A Method Of Scoring Aggregation Based On Social Choice Theory

Evaluations are widely present in our lives,and all processes involving value judgments produce various forms of evaluation.The result of evaluation is a medium for the evaluator to convey its own value judgment to the public.With the rise of the Internet,the evaluation method based on online scoring has been favored by the public,not only its number is exploding,but also the public relies on the scoring information more and more.However,the dramatic increase in the online scoring has gradually revealed two problems: First,due to differences in the evaluator's cognition,each person's understanding of the scoring criteria is different,and the information transmitted by the same score may not be the same,so the scores given by different evaluators can not be directly compared.The traditional way,which directly put the results of adding or averaging all the scores as the final evaluation results,is likely to convey misleading information to the public.Secondly,all the scores are treated equally and aggregated or averaged directly in the present evaluation system without searching the source behind these scores,so that it is difficult to resist the occurrence of malicious scores,and the final evaluation results are easy to manipulate.In order to solve the two problems mentioned above,this paper proposes a scoring aggregation method based on social choice theory.The purpose of social choice is to aggregate the preferences of social members into social total preferences,and the purpose of evaluation is to aggregate the scores of the evaluators into the overall evaluation results.So by changing the scores into the order of preference,the evaluation becomes an issue of social choice.Since the score is converted into a preference,the score value no longer has absolute meaning,and the problem that scores by different evaluators are incomparable is solved.In order to make the social choice function suitable for general evaluation problems,this paper has made corresponding improvements on the basis of the original Schulze method.Firstly,the Borda score is introduced to reduce the requirements for the integrity of the score data,so that it can be widely applied.Then the average score is added to make an attempt to combine the cardinality and ordinality of the score.We also approved that this method satisfies the transitivity,Pareto and neutrality,and does not satisfy the majority criteria,which makes it possible that the evaluation results no longer can be directly manipulated by changing individual scores,and the phenomenon of brushing and scoring will be somehow alleviated.Finally,the results of two examples show that not only the proposed method can apply well under missing scores,but also the results of the evaluation can reflect the preference and willingness of the scorers more accurately.