Dynamic Consumer Profiling and Tiered Pricing Using Software Agents

Shopbots or software agents that enable comparison shopping of items from different online sellers have become popular for quick and easy shopping among online buyers. Rapid searches and price comparison by shopbots have motivated sellers to use software agents called pricebots to adjust their prices dynamically so that they can maintain a competitive edge in the market. Existing pricebots charge the same price for an item from all of their customers. Online consumers differ in their purchasing preferences and, therefore, a seller's profit can be increased by charging two different prices for the same good from price-insensitive and price-sensitive consumers. In this paper, we present an algorithm that partitions the buyer population into different segments depending on the buyers' purchase criteria and then charges a different price for each segment. Simulation results of our tiered pricing algorithm indicate that sellers' profits are improved by charging different prices to buyers with different purchase criteria. Price wars between sellers that cause regular price fluctuations in the market, are also prevented when all the sellers in the market use a tiered pricing strategy.     

Property-based collaborative filtering for health-aware recommender systems

Recommender systems aim at solving the problem of information overload by selecting items (commercial products, educational assets, TV programs, etc.) that match the consumers’ interests and preferences. Recently, there have been approaches to drive the recommendations by the information stored in electronic health records, for which the traditional strategies applied in online shopping, e-learning, entertainment and other areas have several pitfalls. This paper addresses those problems by introducing a new filtering strategy, centered on the properties that characterize the items and the users. Preliminary experiments with real users have proved that this approach outperforms previous ones in terms of consumers’ satisfaction with the recommended items. The benefits are especially apparent among people with specific health concerns.     

Recommendation technology: will it boost e-commerce?

Recommendation engines are becoming a critical part of many e-commerce sites. The approach uses complex algorithms to analyze large volumes of data and determine what products that potential consumers might want to buy based on their stated preferences, online shopping choices, and the purchases of people with similar tastes or demographics. Recommendation technology must also be able to reach out to small and medium-size businesses and be robust and cost-effective.     

Collaborative filtering and deep learning based recommendation system for cold start items

Recommender system is a specific type of intelligent systems, which exploits historical user ratings on items and/or auxiliary information to make recommendations on items to the users. It plays a critical role in a wide range of online shopping, e-commercial services and social networking applications. Collaborative filtering (CF) is the most popular approaches used for recommender systems, but it suffers from complete cold start (CCS) problem where no rating record are available and incomplete cold start (ICS) problem where only a small number of rating records are available for some new items or users in the system. In this paper, we propose two recommendation models to solve the CCS and ICS problems for new items, which are based on a framework of tightly coupled CF approach and deep learning neural network. A specific deep neural network SADE is used to extract the content features of the items. The state of the art CF model, timeSVD++, which models and utilizes temporal dynamics of user preferences and item features, is modified to take the content features into prediction of ratings for cold start items. Extensive experiments on a large Netflix rating dataset of movies are performed, which show that our proposed recommendation models largely outperform the baseline models for rating prediction of cold start items. The two proposed recommendation models are also evaluated and compared on ICS items, and a flexible scheme of model retraining and switching is proposed to deal with the transition of items from cold start to non-cold start status. The experiment results on Netflix movie recommendation show the tight coupling of CF approach and deep learning neural network is feasible and very effective for cold start item recommendation. The design is general and can be applied to many other recommender systems for online shopping and social networking applications. The solution of cold start item problem can largely improve user experience and trust of recommender systems, and effectively promote cold start items.     

Combining global and local matching of multiple features for precise item image retrieval

With the fast-growing of online shopping services, there are millions even billions of commercial item images available on the Internet. How to effectively leverage visual search method to find the items of users’ interests is an important yet challenging task. Besides global appearances (e.g., color, shape or pattern), users may often pay more attention to the local styles of certain products, thus an ideal visual item search engine should support detailed and precise search of similar images, which is beyond the capabilities of current search systems. In this paper, we propose a novel system named iSearch and global/local matching of local features are combined to do precise retrieval of item images in an interactive manner. We extract multiple local features including scale-invariant feature transform (SIFT), regional color moments and object contour fragments to sufficiently represent the visual appearances of items; while global and local matching of large-scale image dataset are allowed. To do this, an effective contour fragments encoding and indexing method is developed. Meanwhile, to improve the matching robustness of local features, we encode the spatial context with grid representations and a simple but effective verification approach using triangle relations constraints is proposed for spatial consistency filtering. The experimental evaluations show the promising results of our approach and system.     

