面向社会化媒体消费对象识别研究.docVIP

面向社会化媒体的消费对象识别研究( 付博1, 刘挺1+ 1(哈尔滨工业大学社会计算与信息检索研究中心,黑龙江 哈尔滨 150001) A Study of Consumption Object based in Social Media * FU Bo1+, LIU Ting1 1(Research Center for Social Computing and Information Retrieval, School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001) + Corresponding author: E-mail: tliu@ Abstract: Understanding consumption need and preferences behind a user’s activities can greatly personalize and thus recommend products. For example, business can decide whether to recommend commercial content based on user’s intent to purchase. In this paper, we focus our attention on capturing consumption object from microblogs and website, i.e., when a user posts a message, whether is containing buy wishes. A buy wishes is “a desire or hope for a product or service to purchase.” We called the product or service as a consumption object. We present the framework of building machine learning models to recognition consumption object. First, we leverage this resource to conduct the buy wish detectors for natural language text. A bootstrapping algorithm is proposed and empirically evaluated on the buy wish corpus. We show that the supervised SVM classifier using words features can achieve promising classification accuracy. Then, we use a novel monolingual word alignment method to recognize consumption object. We demonstrate the effectiveness of the proposed approach in consumption object detecting. Key words: consumption intention; consumption object; information extraction; social media. 摘 要: 在社会化电子商务的时代，用户经常在社交媒体网站与电子商务网站（如：新浪微博和一淘等）上发布具有消费意图的内容信息。然而，很少有研究将消费意图内容中的消费目标识别起来。本文通过分析用户在社交媒体和电子商务网站中发布的具有消费意图的文本内容，用以识别此用户会购买哪种类别的产品，即挖掘用户的消费对象(例如手机)。通过分析性能不同的特征集和迁移学习算法，利用检索的方式对用户的购买产品对象进行识别。基于用户的行为来理解用户的消费需求和喜好能够进行更好的个性化服务和产品推荐。实验证明，我们的方法是有效的。 关键词: 消费意图；消费对象；迁移学习；单语词对齐；信息抽取；社会化媒体。 目前越来越多的用户乐于在社会媒体和电子商务网站中，通过网络的方式发布自己的消费需求，包括已经购买的产品、未来需要购买的产品，以及隐藏在内心中喜爱的产品等。国外流行的社交媒体Twitter和国内的新浪微博已成为用户发表自己观点的社会化平台，每天大约有1亿的信息发表在新浪微博中。在这些大量的信息中，有些信息显示了用户希望去购买某些