Research On Metaphor And Exaggeration Identification And Element Extraction For Appreciation Question In College Entrance Examination

Regarding the Chinese college entrance examination,reading comprehension articles can be divided into multiple genres,such as poetry,prose,etc.This article is based on the rhetoric of the college entrance examination prose reading comprehension of the national key research and development plan project as the research background.Among them,rhetoric is a method of modifying and decorative language.By using some language skills,the language makes the language more expressive,infected,infected Power and attractive,and deeper and more vividly shows the author’s thoughts and emotions.The metaphor and exaggerated rhetoric often appear in the college entrance examination questions.For this reason,this article mainly researches metaphors and exaggerated quit and has developed a system of answering systems based on metaphors and exaggerated rhetoric appreciation.The main research work of this article is as follows:(1)Analysis of prose and appreciation and construction of related datasetsThrough the summary and analysis of reading comprehension questions in the Chinese college entrance examination Chinese test paper in the past ten years,it is found that the metaphor and exaggerated rhetoric of prose and genre of prose genres have more inspections,and when answering.The role of elements and rhetoric is the key to answering the question.Therefore,the components and definitions of metaphorical and exaggerated sentences are analyzed,comparing the differences in the semantic level of the two related elements.Organize 4918 data sets of prose metaphors.Based on this,mark the target and source domains in the metaphor.Due to the less research on exaggerated words,3845 exaggerated data sets were constructed,which counted the 1830 physical words in the exaggerated sentence,and marked the physical word description information.Establishing these data sets can provide strong data support for exploring and solving subsequent issues.(2)Metaphorical sentence recognition and element extractionFor metaphorical sentence recognition and metaphorical element extraction tasks,most of them are trained by single tasks.The semantic characteristics are not fully expressed,and errors will accumulate.To this end,this article uses a joint learning model and,at the same time,optimizes the loss function to identify the metaphorical sentences and extract the elements.Because most of the target domains and source domains in the metaphorical sentence are nouns,the word characteristics are merged into the vector,enhancing the sentence’s vector.It is compared with the existing model on the prose metaphorical data set.The model of this article has achieved good performance.(3)Exaggerated sentence recognition and element extractionExaggeration is also a way of language expression.It emphasizes certain perspectives,emotions,or facts by exaggerating or narrowing the real situation.From the standpoint of metaphors,this article studies metaphor exaggeration and uses the entity in exaggerated sentences as a breakthrough.Because encoding the pre-trained model tends to capture general text representations,it needs more field knowledge.In order to make up for the problem of insufficient information in related fields,prose entity description information was added to the encoding layer for enhancement.In addition,the constructed knowledge base enhances prose entity coding.Similarly,multi-task learning is used to associate the exaggerated sentence recognition and factor drawing tasks and add the entity description information to be represented by rich semantics.Using this method to compare experiments with other methods on the exaggerated dataset,the method proposed in this article is effective.(4)Answering system based on metaphor and exaggerated rhetoric appreciation.By using the metaphorical and exaggerated sentence recognition and element extraction methods proposed in this article,combined with the characteristics of the college entrance examination prose reading understanding the appreciation of the appreciation,the answering system based on metaphor and exaggerated resignation appreciation.According to the input articles and appreciation issues,the system locates the sentences to be appreciated and outputs the rhetoric identification results and the key element entity.Design metaphor and exaggerated rhetoric answering templates combine the identification results,key elements,and answer templates to adjust the final answer.After comparison and judgment with the reference answer,you can get a good score.