基于证据网络的C2C商品购买风险动态评估模型

C2C(consumer to consumer)电子商务市场中的风险认知是消费者网络购物决策的重要依据之一.首先,分析电子商务市场中商家信誉、商品销量和在线商品评论对消费者网购决策的影响;然后,运用信度规则模型推理节点关系,建立基于证据网络的C2C商品购买风险动态评估模型;最后利用从某电子商务网站收集的观测信息,运用多元线性回归分析方法验证购买风险动态评估模型的可行性.借助本模型,可以帮助消费者在网购商品过程中较好地认知商品购买的风险.     

基于深度学习的女装图片分类探索

互联网商品图像的属性分类是人工智能领域的重要研究课题之一,针对商品图像属性分布不 平衡以及不同属性间存在相关性等问题,该文以女装图像为分类目标,提出了一种基于卷积神经网络的商品图像分类方法。首先,从电商网站获取大量商品图像,并进行人工标注;然后,基于卷积神经 网络框架,采用了一种有效的采样策略,通过增加新的损失函数,实现了基于多任务学习方法的商品图像属性准确分类;最后,通过对不同策略下分类结果的对比分析,验证了该方法的有效性。结果显 示,所提出方法具有较高的分类精度。     

A multi-attribute probabilistic matrix factorization model for personalized recommendation

Recommendation systems can interpret personal preferences and recommend the most relevant choices to the benefit of countless users. Attempts to improve the performance of recommendation systems have hence been the focus of much research in an era of information explosion. As users would like to ask about shopping information with their friend in real life and plentiful information concerning items can help to improve the recommendation accuracy, traditional work on recommending based on users’ social relationships or the content of item tagged by users fails as recommending process relies on mining a user’s historical information as much as possible. This paper proposes a new recommending model incorporating the social relationship and content information of items (SC) based on probabilistic matrix factorization named SC-PMF (Probabilistic Matrix Factorization with Social relationship and Content of items). Meanwhile, we take full advantage of the scalability of probabilistic matrix factorization, which helps to overcome the often encountered problem of data sparsity. Experiments demonstrate that SC-PMF is scalable and outperforms several baselines (PMF, LDA, CTR, SocialMF) for recommending.     

基于兴趣度的Web页面关联规则的研究

随着互联网上的信息迅速增长,如何快速准确地寻找到信息越来越受到人们的重视。文中给出了几种计算用户兴趣度的方法,并利用其中一种计算用户兴趣度的方法,论述了基于兴趣度的Web页面关联规则。论述了关联规则和一般的Apriori算法,并利用了"壹支持数下K—关联规则",对一般的Apriori进行了改进,主要是将兴趣度用于Apriori算法中。实验结果证明,该方法用于在网上寻找用户感兴趣的信息具有较好的准确率。     

