Application And Research On Entity And Relation Extraction For Psychological Counseling Text

With the rapid development of the Internet,online psychological counseling services are more and more popular,and the data generated is increasing day by day.These data contain a large number of patient event information,the extraction and sorting of these event information can improve the work efficiency of psychological counselors.At present,Entity and Relation Extraction is effective in various fields of information extraction tasks,but it’s application and research in the field of psychological counseling are limited.Therefore,this paper studies information extraction of Psychological counseling text using Entity and Relation Extraction algorithm.The Entity and Relation Extraction algorithm Cas RelBiLSTM-Biaffine is proposed to extract patient event information from psychological counseling texts.The main work of this paper is as follows:(1)Psychological counseling data set construction.Conventional entity relationship extraction is mainly aimed at entities such as names,place names and their semantic relationships to complete entity relationship triplet extraction,but psychological counseling text is the daily language of patients,the structure is complex,entity type is not clear.To solve this problem,this paper proposes to take event information as entity and event logical relation as triplet form of relation.Through this form to complete the data annotation work,finally get the event as the entity of psychological counseling data set Ev RTD.(2)Entity and Relation Extraction algorithm.An Entity and Relation Join Extraction model Cas Rel-BiLSTM-Biaffine is proposed.This model is based on Cas Rel model of cascade pointer network.In Cas Rel coding layer,pretraining model Ro BERTa is replaced and BiLSTM is added to enhance context feature representation.Biaffine mechanism is introduced to capture entity boundary in entity recognition layer.The problems of long entity recognition and triplet overlap in psychological counseling text are solved by the above changes.Through experimental comparison,Casrel-Bilstm-Biaffine model performs better than other comparison models in Ev RTD data set,and F1 value of Cas Rel is increased by 3.0% compared with that of the basic model.In SEO and multi-relationship experiments,F1 value increased by 3.8% and 3.1% respectively.To prove the validity of Cas Rel-BiLSTM-Biaffine model in extracting event information from psychological counseling texts.(3)Design and implementation of information extraction system for psychological counseling events.The Cas Rel-BiLSTM-Biaffine model is used to construct the information extraction system of psychological counseling events,process the psychological counseling texts of patients,extract the event information in the texts and store it as the patient event set,so as to assist the work of psychological counselors.In order to achieve this goal,this paper uses the relevant system construction technology,combined with the actual application requirements,to complete the construction of the system.