Querying structured documents based on human visual memory

Current information retrieval (IR) is mainly focused on retrieving documents that have not been read before, and to our knowledge, no research has been conducted for querying against already read documents. For such documents, querying based on visual memory is often helpful. In this paper, a framework for querying structured documents by making use of human visual memory is proposed, and an index structure suitable for such a framework is presented. Specifically, we discuss our proposal in the context of personalization, where a vast amount of primary source documents are reconstructed for end-users reflecting their preferences. We also mention some clues to implement our framework as well as future directions