Research On The Quality Of Robotic News Content Based On Text Similarity

Today,artificial intelligence has penetrated into all aspects of people's lives.Artificial Intelligence is about to replace human beings" are in prevalence.In such an AI age,the Journalism also innovated their own AI applications-"robot" journalism.Its innovation also set off vigorous discussions that whether News-bots would replace journalists or not.Hammond,technical director of Narrative Science,a US robot-writing company,predicted that 90 percent of the news would be written by computer in next 15 years.As to whether News-bots will replace journalists,most of the judgments in the current research come from the subjective judgment of the researcher,lack of objective and quantitative comprehensive research.This paper argues that the study of whether News-bots will be substitute for journalists,first of all should be based on the extent to which the News-bots can reach journalists' writing level.There are famous assumptions in the field of human-computer interaction:The higher degree of the computer anthropomorphism,the better the human evaluation of the computer.Based on the assumptions above,this paper puts forward its own basic assumptions:There is correlationship between robot and artificial news and audience's evaluation to robotic news content.This paper uses the concept of similarity in natural language processing to describe the similarity of the two.For the content quality,this article combined with the pre-test to establish content evaluation system on the basis of references.I collected Tencent Dreamwriter's a series of reports about gold medals in the 2016 Rio Olympic Games,further more,the other three major domestic Internet portal-Neteasy,Sina,Sohu's artificial news reports were taken as samplesAmong these samples,five diferentiated robotic news and five normal news were used as control group to accept double-blind evaluation.Based on this study's evaluation system,the robot and artificial news were evaluated in the form of questionnaire.Through a series of analyses,the data results verify all the hypotheses presented in this paper,that the similarity between robot and artificial news is positively correlated with its content quality.In the data difference,this paper also finds that the factors that affect the similarity between the robot and the artificial mainly focus on the objective factors such as objectivity,information richness,and information validity.However,emotional indicators such as fun,style of writing and other factors in different similarity of the robotic news did not show much difference.We can see the robotic news in these areas and the distance from the human.This paper also summarizes four improving measures for robotic news from the data results to provide references to the News-bots industry.