基于细粒度特征交互的推荐模型

随着电子商务的兴起,用户在网购的同时留下了大量的评论。用户评论通常包含丰富的用户兴趣和项目属性等语义信息,反应了用户对项目特征的偏好。近年来,许多基于深度学习的方法通过利用评论进行推荐,并取得了巨大成功。这些工作主要是采用注意机制来识别对评分预测很重要的词或方面。它们单一的从评论中提取特征信息,并通过用户和物品的特征交互得到预测分数。然而,过度的聚合可能会导致评论中细粒度信息的丢失。此外,现有的模型要么忽略了用户和项目评论的相关性,要么只在单个粒度上构建评论特性交互,这导致用户和项目的特征信息不能被有效而全面地捕获。针对上述问题,在本文我们考虑通过从评论的多个粒度捕获特征信息,然后为用户和物品进行多粒度下的特征交互,可以实现更好的评分预测和解释性。
为此,我们提出了一种新的用于评分预测的细粒度特征交互网络（FFIN）。首先,模型并没有将用户的所有评论聚合成一个统一的向量,而是将用户和物品的每条评论单独建模,通过堆叠的扩展卷积分层地为每个评论文本构建多层次表示,充分地捕获了评论的多粒度语义信息;其次,模型在每个语义层次上构建用户和物品评论的细粒度特征交互,这有效避免了单粒度交互导致的次级重要信息被忽略的问题;最后,由于用户的评论行为通常是主观且个性化的,我们没有使用注意力机制来识别重要信息,而是通过类似于图像识别的层次结构来识别高阶显著信号,并将其用于最终的评分预测。我们在6个来自Amazon和Yelp的具有不同特征的真实数据集上进行了广泛的实验。我们的结果表明,与最近提出的最先进的模型相比,所提出的FFIN在预测精度方面获得了显著的性能提升。进一步的实验分析表明,多粒度特征的交互,不仅突出了评论中的相关信息,还大大提高了评分预测的可解释性。     

Session-based item recommendation in e-commerce: on short-term intents,reminders, trends and discounts

Many e-commerce sites present additional item recommendations to their visitors while they navigate the site, and ample evidence exists that such recommendations are valuable for both customers and providers. Academic research often focuses on the capability of recommender systems to help users discover items they presumably do not know yet and which match their long-term preference profiles. In reality, however, recommendations can be helpful for customers also for other reasons, for example, when they remind them of items they were recently interested in or when they point site visitors to items that are currently discounted. In this work, we first adopt a systematic statistical approach to analyze what makes recommendations effective in practice and then propose ways of operationalizing these insights into novel recommendation algorithms. Our data analysis is based on log data of a large e-commerce site. It shows that various factors should be considered in parallel when selecting items for recommendation, including their match with the customer’s shopping interests in the previous sessions, the general popularity of the items in the last few days, as well as information about discounts. Based on these analyses we propose a novel algorithm that combines a neighborhood-based scheme with a deep neural network to predict the relevance of items for a given shopping session.     

融合注意力机制的深度协同过滤推荐算法

传统基于项目的协同过滤算法在计算项目之间相似度时只考虑历史项目的评分，而忽略了历史项目偏好对其的影响，以至于推荐精度不够理想。针对此问题，提出了一种融合注意力机制的深度电影推荐算法。根据得到的隐性反馈，在特征级注意力框架上，从项目内容特征提取网络开始，学习项目特征的偏好；将项目特征偏好与项目特征加权得到项目内容特征向量；在项目级特征注意力框架中，通过项目内容特征向量学习对项目偏好的评分，从而产生最终的推荐结果。实验结果表明，提出的推荐算法在MovieLens 100K和MovieLens 1M两个公开数据集上的准确率和推荐个性化较传统算法均有不同程度的提高，表现出较为优越的推荐性能。     

Perceived security determinants in e-commerce among Turkish university students

Perceived security is defined as the level of security that users feel while they are shopping on e-commerce sites. The aims of this study were to determine items that positively influence this feeling of security by users during shopping, and to develop guidelines for perceived security in e-commerce. An experiment allowed users with different security assurances to shop on simulated e-commerce sites. The participants were divided into three groups, shopping for cheap, mid-range, and expensive products, respectively. Following the shopping environment, the virtual shopping security questionnaire (VSSQ), consisting of fourteen perceived security items, was presented to the users. The VSSQ was presented to the participants to validate these perceived security items. The VSSQ had a Cronbach's alpha internal reliability value of 0.70. With the exception of two items, there were no significant differences in item ratings between the groups of different shopping item values. A factor analysis procedure determined two main factors concerning perceived security in e-commerce. The perceived operational factor includes: the site's blocking of unauthorized access; emphasis on login name and password authentication; funding and budget spent on security; monitoring of user compliance with security procedures; integration of state-of-the-art systems; distribution of security items within the site; website's encryption strategy; and consolidation with network security vendors. The perceived policy-related factor includes: the website's emphasis on network security; top management commitment; effort to make users aware of security procedures; the website's keeping up-to-date with product standards; the website's emphasis on security in file transfers; and issues concerning the web browser.     

