Research And Implementation Of Automatic Question Answering System For Financial Field

In recent years,"Fintech-Financial Technology" has become a hot spot in the financial field.Enabling technology to empower finance has become a new development trend in the industry.Automatic Question-Answering(QA)is one of the typical representatives.Thanks to the advancement of technologies about knowledge graph,the Knowledge Based Question Answering(KBQA)stands out and has become a research hotspot in recent years.KBQA uses the knowledge graph as the data source,accepts the user's natural question as input,understands the user's inquiry intention,searches for the knowledge information in the knowledge graph,and returns the corresponding result.KBQA is more in line with the user's behavioral habits,and users can directly obtain high-precision answers through KBQA method,which greatly improves the user's information acquisition efficiency and experience.However,the implementation of the current KBQA system for the financial field still faces some problems:(1)There is a lack of feasible KBQA methods.The method of relying only on the deep learning model is time-consuming and unstable to meet the needs of online business;(2)The lack of an open knowledge graph in the financial field makes it difficult to establish a knowledge system.To solve above problems,this paper designs and implements a multi-level decision-making automatic question answering method,and uses the company's industry information as the entry point of the financial domain knowledge to construct the company's industry information knowledge graph,and finally realize the automatic question answering system for the financial domain.The specific content includes the following three points:(1)Research and implement a multi-level decision-making automatic knowledge based question answering method.This article first improves on the automatic question answering IE-SQL model,strengthens the semantic network characteristics of the model structure,strengthens the relationship between the model's multi-tasks,and improves the model's expressive ability;this article combines the traditional automatic question answering method with the improved IE-SQL model to realize the automatic question answering method of multi-level decision-making,and effectively improve the calculation efficiency of the method;(2)Construct the company's industrial information knowledge graph.This paper uses data crawling,entity recognition,relationship extraction and other methods to realize the collection of company industry data,and uses the traditional probability model method to realize the knowledge fusion between data.Then this paper constructs the company's industry information knowledge graph,and uses this knowledge graph to experimentally verify the multi-level decision-making method;(3)Build an automatic question answering system for the financial field.This paper uses the established knowledge graph as the data foundation,with the multi-level decision-making method as the core function,designs and implements an automatic question answering system for the financial field,and provides users with automatic question answering service.