数字消费市场政策设计中的关键行为问题

Key Behavioral Issues in Digital Consumption Market Policy Design

当前我国数字消费市场蓬勃发展,网络视频、网络直播、网络游戏、网络文学等新业态产生了海量真实行为数据,为推进相关政策设计的前瞻性干预研究创造了条件,但仍存在以下关键行为问题亟待解决:一是行为数据通常是多模态的,现有挖掘方法存在“黑箱”问题,给挖掘规律解读造成困难;二是感性因素的早期表现不够显著,难以保障早期因果推断的可靠性;三是早期干预对象不易确定,柔性措施干预力度不易把控,干预效果不易评估。为推动该方向研究,本文建议进一步丰富多模态数据挖掘所需样本库和训练集,促进形成经验性数据挖掘与规范性经济分析相互补的研究范式,完善干预效果评价体系,以逐步激活真实行为数据在行为源头干预方面的研究潜能。

The rapid expansion of China's digital consumption market has led to the proliferation of realworld behavioral data across diverse emerging sectors,such as online video,live streaming,online gaming,and internet literature.This presents an opportunity for forward-looking intervention research in policy design.However,several key behavioral issues still warrant attention.Firstly,behavioral data often exhibit multi-modal characteristics,and existing mining methods face challenges related to interpretability,commonly known as the“black box”problem,which may impede effective discernment of underlying patterns in the data.Secondly,early indications of emotional factors may not be overtly evident,leading to potential difficulties in ensuring the reliability of initial causal inferences.Thirdly,identifying appropriate targets for early intervention and managing the impact of flexible measures pose complexities in accurately evaluating intervention effects.To advance research in this field,this paper proposes policies to enrich sample libraries and training sets specifically tailored for multi-modal data mining,advocates a research paradigm that integrates empirical data mining with normative economic analysis,and enhances the assessment system for measuring intervention effectiveness.These endeavors will effectively harness the potential of real-world behavioral data for early-stage interventions,offering valuable insights for policy design.