Intelligent Q&A Based On Beautiful Rural Knowledge Graph

The construction of beautiful rural areas is a key aspect of rural revitalization in China,and rural information is essential for the development of each village.Accurately querying rural information and intelligently answering user questions are critical for achieving the goals of building a "beautiful rural" and sustainable development.Intelligent Q&A is one of the most extensively researched applications of artificial intelligence.Currently,the query of beautiful rural information still relies on traditional search engines,which is difficult to meet users’ demands for accurate and intelligent information retrieval.This article aims to create intelligent Q&A of the knowledge graph of beautiful rural areas,and performs study on the following three issues:Construction of a beautiful rural data set and knowledge graph: At the moment,there is a lack of open beautiful rural data set,and the majority of its data originates from news articles and Baidu Baike.Therefore,it is crucial for study in this area to construction of beautiful rural data set and knowledge graph.This phase collects data from different sources using the Scrapy crawler and the Han LP tool,creates a knowledge system and ontology layer for beautiful rural regions,and finally imports the data to construct a relatively full knowledge graph of beautiful rural areas.The findings from this section will be used in future study.Deep Learning-Based Knowledge Graph Entity Recognition Model: Although deep learning methods have shown strong performance in entity recognition tasks,contemporary mainstream models rely on the size of annotated datasets,resulting in lower recognition accuracy in the field of sparse data.To address the above issues,this work proposes the BERT Bi LSTM-GCN-CRF model,which incorporates GCN into the existing model.On the one hand,it investigates the relationship between character and character,words and words,and on the other hand,it combines with the features extracted by Bi LSTM to improve understanding of Semantic information.Validation was conducted on both the public dataset CLUENER2020 and the Beautiful Rural dataset,and it outperformed other models in entity recognition tasks.Beautiful rural intelligent Q&A based on multi-level template matching: Currently,supervised learning approaches based on annotation data rely on enormous amounts of annotation data;nevertheless,rule-based templates frequently result in decreased accuracy due to template matching mistakes.The accuracy will fall further due to the sparsity of data in the field of beautiful rural areas and the complexity of entity attributes.As a result,an intelligent Q&A method based on multi-level template matching is proposed for beautiful rural areas.This method classifies questions by embedding knowledge graphs and generates similar questions,thereby reducing the number of template construction and the probability of direct matching errors,effectively increasing the accuracy of template matching question answering.The experimental results reveal that when compared to methods based on BERT annotated data and Bayesian classification templates,this method improves accuracy by 10.78% and 4.79%,respectively.Finally,a beautiful rural intelligent Q&A system was created,which helps users to deploy the application by allowing them to more quickly and accurately query relevant information.