Topic Mining Research Based On Learner's Background Information

Nowadays,information technology is increasingly developing.The world is changed by Internet.Massive open online course becomes the hot topic in educational field.A varieties of learning platform becomes the most important that lots of learners publishes their text and restores their learning data.The educational basic theories considered that the best education is promoting three elements(educator;learner;educational influence)cooperate and understand each other.My paper based on educational theories,according to the text data of learners from the course discussion area of China Center Normal University Learning Cloud Platform,combines the data of learner's background data;try to study the text data of learners form the learning platform.The purpose is to find out that which dilemma they may meet and which topic they may be interest.I hope to build the contact for learner,educator and educational manager so that education can be timely adjustment and the best education can be created.By the massive educational data produced,data analysis came into being.Text data analysis is the important technology for study the interest of learners.In the text data analysis,deal with Chinese text is the difficult point.Generally,scholars take the TF-IDF algorithm to characterize the text or participle the Chinese text.The traditional text analysis means is the means that label the text data,then classify these text data combine the machine leaning.Generally,these means is to divide the text data into two categories.For example,these texts are the exploratory or non-exploratory?The specific algorithm is naiveBayesian,support vector machine,etc.Subsequently,the probability topic model obtains the extensive note.The model is the typical hierarchical model for the text data based on Bayesian probability theory.It is the model that builds the connection with the text and topic and word.Latent Dirichlet Allocation semantics model is the typical topic model,the purpose is to automatically mine the interest topic of learners.This model has obvious advantage comparison to the traditional classification.It will have more flexible ways to carry out study and apply to the epoch data explosion.This paper put forward a Latent Dirichlet Allocation Model with parameters expended.This model will mine text topic combine with the background data of learners.It can automatically mines the topic in the massive text of learners and understand the interest topic of different sex and discipline link to the background data of learners.The experimental results of topic miningdemonstrate the learners'dilemma and the differences of interest for learners have different sex and discipline.For example,learners major in math prefer to study the knowledge of psychology.In the topic about prenatal education,the sex of learners is almost female.The topic mining combine with the background data of learners can study the interest for different learners.Their group differences can be seen clearly.Data visualization is a hotspot in recent years.This paper lists the commonly used data visualization techniques and text visualization techniques.The text visualization of the experimental result clearly shows the hotspot for multiple categories of learners.Which greatly improves the readability and comprehensibility of text data?In the future,it is expected that more research can combine the learners' data with their text data.It will be an important research point for adding time parameters.The study of data mining for learners will be the key to the integration of information technology and education.