Documents are then indexed with elecept indices, while the generative grammar is used to facilitate the query operation. More extensive color artistry concept queries can then be supported by post-coordination of the elecept indices.