基于用户兴趣和评分差异的改进混合推荐算法

为进一步提高协同过滤推荐算法的推荐精度,提出一种基于用户兴趣和评分差异的改进混合推荐算法.利用词频-逆向文件频率(TF-IDF)思想对稀疏矩阵进行填充,在计算用户相似度时在传统的修正余弦相似度计算公式中引入两个不同的影响因子来改善用户评分差异的影响,使用两种不同的时间衰减函数用于修正时间因素对用户和项目之间以及用户与用户之间的影响.实验结果表明,该算法能够缓解数据稀疏的问题,有效修正用户评分差异和用户兴趣变化对推荐结果的影响,其推荐精度均优于现有其它改进算法.     

A process model for interactive systems

Designing user interfaces and designing computational software are very different processes. The differences lead to late discovery of design conflicts, which drives up development costs. A unifying methodology that could provide early discovery and resolution of design conflicts must account for the governing principles of both processes. Disciplined long-term investigation of candidate methodologies requires that these governing principles be fixed and that evolving development methods comprising each process be accommodated. This article describes an application of general systems theory to integrate these principles, proposes a process model that fixes them as explicit elements of a process program, argues the feasibility of the model and its worthiness for further study, and describes its initial implementation.     

How scales influence user rating behaviour in recommender systems

Many websites allow users to rate items and share their ratings with others, for social or personalisation purposes. In recommender systems in particular, personalised suggestions are generated by predicting ratings for items that users are unaware of, based on the ratings users provided for other items. Explicit user ratings are collected by means of graphical widgets referred to as ‘rating scales’. Each system or website normally uses a specific rating scale, in many cases differing from scales used by other systems in their granularity, visual metaphor, numbering or availability of a neutral position. While many works in the field of survey design reported on the effects of rating scales on user ratings, these, however, are normally regarded as neutral tools when it comes to recommender systems. In this paper, we challenge this view and provide new empirical information about the impact of rating scales on user ratings, presenting the results of three new studies carried out in different domains. Based on these results, we demonstrate that a static mathematical mapping is not the best method to compare ratings coming from scales with different features, and suggest when it is possible to use linear functions instead.     

Combining user preferences and user opinions for accurate recommendation

Recommendation systems represent a popular research area with a variety of applications. Such systems provide personalized services to the user and help address the problem of information overload. Traditional recommendation methods such as collaborative filtering suffer from low accuracy because of data sparseness though. We propose a novel recommendation algorithm based on analysis of an online review. The algorithm incorporates two new methods for opinion mining and recommendation. As opposed to traditional methods, which are usually based on the similarity of ratings to infer user preferences, the proposed recommendation method analyzes the difference between the ratings and opinions of the user to identify the user’s preferences. This method considers explicit ratings and implicit opinions, an action that can address the problem of data sparseness. We propose a new feature and opinion extraction method based on the characteristics of online reviews to extract effectively the opinion of the user from a customer review written in Chinese. Based on these methods, we also conduct an empirical study of online restaurant customer reviews to create a restaurant recommendation system and demonstrate the effectiveness of the proposed methods.     

面向用户群体的Web服务推荐

针对Web服务推荐中服务用户调用Web服务的服务质量数据稀疏性导致的低推荐质量问题,提出了一种面向用户群体并基于协同过滤的Web服务推荐算法(WRUG)。首先,为每个服务用户根据用户相似性矩阵构建其个性化的相似用户群体;其次,以相似用户群体中心点代替群体从而计算用户群体相似性矩阵;最后,构造面向群体的Web服务推荐公式并为目标用户预测缺失的Web服务质量。通过对197万条真实Web服务质量调用记录的数据集进行对比实验,与传统基于协同过滤的推荐算法(TCF)和基于用户群体影响的协同过滤推荐算法(CFBUGI)相比,WRUG的平均绝对误差下降幅度分别为28.9%和4.57%;并且WRUG的覆盖率上升幅度分别为110%和22.5%。实验结果表明,在相同实验条件下WRUG不仅能提高Web服务推荐系统的预测准确性,而且能显著地提高其有效预测服务质量的百分比。     

The role of user models in cooperative interactive systems

For interactive systems to communicate in a cooperative manner, they must have knowledge about their users. This article explores the role of user models in such systems, with the goal of identifying when and how user models may be useful in a cooperative interactive system. User models are classified by the types of knowledge they contain, several user modelling characteristics that serve as dimension for an additional classification of user models are presented, and user model representations are discussed. These topics help to characterize the space of user modelling in cooperative interactive systems-addressing how they can be used-but do not fully address when it is appropriate to include a user model in an interactive system. Thus, a set of design considerations for user models is presented, while a final example illustrates how these topics influence the user model for a hypothetical investment consulting system.     

