Language models used in current automatic speech recognition systems are trained on general-purpose corpora and are therefore not relevant to transcribe spoken documents dealing w...
—The Web 2.0 era is characterized by the emergence of a very large amount of live content. A real time and finegrained content filtering approach can precisely keep users upto-...
Weixiong Rao, Lei Chen 0002, Pan Hui, Sasu Tarkoma
A new segmentation algorithm for multifont Farsi/Arabic texts based on conditional labeling of up and down contours was presented in [1]. A preprocessing technique was used to adju...
Mona Omidyeganeh, Reza Azmi, Kambiz Nayebi, Abbas ...
This paper introduces a content-based information retrieval method inspired by the ideas of spreading activation models. In response to a given query, the proposed approach comput...
Combining the output from multiple retrieval sources over the same document collection is of great importance to a number of retrieval tasks such as multimedia retrieval, web retr...