Research On Key Technologies Of Personalized English Learning System

Nowadays it is necessary for people to be equipped with a good knowledge ofEnglish.English learning is an integrated process joining vocabulary memorizing,grammar rules learning and exercises of listening, and all these sections areinterrelated. The ideal process of English should combine those sections as a wholerather than independent parts. Characteristics of English learning put forward ademand for assisted English learning system: How to build a multifunctionalintegrated English learning environment to allow users to learn English moreconveniently. Meanwhile,with the continuous development of the Internet,theamount of English learning materials enriched constantly,so it becomes veryimportant to help users to find English learning materials from vast amount of datawhich are suitable for themselves,it directly affects the learning time cost andlearning interests of users. We put theories of natural language processing,information retrieving and machine learning into practice to satisfy the demandsfaced in personalized resource organization and query in English learning assistingsystem. Against the English learning problems mentioned above,this paper mainlyincludes the following aspects:Firstly,to organize resources effectively and provide data support to analyzethe English level of users,we propose a classification method for English documentsbased on the degree of difficulty in English reading.Specifically speaking, we takeensemble learning framework as a basis,and train a machine learning model basedon different feature spaces(grammatical features, unigram model, non-text feature)to improve classification results.We prove and analyze the effectiveness of ouralgorithm through adopting accuracy,recall and F-measure,we finally get andifficulty level discriminant model which is effectively used in practice.Secondly,to achieve the purpose of organizing resources based on different Englishlevels of users, we propose a document retrieval ranking algorithm based onmultiple similarity pattern, which is convenient for users to find matching text oftheir queries quickly and accurately. This paper mainly uses the techniques of usermodeling based on the pre part of our work,we also analyze and model againstuser’s English level(according to user’s historical information),then combining with the query terms and document based on text similarity measurement, we design aranking algorithm fuses the two similarity models eventually.Thirdly,we converge the above algorithm in local text retrieval framework andestablish a dynamic English resources retrieval system,we also propose a systematicsolution program to evaluate and gain English resources as well as construct indexfiles and query rapidly.Meanwhile, in order to solve problems in learning Englishfaced in reciting words and listening practice,the system introduces the functionof reading and listening of word lists.Finally,we construct a multifunctionalintegrated and personalized English learning environment for users.