CAJ | 학술논문

近年来兴起的兴趣社交网络中因其内在的商业价值引起了人们的注意。兴趣社交网络中涌现出大量的商品供需信息,带来的问题之一就是商品供需信息传播匹配效率低下。为解决上述问题,本文依托社会网络分析方法,探讨在具备小世界特性的兴趣社交网络中商品供需匹配的特点,研究商品传播和匹配的规律和提高匹配效率的策略。本文选取豆瓣网关注关系为实证数据,提出在兴趣社交网络中友邻信息相关度与供需信息传播匹配效率正相关的假设,并采用计算机仿真方法来验证假设。论文获取豆瓣网的社交网络数据、包括网络用户藏书数据和图书的标签数据,针对随机选取的1000个用户和200本图书作为实验样本,分别设计两套友邻信息相关度度量算法和两套兴趣社交网络中商品供需信息传播匹配算法,采用进行仿真模拟实验进行了比较。实验结果证实了人们对于商品检索的直观判断,得出在兴趣社交网络中,商品供需信息传播匹配效率与友邻信息相关度正相关的结论。
근년래흥기적흥취사교망락중인기내재적상업개치인기료인문적주의。흥취사교망락중용현출대량적상품공수신식,대래적문제지일취시상품공수신식전파필배효솔저하。위해결상술문제,본문의탁사회망락분석방법,탐토재구비소세계특성적흥취사교망락중상품공수필배적특점,연구상품전파화필배적규률화제고필배효솔적책략。본문선취두판망관주관계위실증수거,제출재흥취사교망락중우린신식상관도여공수신식전파필배효솔정상관적가설,병채용계산궤방진방법래험증가설。논문획취두판망적사교망락수거、포괄망락용호장서수거화도서적표첨수거,침대수궤선취적1000개용호화200본도서작위실험양본,분별설계량투우린신식상관도도량산법화량투흥취사교망락중상품공수신식전파필배산법,채용진행방진모의실험진행료비교。실험결과증실료인문대우상품검색적직관판단,득출재흥취사교망락중,상품공수신식전파필배효솔여우린신식상관도정상관적결론。
Building trust on the Internet has become a challenging issue in the E-commerce era. In addition, the increase of unstructured and overloaded information creates inefficient dissemination of supply-demand information. Trust issue may be tackled by online social networks （e.g. Faeebook） based on emotional connections. However, emotion-based societies do not support commercial atmosphere. Faekbook also fails to implement its F-commerce strategy. Under these circumstances, Interest-based social network may be able to solve the abovementioned problems. When searching for a desired product, does an individual often consult friends about the characteristics of products that they are interested in buying？ The paper tries to answer the question. Traditional social network researches, mostly based on empirical surveys, focus on structural issues such as strong or weak tie, and neglect the relationship between individual characteristics and need. Small World theories have enriched researches on information dissemination in social network based on local information hypothesis. However, these theories are proposed based on heuristic clues and nodes tend to have inclinations without considering real world situations. This paper tries to answer the above questions by employing social network analysis and computer simulation based on real world Interest-based social network data. Hypothesis is proposed that need information dissemination efficiency is positively related to neighbor＇s characteristics in relation to the need information. These hypotheses represent steps required to reach the supply in the network and individual＇s bookmarks. A piece of need （a book enquiry） is being relayed via linked nodes （individuals） to the matching one in the social network. If the path of high-relevant nodes statistically requires fewer steps to reach the supply node than that of low-relevant ones, the relevance can be confirmed under repeated experiments. With Douban.com chosen as an empirical object, a sub-graph of the follower＇s network, including book collections of users together with tags, is retrieved. A number of indicators of the networks has proven its characteristics of small-world and scale-free. For comparison, a set of top 100 most collected books and a set of 100 systematically sampled books are retrieved as need information. We designed two models to measure the relationship between neighbor＇s interest and demand information. A total of 1000 starting nodes are randomly chosen, based on which a high-relevant and a low-relevant path algorithm relay the need information. The simulation experiment generates four groups of data and the results confirm the proposed hypothesis. Thus, we conclude that the efficiency of matching supply-demand information in online interest-based social network is highly related to neighbor＇s interest in relation to the relayed need information. This paper confirms the well-known yet implicit experience based on scientifically empirical study. Based on real world Interest-based social network data, computer simulation is employed in the study. Moreover, the method used to test the proposed hypotheses is adapted from the Portfolio Analysis method. The statistical method compares the required steps to reach supply node between path of high-relevant nodes and path of low-relevant ones. This study provides a new methodology to study information dissemination in social networks.