Design And Implementation Of Chinese Metaphorical Labeling Corpus System

Metaphor is used to describe and understand abstract concepts,it is not only a linguistic phenomenon,but also a cognitive way.Metaphor is ubiquitous in human language.In recent years,with the rapid rise of social media such as Twitter,Weibo and forums,metaphorical texts appear on more diversified platforms and get more and more attention.In this case,people urgently need to use natural language technology to process massive metaphorical information.Therefore,the study of metaphor has been widely concerned by scholars and has gradually become an important research direction in the field of natural language processing.Metaphorical semantic resources are the basis of metaphorical computing.Only with the support of large-scale and high-quality semantic resources can metaphorical research be carried out more deeply and extensively.Although the demand is urgent,the resources of metaphor corpus at home and abroad are relatively scarce at present.The overall scale is small,and the quality is questionable.Therefore,few annotation information is provided for computer processing,especially the semantic resources of Chinese metaphor are extremely scarce.Therefore,based on practical problems and requirements,this paper designs and implements a Chinese metaphorical labeling corpus system.This paper mainly includes the design and implementation of metaphor recognition algorithm and system.The metaphor recognition algorithms mainly consist of two algorithmic frameworks: rule-based and semantic-based,which are metaphor recognition from different perspectives.The system is mainly composed of five modules,including crowdsourcing platform module,metaphor recognition module,information analysis module,resource download module and personal center module.The crowdsourcing platform module is used to provide services for users to manually annotate data.The metaphor recognition module is used to provide the service of the user's metaphor recognition.The information analysis module provides users with the service of corpus internal information analysis.the resource download module provides users with the service of corpus download.And the personal center module is used to provide user information management services.The algorithm is mainly built with Keras framework,and the language model is pre-trained with Word2 vec and BERT.The system backend is implemented using Django framework and Python language.The front end uses Echarts to present the dataset analysis results.The database uses MongoDB to store file data easily.Finally,through the system in this paper,users can complete such task as metaphor recognition,corpus tagging and corpus downloading.This paper can meet the practical needs of researchers in the construction of metaphorical corpus and has certain practical use value.