Research On User Portrait In Android Platform Based On Emotional Analysis

People are depending on the network to perform their daily activities more and more in the network information era.These activities will leave traces on the Internet,and these trace information can be analyzed and transformed to generate user portraits in order to obtain the privacy information of the target user.However,although user portrait technology has made great progress and has been widely used,these technologies are mainly used for interest and behavior analysis in the commercial field.The analysis of the user's intrinsic characteristics is rare,and the method of refining the intrinsic characteristics into emotional intelligence is even less,many technical gaps still exist in this field.The reason for this phenomenon is the lack of quantitative methods for the intrinsic characteristics of target users,which makes a large number of data containing target information be simply processed and discarded.The traditional psychological models are often able to portray the intrinsic characteristics of the target,but these models remain in the manual operation stage because of technical problems.This makes the use of psychological models to automatically quantify the various indicators of the target and for user portraits very few now,and there are some problems such as the unmatching between the parameters needed for model quantization and the existing data content.In order to portray the intrinsic characteristics of the target user,this paper combines information science with psychology,and proposes a method of privacy mining and user portrait for target users.Firstly,the method collects the trace information left by the target user on Android platform,and preprocesses the information,which is divided into two parts:data information and text information.Then,data statistics method is used for data information,and Chinese word segmentation and emotional analysis method is used for text information.In the emotional analysis,a method of calculating the emotional value of sentences is proposed.Finally,the data statistics are calculated to complete the user portraits of social relations and social behaviors,and the emotional value of sentences is brought into Baron's emotional intelligence model to complete the user portraits of emotional intelligence level.In the aspect of affective analysis,this paper proposes an affective analysis method based on dictionary and statistics,which is used to calculate the emotional value of sentences in text data.Firstly,emotional dictionary queries are used to classify the extreme value,intensity value and emotion of emotional words,and each numerical value is used to calculate the emotional value.Then,the similarity of unlisted words is calculated with the whole affective dictionary as the original word,and the words that exceed a certain threshold are judged as affective words and the affective values are calculated.Finally,all the emotional words are restored to all the sentences,the emotional values of the sentences are calculated,and the situation of adverbs or conjunctions is corrected according to grammatical rules.In addition,this paper evaluates the experimental performance of the method in two forms,using precision,recall,F-measure and result accuracy to evaluate its emotional classification ability and emotional value calculation ability,which verifies the feasibility,effectiveness and accuracy of the method.In the aspect of user portrait,this paper proposes a user portrait method based on privacy data,according to the user data collected from Android platform.This method generates user portraits of target user in social relations,social behavior and emotional intelligence through data statistics and model construction.The social relationship and social behavior portraits are based on the social data information of the target users,and construct the social relationship charts and active time scatter diagrams of the target users respectively.Emotional intelligence portrait is based on Baron's Emotional Intelligence Model.It constructs a model of user text data that has been completed Chinese word segmentation and emotional analysis.It quantifies the emotional intelligence ability by using the generating rules of each index,and generates the emotional intelligence value of the target.This method realizes the automatic quantification of emotional intelligence index for the first time.It evaluates the experimental performance and verifies the accuracy from three aspects:questionnaire,group distribution and scoring trend.It also analyses the selection of parameters and the results of emotional intelligence portrait,which proves that the method is correct,feasible and interpretable.