Research On The Question Answering Model And Application Of Artificial Intelligence Oriented To Intelligent Logistics

Logistics is an important infrastructure for national economic development.After more than 30 years of development,the logistics industry has become a pillar industry of the national economy and an important modern service industry.It is also one of the fastest growing industries.However,as China's demographic dividend gradually disappears,labor costs continue to rise,and the shortage of labor forces the traditional logistics industry to gradually transform into an unmanned and intelligent direction.The logistics industry's demand for intelligent logistics robots is increasingly urgent.In recent years,AI technology has been widely used in logistics scenarios.In terms of software,intelligent voice assistant robots and logistics customer service robots have begun to be used in daily life.The logistics company embeds the intelligent question answering system in its We Chat applet to provide users with real-time online consulting services.However,there are some problems in use,such as the system is unresponsive,the answer is not what you ask,only supports a single round of QA;and the response effect for slightly complex questions is not satisfactory.There are two main reasons for these,the one is the inadequate system design,and another is that the efficiency of information retrieval needs to be improved.This thesis mainly answers three questions.What kind of artificial intelligence QA system is needed in the field of logistics? How to improve the accuracy of information retrieval through key algorithm design? How to optimize the logistics QA system from the aspect of system design?First of all,the research status and hot spots in the fields of intelligent logistics and intelligent question-and-answer are analyzed by using bibliometric method and scientific knowledge mapping method to provide evidence for the starting point of this research.Then,based on the relevant theories of question-and-answer system,combining the logistics service process and logistics question-and-answer situation,the operation mechanism of question-and-answer system in logistics field is explained.On this basis,the design idea of AI Q&A mode for intelligent logistics field is further put forward.Then,the sentence similarity algorithm in the AI QA mode is optimized.The idea is to add part of speech and word order factors to the baseline SIF model.The improved algorithm is further verified by three indexes of accuracy rate,recall rate and F value.It is found that the SIF algorithm is much better than the TF-IDF algorithm based on statistics and the algorithm based on word order by 18% and 29%,and the improved SIF algorithm in this study has increased by 4.5% in accuracy rate.At last,the paper puts forward the optimization suggestions of AI QA mode from "soft" and "hard" aspects,and on this basis,discusses the theoretical and practical value of the research conclusion,and summarizes the deficiencies in the research.