从评价网络提升信任启动的准确度

Promoting accuracy of trust bootstrapping from rating network

为了减少电子商务平台中商品只有很少用户评价时评估商品信任值容易受到不公正、恶意评价的影响,提出一种基于评价网络评估评价可信度的信任启动方法。评价的可信度通过评估评价者对其他商品的评价得出,与评价者评价的数量、金额以及被评价商品的价格等因素有关。无评价商品的信任值来源于商品所在的商铺和商品的申明属性。当商品的评价具有足够高的可信度时,它的信任值就由足够可信的评价决定,否则部分由评价决定或按无评价的商品处理。计算、分析和测试结果表明提出的从评价网络评估评价信任度的方法同常规方法、k均值聚类方法相比产生的误差小而且对恶意评价比率不敏感,能有助于用户从电子商务平台选择值得信赖的上架初期的商品。

To reduce the influence of evaluating trust of a commodity may be affected easily by unfair and malicious rates when the commodity has only few rates on e-commerce platform, a trust bootstrapping method based on assessing the credibility of the rate was presented. The credibility of a rate was got through evaluating the rates for other commodities and related to the factors of the number of rates by the rater, the rater＇s transaction amount and the price of the rated commodity. The trust value of a commodity without a rate was derived from the shop to which this commodity belonged and the declared attributes of this commodity. The trust value of a commodity which owned rates with sufficient high credibility was determined by these rates with high credibility. Otherwise the trust value was determined partly by rates or was processed according to a commodity without a rate. Calculation, analysis and experimental results show that this presented method, evaluating the credibility of a rate by its rating network, compared with the conventional method and k-means clustering method, has the smallest error and is not sensitive to the ratio of malicious rates. This method can help users select reliable commodities sold at the initial stage on e-commerce platforms.