基于高斯pLSA模型与项目的协同过滤混合推荐

协同过滤是推荐系统中常用的一种技术。以往的推荐算法往往只从用户或商品的角度单一地进行推荐,在推荐准确率上存在瓶颈和局限性。提出了一种新的混合推荐方法——结合基于高斯概率潜在语义分析模型与改进的基于项目的协同过滤算法,通过建立用户群体混合模型和基于目标项目的邻居集进行预测推荐。实验证明该算法与其他协同过滤算法相比具有更高的准确率。     

Consumer informedness and diverse consumer purchasing behaviors: Traditional mass-market,trading down,and trading out into the long tail

As truly informed consumers are increasingly able to find exactly what they want and willing to pay premium prices to obtain products with perfect fit for them, companies have responded with new product portfolio strategies and new pricing strategies, based on the concepts of resonance marketing and hyperdifferentiation. This is not just consumers’ pursuit of products that are better, but rather better for them. It is not trading up, but rather trading out. In this paper we offer a more complete explanation of changes in consumer behavior, based on consumers’ new-found informedness, and an understanding of consumers’ pursuit of products that truly meet their individual wants and needs, cravings and longings.This paper also contributes to a deeper understanding of how online reviews are linked to sales. Recent empirical studies suggest that consumers use information in different ways in different shopping experiences, and that consumers’ purchasing behavior varies across different online shopping experiences; consequently, the best predictors of the success of different online products will therefore vary depending on what consumers are buying and why and how they are buying it.     

An improvement for semantics-based recommender systems grounded on attaching temporal information to ontologies and user profiles

Recommender systems in online shopping automatically select the most appropriate items to each user, thus shortening his/her product searching time in the shops and adapting the selection as his/her particular preferences evolve over time. This adaptation process typically considers that a user's interest in a given type of product always decreases with time from the moment of the last purchase. However, the necessity of a product for a user depends on both the nature of the own item and the personal preferences of the user, being even possible that his/her interest increases over time from the purchase. Some existing approaches focus only on the first factor, missing the point that the influence of time can be very different for different users. To solve this limitation, we present a filtering strategy that exploits the semantics formalized in an ontology in order to link items (and their features) to time functions. The novelty lies within the fact that the shapes of these functions are corrected by temporal curves built from the consumption stereotypes into which each user fits best. Our preliminary experiments involving real users have revealed significant improvements of recommendation precision with regard to previous time-driven filtering approaches.     

基于对抗采样的社交推荐算法

推荐系统利用用户的历史记录、物品的基础信息等数据进行建模来捕获用户的偏好,有效缓解了信息过载等问题,虽然其已应用广泛,但整个推荐领域面临的挑战却依旧存在,其中数据稀疏这一问题对于推荐性能有举足轻重的影响。近年来,大量研究表明基于社交信息的推荐算法能够有效缓解数据稀疏问题,但它们也仍然存在一定的局限。线上的社交网络是非常稀疏的,并且线上社交网络中的“朋友”通常包括同学、同事、亲戚等,因此,拥有显式朋友关系的用户不一定拥有相似的偏好,即直接利用显式朋友的兴趣偏好进行推荐会存在噪声问题。此外,大部分基于隐式反馈的算法通常直接对用户没有交互过的物品进行随机采样,然后将其作为用户实际交互过的物品的负样本来优化模型,然而用户没有交互过的物品并不代表用户不喜欢,这种粗粒度的采样策略忽略了用户的真实偏好,同样也带来了一定程度的噪声。生成对抗网络（GANs）因其在训练中捕获复杂数据分布的能力以及强大的鲁棒性被广泛应用到推荐系统中,为了减弱上述噪声问题带来的影响,本文基于生成对抗网络提出了一种细粒度的对抗采样推荐模型（ASGAN）,包括一个生成器和判别器。其中,生成器首先利用图表示学习技术初始化社交网络,接着为用户生成一个与其偏好相似的朋友,然后再从该朋友喜欢的物品集中同时生成该用户喜欢的物品和用户不喜欢的物品。判别器则尽可能区分出用户实际交互过的物品和生成器生成的两类物品。随着对抗训练的进行,生成器能更有效地进行社交朋友采样和物品采样,而判别器能够良好地捕获用户的真实偏好分布。最后,在三个公开的真实数据集上与现有的六个工作进行对比,实验结果证明：ASGAN拥有更好的推荐性能,通过重构社交网络和细粒度采样有效缓解了社交信息和物品采样策略带来的噪声问题。     

