基于知识增强的企业技能智能问答应用研究

Research on Application of Enterprise Skill Intelligent Question Answering Based on Knowledge Enhancement

为解决公司员工对公司专业技术知识进行频繁查询而无法得到及时响应和准确答案的问题,本文以知识图谱结合神经网络为基础来进行知识增强,构建出一种针对公司特定知识领域的智能问答模型。主要利用双向LSTM神经网络将复杂多变且不标准的自然语言查询语句转换为公司专业知识库中对应的标准问题以解决查询不标准情况。然后结合由公司专业知识库中的问题、相似问题和答案等实体构建的知识图谱使得返回答案比较准确。通过相关实验,本文模型准确率达到83.44%,能及时响应和准确回答用户查询,节省了大量资源。

In order to solve the problem that employees of the company frequently inquire about the company’s professional technical knowledge and cannot get timely responses and accurate answers,this paper uses knowledge graphs combined with neural networks to enhance knowledge and build an intelligent question answering model for the company’s specific knowledge areas.The bidirectional LSTM neural network is mainly used to convert complex and non-standard natural language query sentences into corresponding standard questions in the company’s professional knowledge base to solve non-standard queries.Then combined with the knowledge graph constructed by entities such as questions,similar questions and answers in the company’s professional knowledge base,the returned answers are more accurate.Through relevant experiments,the accuracy of the model in this paper reaches 83.44%,which can respond to user queries in a timely and accurate manner,saving a lot of resources.