DaisyViz: A model-based user interface toolkit for interactive information visualization systems

While information visualization technologies have transformed our life and work, designing information visualization systems still faces challenges. Non-expert users or end-users need toolkits that allow for rapid design and prototyping, along with supporting unified data structures suitable for different data types (e.g., tree, network, temporal, and multi-dimensional data), various visualization, interaction tasks. To address these issues, we designed DaisyViz, a model-based user interface toolkit, which enables end-users to rapidly develop domain-specific information visualization applications without traditional programming. DaisyViz is based on a user interface model for information (UIMI), which includes three declarative models: data model, visualization model, and control model. In the development process, a user first constructs a UIMI with interactive visual tools. The results of the UIMI are then parsed to generate a prototype system automatically. In this paper, we discuss the concept of UIMI, describe the architecture of DaisyViz, and show how to use DaisyViz to build an information visualization system. We also present a usability study of DaisyViz we conducted. Our findings indicate DaisyViz is an effective toolkit to help end-users build interactive information visualization systems.     

An instrumentation system to measure user performance in interactive systems

Many software tools are interactive in nature and require a close match between the user's knowledge of how a task is to be performed and the capabilities the tool provides. This paper describes the current status of an instrumentation and analysis package to measure user performance in an interactive system. A prototype measurement system is considered to evaluate a screen editor and to develop models of user behavior.     

Discovering user profiles for Web personalized recommendation

With the growing popularity of the World Wide Web, large volume of user access data has been gathered automatically by Web servers and stored in Web logs. Discovering and understanding user behavior patterns from log files can provide Web personalized recommendation services. In this paper, a novel clustering method is presented for log files called Clustering large Weblog based on Key Path Model (CWKPM), which is based on user browsing key path model, to get user behavior profiles. Compared with the previous Boolean model, key path model considers the major features of users‘ accessing to the Web: ordinal, contiguous and duplicate. Moreover, for clustering, it has fewer dimensions. The analysis and experiments show that CWKPM is an efficient and effective approach for clustering large and high-dimension Web logs.     

The graph-based behavior-aware recommendation for interactive news

Applied Intelligence - Interactive news recommendation has been launched and attracted much attention recently. In this scenario, user’s behavior evolves from single click behavior to...     

Building a mobile movie recommendation service by user rating and APP usage with linked data on Hadoop

Movie recommendation systems are important tools that suggest films with respect to users’ choices through item-based collaborative filter algorithms, and have shown positive effect on the provider’s revenue. Given that mobile Apps are rapidly growing, the recommender is implemented to support web services in frontend Apps. Among those films recommended, users can give ratings and feedback, collecting film information from linked data concurrently. In order to solve cold-start problems, Cluster-based Matrix Factorization is adopted to model user implicit ratings related to Apps usage. Knowing that user rating data processing is a large-scale problem in producing high quality recommendations, MapReduce and NoSQL environments are employed in performing efficient similarity measurement algorithms whilst maintaining rating and film datasets. In this investigation, the system analyzes user feedbacks to evaluate the recommendation accuracy through metrics of precision, recall and F-score rates, while cold-start users make use the system with two MovieLens datasets as main rating reference in the recommendation system.     

