[Objective] Literature collection and reading is a key task of scientific research work, and the concentration status that researchers devote to it is directly related to the research efficiency. At present, the concentration of literature reading is mostly evaluated by manual methods or eye tracking methods, in order to realize the automatic detection and real-time feedback of the concentration evaluation process, this paper combines technology and concentration evaluation research, which is also meaningful to study the application of intelligent technology in smart knowledge service. [Methods]Detecting the head posture by the vertical and horizontal rotation angle of the reader's head, the closing eyes or yawning status by the eye and mouth closure and to score the fatigue and the emotion based on these expression recognition results. Based on the scores, the fuzzy comprehensive evaluation algorithm is used to determine the weights of relevant factors and integrate the model to output the reader's concentration status at different moments in their reading process. [Results]In this paper, the model is applied to the actual reading scene to simulate and evaluate the reading concentration of head tilt, fatigue and negative emotional states, and the results are 26.3%, 25.2% and 6.8% lower than the normal state, respectively. [Limitations]The model in this paper is limited by the limitations of visual recognition technology, the accuracy has a certain room for improvement, and there are some extreme reading examples that need to be optimized. [Conclusions]The proposed methods can be applied to downstream tasks in multiple fields, which can assist researchers to adjust literature reading strategies in time to improve reading efficiency. And it is also helpful for libraries and other related departments to formulate book acquisition strategies, thus reducing the waste of book resources.