Construction And Application Of Intelligent Recommendation System In Rehabilitation Training For Autistic Children

Autism is a developmental disorder caused by a neurological disorder,which has been gaining attention worldwide in recent years as its incidence and prevalence have increased.In 2006,the country introduced a series of preferential policies to assist families of patients.Because of the specific nature of the disease,the social resources currently devoted to it still do not cover the majority of the patient population.As of the end of 2018,there were more than 1,000 autism institutions registered with the China Disabled Persons Federation,but many were founded by parents,and the regularity of the institutions cannot be guaranteed.Increased investment in education and improvement of past educational methods are therefore key to alleviating the social problems caused by the disease.In recent years,the rapid development of artificial intelligence technology,in the field of transportation,film and television,heavy industry and other fields have been applied to bring about a great change in society,the recommended technology as an important branch of artificial intelligence technology,becoming an important technical means of profitability for commercial companies.It also demonstrates significant application value in areas such as book recommendations for school libraries and friend recommendations for social networks.This paper collects data on the PEP3 scale assessment of children with autism,and based on the restricted Boltzmann machine’s collaborative filtering model,two parts of the work are done around the pre-construction and post-optimization of the recommendation system system system:the first part is a study on the cold start of the recommendation system,i.e.,the initial data set of children’s behavior is generated based on the information already available before the recommendation system is applied.First,the data pre-processing of the text-based rehabilitation training program was performed,topic classification was completed through the topic model,and then the children’s ability deficits were assessed using the PEP3 scale,and a matrix of relationships between abilities and topics was obtained through the word2vec algorithm and similarity calculations,and the initial data set of children’s behavior was generated with reference to the document’s topic classification results.The second part is the design of the recommended scheme,i.e.,building a suitable recommended model based on the available initial data.First,the initial dataset generated by cold start is used as the training set and the restricted Boltzmann machine(RBM)model is used as the recommended model,considering the structural characteristics of RBM,a large number of visible layer neurons will cause the model’s accuracy to decrease,so the implicit semantic model(LFM)is introduced,the prediction results of the RBM model are reordered,and finally the Top-N recommendation is completed based on the prediction score.In this paper,experimental comparison and results analysis were performed on the dataset generated during the cold start step.The mixed recommendation model based on RBM-LFM showed an overall improvement in the accuracy and recall rate of the recommendation results compared to the single RBM recommendation model.The innovative point of this paper is to attempt to combine automated recommendation techniques with research in the field of autism to provide a solution for changing the traditional approach to rehabilitation education.Secondly,in the cold start session of the recommended system,the problem of difficulty in completing the cold start of the system due to the complexity of autism was solved by the thematic model construction and the application of the word2vec algorithm to link the text-based rehabilitation training program with the child’s assessment results to generate the initial behavioral data set of the child.The recommended system provides individualized rehabilitation training programs for children based on the results of the scale assessment,so that parents can correct their children’s developmental problems anytime and anywhere according to the rehabilitation training program,which can not only greatly reduce labor costs,but also break through the limitations of where children are educated,improve the efficiency of rehabilitation training,and relieve the social and family pressure caused by the disease.