Research on Global Entity Relation Extraction in Music

With the rapid growth of information on the Internet , information extraction technology is increasingly concerned about the entity relationship extraction is very important subtasks in these tasks . The study found that due to the traditional entity-relationship extraction technology for the sentence level , only in one sentence extraction , which lost a lot of relationships . This paper presents a global entity relationship extraction, relation extraction and classification of any two entities in the chapter . Start from the field of music , through detailed statistics and analysis , entity relationship extraction will be the impact of the relationship between entities , such as the equivalence relation between the entity and the parallel relationship . By equivalence relationship between the entity and the non - equivalence relations fusion and simple reasoning can achieve global entity relation extraction . Firstly, using the rule-based and use of the dependency syntactic and the combination of both a variety of methods to identify all representations of the entities of the field of music ; rule-based and based on this , various methods based on the binary classification and a combination of both . the field of music , refers digestion study ; Secondly, the the convolution tree trees nuclear and mixed nuclear entity relation extraction in the field of music ; Finally, the relationship between the effective integration and simple reasoning , global entity relation extraction . The evaluation results show that our global entity relation extraction than traditional entity-relationship extraction F value increased by 13.8 % , and our technology can be applied to any field . Finally, we also designed and implemented the the coreference resolution experimental application platform and text mining technology integration platform , to provide better services for the areas of research and application of natural language processing .