| Alzheimer's disease is a common senile disease,it had reported that have caused 1.9 million deaths in 2015.Although the cost of treatment for Alzheimer's Disease is pretty high,there have been no related medicine can cure or even ease the condition until now.In order to find effective treatments,a large number of related experiments are ongoing,so a large number of papers are constantly emerging.However,the New England Journal announced last year that clinical trials of the most likely specific drug failed.Therefore,it is necessary to use the knowledge graph method to sort out past research ideas and to discover new knowledge and research directions.In this paper,a total of 132,749 abstracts were processed and analyzed in the experiment to construct a knowledge graph of Alzheimer's disease,which from 2000 to 2019 in the PubMed database.During the construction process,we firstly download data from the PubMed website,and then extract the data is according to the format.After that,we use the topic model,word vector model,clustering model,and named entity recognition to process the data in turn,and get the knowledge of entities,relationships and attributes.Finally,the graph database is to construct the extracted knowledge graph.We also take the time dimension into consideration,and then performing related knowledge discovery according to the change of the entity in the time dimension.There were 775 entities were identified which belong to 16 types.Based on the cooccurrence of entities in the literature,six relationships of "treatment","cause","complication","diagnosis" and "research geography" were defined in the other hand,which total number is 664.At the same time,we conducted entity trend research from five aspects of diseases,drugs,symptoms and processes,research areas,and genes.Finally,a knowledge graph of Alzheimer's disease was constructed successfully,which can provide direct or indirect theoretical basis and data support for related researchers' research. |
PDF Full Text Request