Retrieving electronic ink by content

This paper deals with a new emerging multimedia data, namely, electronic ink. As pen-based computers and personal digital assistants (PDAs) become more popular, searching electronic ink becomes an important issue. We treat the ink object as a one dimensional image which is described by a set of features. This is different from the traditional methods that use handwritten recognition techniques to convert it into ASCII. The search retrieves a set of ink objects that most resembles the query object. We describe a multi-stages filter for searching large repository of electronic ink. The first stage is an R-tree based index used to prune the search space. The output of the first stage is a set of words that have some common features with the query. A sequential search algorithm is then used to extract the most similar word to the query string. Our schema is 12 time faster than the sequential and improves the retrieval rate by up to 50%