Research on key technologies in constructing affective disorder knowledge graph
The key technologies for constructing a knowledge graph of affective disorder mainly include multi-source data corpus construction,information extraction,knowledge fusion,knowledge graph storage and visualization,etc.Firstly,multiple sources of data are used to integrate and construct a corpus of knowledge on affective disorder.After data processing,a normalized corpus,Normal,is obtained.Secondly,a mixed Maximum Entropy Markov Model-Convolutional Neural Network(MEMM-CNN)information extraction method was proposed in the information extraction stage.The corpus dataset was first trained by MEMM,and the model was further trained by CNN.Repeat the iteration to complete the three task units in information extraction and obtain the final triplet set.Then,in the knowledge fusion stage,an adaptive weighted estimation knowledge fusion algorithm based on ρ-value(SAW)was designed,which calculates the triplet credibility through the credibility function and compares it with the credibility threshold to complete triplet polarity recognition and output the trusted database triplet set.Compared with traditional data fusion algorithms such as VOTE and ACCU,this algorithm can significantly improve the recognition quantity and polarity recognition accuracy of triples in the trusted database.At the same time,it has advantages such as good noise resistance and short response time.Finally,use the Neo4j graphical database to dynamically generate a knowledge graph of affective disorder.
万方数据
维普
