Recommend Model Reaearch Based On The Context-Aware And User Preference For The Blind Route

As the specially disadvantaged group of society, the blind with severe visual impairments do not have access to information resources fairly, like the normal persons, so they always have more problems in information acquisition and interaction. Especially, when their context information changes, they will be at a loss. Therefore, it is helpful to study how to recommend the blind the most suitable behavior so that they can gain the biggest perceived benefits when the context information changes, according to their preferences. Besides, to a certain extent, the solution also can solve the problem of the information barriers for the blind, which can make them truly enjoy the convenience brought from the information technology. Thus, it is a significant research subject in the information age.Based on the context-aware technology and BP neural network technology, this paper does some research on the personalized recommendation model for the blind, and the main research contents include the following aspects.Firstly, this paper analyzes all the context information related to the blind preferences in detail, and constructs the context information set expressed by tuples, including the external context information set and the blind’s schedule information set. Among them, the external context information set has three aspects, which are the state of activity areas, the conversion state between different activity areas and the emergency. And the blind’s schedule information set is consist of the blind’s fixed and temporary schedule. Besides, this paper does the preprocessing for them, deletes those recommend behaviors which do not conform to the actual conditions from the initial recommended list. At last, the second recommended list generates.Secondly, this paper deeply analyzes all the activity areas, extract their attributive characters, and then build their attributive character vectors in different situations. Based on them, we divide all the recommend behaviors. At the same time, considering users’different influences on the characters of different activity areas, we use a bias function to measure the blind’s preference degree for each attributive character. At last, combining with bias function and preference function, we build the blind preference model based on context-aware. Thirdly, on the basis of the above study, this paper proposes a model framework based on context-aware and user preference for the blind route. In this model, we use BP neural network to train the blind preference model and validate it. Thus, we can get different preferences in different situations through the network. And then, we can get the finally recommended list successfully due to analyzing the second recommended list and the predicted preference about the special activity area comprehensively, and then recommend the appropriate behavior so that the blind gain the biggest perceived benefits. In the last of this paper, we do the model simulation with all the collected information of the blind students, and extract and analyze contrastively the different experiments of two blind students which can illustrate the usefulness and effectiveness of the proposed model.