论文部分内容阅读
【目的】对商品检索中的购物任务进行识别,并对多任务会话行为特征进行分析。【方法】利用淘宝商品分类体系以及自建的商品词表,根据商品检索的检索式进行购物任务识别,数据集为2 754个用户的19 704个检索会话。【结果】影响每个购物任务所用检索式数的因素包括商品分面、数量的多少以及描述难易程度;有主要任务和次要任务之分的多任务会话中,任务之间的关系更为紧密。【局限】购物任务识别方法有待完善,只以检索式作为研究对象无法全面反映用户行为特征。【结论】本研究可以帮助理解购物中的商品检索行为,并为设计更好的商品推荐算法、预测用户购物过程、行为等提供依据。
【Objective】 To identify shopping tasks in product retrieval and to analyze the behavioral characteristics of multitasking sessions. 【Method】 Using the Taobao product classification system and self-built product vocabularies, shopping task identification was conducted according to the retrieval formula of commodity search. The dataset was 19 704 retrieval sessions with 2 754 users. 【Result】 The factors influencing the number of search queries used in each shopping task include the number of facets, the number of products and the ease of description. Among the multitasking sessions that have primary tasks and secondary tasks, the relationships among tasks are more close. [Limitations] shopping task identification method to be perfected, only the search as a research object can not fully reflect the user behavior characteristics. 【Conclusion】 This research can help to understand the behavior of product retrieval in shopping and provide the basis for designing better product recommendation algorithm and forecasting users’ shopping process and behavior.