CAJ | 학술논문

随着电子商务的发展，推荐系统的开发成为各大互联网商店的一个重要部分。推荐系统中的用户购买相似度推荐算法被普遍应用，该类算法包含两种类型：基于邻居模型的算法和潜在因素算法。但是，随着网上商店的增多，淘宝等大型网上电子商城的兴起，传统的用户购买相似度推荐算法不能有效地处理互联网爆炸式的大数据信息，快速完成推荐系统的推荐工作。针对用户购买商品后的潜在因素，提出了基于MapReduce计算框架的分布式潜在因素算法，该算法能够有效地完成推荐系统的推荐工作，实现系统的实时性。实验结果表明，本文提出的算法具有很高的执行效率以及低误差率。
수착전자상무적발전，추천계통적개발성위각대호련망상점적일개중요부분。추천계통중적용호구매상사도추천산법피보편응용，해류산법포함량충류형：기우린거모형적산법화잠재인소산법。단시，수착망상상점적증다，도보등대형망상전자상성적흥기，전통적용호구매상사도추천산법불능유효지처리호련망폭작식적대수거신식，쾌속완성추천계통적추천공작。침대용호구매상품후적잠재인소，제출료기우MapReduce계산광가적분포식잠재인소산법，해산법능구유효지완성추천계통적추천공작，실현계통적실시성。실험결과표명，본문제출적산법구유흔고적집행효솔이급저오차솔。
With the development of electronic commerce, recommended the development of the system become an important part of each big Internet shop. Recommendation system of recommended users to buy similarity algorithm is widely used, the algorithms include two types: based on the neighbor algorithm and the potential factors of the model. But, along with the increase in online store, the rise of taobao and other large online e-commerce, traditional users buy similarity recommendation algorithm cannot effectively deal with the Internet explosion in big data information, recommendation system user recommendation work done quickly. In view of the potential factors after the user to purchase goods, this paper puts forward a framework of distributed potential factors were calculated based on graphs algorithm, this algorithm can effectively work of recommendation system, to realize the real-time system. Through the experiment results show that the proposed algorithm has high execution efficiency, and low error rate.