| Robot journalism is a product of technological development in the field of journalism and communication along with big data and AI.It is a technology application after radio,television,and the Internet,which has a huge impact on the news industry.The emergence of robot journalism has not only enriched ways of news reports,but also caused structural changes in the traditional news production.The guard of news transmission have switched from human editors to AI algorithms.The speed of news compilation and distribution has gradually decreased from a few hours in the past to a few seconds now.The process of news production has gradually become more and more automatic and intelligent.As a completely new model of news production,robot journalism is quite different from traditional news such as newspapers,television,and even We Media news in the Era of Internet.From the establishment of robot algorithm models to the generation,compilation and distribution of news,robot journalism has its own automatic model of news production.The robot news production mode of National Business Daily is mainly composed of data extraction model and content generation model,which model of news production includes several important parts,such as database establishment,machine learning,content generation,review,compilation and distribution,and finally generates the widely used AI Express today.Currently,robot journalism is widely applied to real-time information news such as finance,sports,geological disasters,weather forecasts and so on.This study deeply researches the model of robot journalism production based on the Xiaoice writing robot,which is mainly used in the National Business Daily financial express.This article studyed an artificial intelligence(AI)robot as a journalism writer on the National Business Daily,who called Xiaoice.To know about theXiaoice,this article conducted an in-depth interview with the person in charge of Xiaoice's financial products and editors in the National Business Daily,and base on Tencent Dreamwriter's news production process,the Xiaoice writing robot production process was divided into three parts,which included database establishment,machine learning,writing and review.The detailed study was conducted to the three parts.Xiaoice released much news on the National Business Daily online every day.This article adopted the systematic sampling method to select the financial news released by Xiaoice on February 21,22 and 23,2022 as samples,and these samples were analyzed from several dimensions,which include news release time,news content category,news content complexity and so on.In addition,this paper fully studies the news production mode of Xiaoice and the characteristics of financial news written on the National Business Daily online.The study found that although Xiaoice releases the financial newsletter faster and more efficiently compared to human editors,it also has problems with poor logic,poor flexibility,and poor ability to summarize and analyze.Firstly,this paper proposed that the robot journalism should refer to the Chinese standard to further strengthen the AI model training of the writing robot in the model training stage,aiming at the problems of weak application of symbols and poor logic in robot news.Secondly,It is proposed that in the model training stage more writing templates should be added to strengthen the model for the weaknesses that are now exposed.Thirdly,in the term of the problem that AI Express cannot have both efficient editing and high accuracy,it is proposed that AI review should be added to share the workload of human editors and improve the review efficiency.Finally,aiming at the human-machine cooperation mode,it is proposed that the ability of writing robots to process structured data should be fully tapped to make up for the editors'deficiency in the data collation,information cross-validation and other parts in news,and it is also helpful to explore a new model of human-machine cooperation.In addition,this article took the gatekeeper theory as an example,and it was pointed out that the gatekeeper in robot news is further pre-positioned and lack of experience in the content of robot news,therefore robot news needs a set of editorial review mechanisms suitable for it,and the impact of robot news on current communication theroy.The innovation of this article lies in the combination of in-depth interview method and content analysis method.By interviewing the R?D person in charge of Xiaoice project on the R?D side,and the editor who works with Xiaoice on the application side,it systematically analyzed the news production mode of Xiaoice on the " National Business Daily",which has compiled and distributed more than 500 AI newsletters in the past two years,is deeply applied.The news production mode strengthens model training,enriches writing templates in a targeted manner,adds AI auditing,and explores a new mode of human-machine cooperation should be improved and explored.With the development of AI still in its infancy,the application of robot journalism has just begun.The establishment and improvement of the production model of robot journalism is also an issue that requires long-term attention. |
PDF Full Text Request