RNN-based demand awareness in smart library using CRFID

To provide more intelligence service in the smart library, we need to better perceive the reader's preferences. In addition to perceiving online records based on readers' search history and borrowing records, advanced information technologies give us more chance to perceive the behavior of readers in the actual reading process and further discover the need for reading. In this paper, we use CRFID and RNN deep learning network to recognize book motions in the reading process, so as to judge readers' need degree for the book, which can provide a basis for library book purchases and readers personalized service. In order to improve the recognition accuracy, we use the RSS as well as acceleration magnitude gathered from CRFID as the input data for RNN, and design a new encoding scheme. We trained and tested the deep learning network using real-world data, recorded during actual reading in our lab environment which mimics a typical reading room, from the experimental results, we conclude that our approach is feasible to recognize different reading phase to perceiving the needs of the readers.