Efficiently using contextual influence to recommend new items to ephemeral groups

Group recommender systems suggest items to groups of users that want to utilize those items together. These systems can support several activities that can be performed together with other people and are typically social, like watching TV or going to the restaurant. In this paper we study ephemeral groups, i.e., groups constituted by users who are together for the first time, and for which therefore there is no history of past group activities.Recent works have studied ephemeral group recommendations proposing techniques that learn complex models of users and items. These techniques, however, are not appropriate to recommend items that are new in the system, while we propose a method able to deal with new items too. Specifically, our technique determines the preference of a group for a given item by combining the individual preferences of the group members on the basis of their contextual influence, the contextual influence representing the ability of an individual, in a given situation, to guide the group’s decision. Moreover, while many works on recommendations do not consider the problem of efficiently producing recommendation lists at runtime, in this paper we speed up the recommendation process by applying techniques conceived for the top-$K$ query processing problem. Finally, we present extensive experiments, evaluating: (i) the accuracy of the recommendations, using a real TV dataset containing a log of viewings performed by real groups, and (ii) the efficiency of the online recommendation task, exploiting also a bigger partially synthetic dataset.     

Location-aware online learning for top-k recommendation

We address the problem of recommending highly volatile items for users, both with potentially ambiguous location that may change in time. The three main ingredients of our method include (1) using online machine learning for the highly volatile items; (2) learning the personalized importance of hierarchical geolocation (for example, town, region, country, continent); finally (3) modeling temporal relevance by counting recent items with an exponential decay in recency.For (1), we consider a time-aware setting, where evaluation is cumbersome by traditional measures since we have different top recommendations at different times. We describe a time-aware framework based on individual item discounted gain. For (2), we observe that trends and geolocation turns out to be more important than personalized user preferences: user–item and content-item matrix factorization improves in combination with our geo-trend learning methods, but in itself, they are greatly inferior to our location based models. In fact, since our best performing methods are based on spatiotemporal data, they are applicable in the user cold start setting as well and perform even better than content based cold start methods. Finally for (3), we estimate the probability that the item will be viewed by its previous views to obtain a powerful model that combines item popularity and recency.To generate realistic data for measuring our new methods, we rely on Twitter messages with known GPS location and consider hashtags as items that we recommend the users to be included in their next message.     

Discovering the associations between test items in an examination

An examination is .an important approach to comprehensively evaluate students' abilities. An examination paper is composed of test items. Most studies assume that test items are independent, but there may, in fact, be associations among test items. How to select test items with low correlations into a test paper is our concern. Meanwhile, for diagnostic learning , students must perform highly relevant exercises to help them efficiently understand the concepts included in the test items. People usually find the associations among test items based on properties of individual test items. This paper proposes a test item association model to represent the correlations of test item pairs by mining the response data of examinees. Then, a hypothesis test and an empirical analysis were used to verify and analyse the relationship between the test item correlation and similarity. Next, the effects of test item parameters on the correlation are investigated by applying the model to the simulated response data which are produced by Item Response Theory (IRT) simulation data generation software. Finally, three applications of this test item association model are listed. After verification, the proposed model can be used for precise teaching.