面向不一致用户评价准则的在线服务推荐

客观上,用户的评价准则是由主观意识决定的,用户之间的评价准则不同导致多个用户对同一服务的评分不具备可比较性,不考虑不同用户评分的不可比较性所获得的服务推荐将难以满足用户个性偏好及其真实需求。为此,提出一种面向不一致用户评价准则的在线服务推荐方法,考虑用户偏好不一致时用户对在线服务的偏好关系,以偏好关系计算用户之间的相似度,并以此获得在线服务推荐结果。首先以用户-服务评分矩阵为基础建立用户对服务的偏好关系,其次根据偏好关系计算用户之间的相似度,然后以用户相似度为基础对用户未评分的服务进行评分预测,最后以预测评分的排序结果作为推荐结果。与经典的协同过滤推荐方法的比较实验,验证了本方法的有效性。实验表明,本方法获得的推荐结果能满足大多数用户的服务偏好,同时获得了比经典的协同过滤推荐方法更好的准确率。     

Semantic user profiling techniques for personalised multimedia recommendation

Due to the explosion of news materials available through broadcast and other channels, there is an increasing need for personalised news video retrieval. In this work, we introduce a semantic-based user modelling technique to capture users’ evolving information needs. Our approach exploits implicit user interaction to capture long-term user interests in a profile. The organised interests are used to retrieve and recommend news stories to the users. In this paper, we exploit the Linked Open Data Cloud to identify similar news stories that match the users’ interest. We evaluate various recommendation parameters by introducing a simulation-based evaluation scheme.     

A multi-formalism approach for model-based dynamic distribution of user interfaces of critical interactive systems

Evolution in the context of use requires evolutions in the user interfaces even when they are currently used by operators. User Centered Development promotes reactive answers to this kind of evolutions either by software evolutions through iterative development approaches or at runtime by providing additional information to the operators such as contextual help for instance. This paper proposes a model-based approach to support proactive management of context of use evolutions. By proactive management we mean mechanisms in place to plan and implement evolutions and adaptations of the entire user interface (including behaviour) in a generic way. The approach proposed handles both concentration and distribution of user interfaces requiring both fusion of information into a single UI or fission of information into several ones. This generic model-based approach is exemplified on a safety critical system from space domain. It presents how the new user interfaces can be generated at runtime to provide a new user interface gathering in a single place all the information required to perform the task. These user interfaces have to be generated at runtime as new procedures (i.e. sequences of operations to be executed in a semi-autonomous way) can be defined by operators at any time in order to react to adverse events and to keep the space system in operation. Such contextual, activity-related user interfaces complement the original user interfaces designed for operating the command and control system. The resulting user interface thus corresponds to a distribution of user interfaces in a focus+context way improving usability by increasing both efficiency and effectiveness.     

Preference-based reasoning in BDI agent systems

An important feature of BDI agent systems is number of different ways in which an agent can achieve its goals. The choice of means to achieve the goal in made by the system at run time, depending on contextual information that is not available in advance. In this article, we explore ways that the user of an agent system can specify preferences which can be incorporated into the BDI execution process and used to guide the choices made. For example, a user of a travel system can specify a preferred airline, or a particular kind of accommodation, and the system will use this information to satisfy the goal and preferences, if possible. Preferences are specified in terms of properties of goals and resource usage, and are used to make two types of decisions: (a) select a plan when there is a choice and (b) determine the order in which subgoals of a plan should be pursued when their order is not fixed by design. We have implemented our preference framework in Jadex, and provide detailed case studies within the context of a holiday travel agent application.     

Graph bandit for diverse user coverage in online recommendation

We study a recommendation system problem, in which the system must be able to cover as many users’ preferences as possible while these preferences change over time. This problem can be formulated as a variation of the maximum coverage problem; specifically we introduced a novel problem of Online k-Hitting Set, where the number of sets and elements within the sets can change dynamically. When the number of distinctive elements is large, an exhaustive search for even a fixed number of elements is known to be computationally expensive. Even the static problem is known to be NP-hard (Hochba, ACM SIGACT News 28(2):40–52, 1997) and many known algorithms tend to have exponential growth in complexity. We propose a novel graph based UCB1 algorithm that effectively minimizes the number of elements to consider, thereby reducing the search space greatly. The algorithm utilizes a new rewarding scheme to choose items that satisfy more users by balancing coverage and diversity as it construct a relational graph between items to recommend. Experiments show that the new graph based algorithm performs better than existing techniques such as Ranked Bandit (Radlinski et al. 2008) and Independent Bandits (Kohli et al. 2013) in terms of satisfying diverse types of users while minimizing computational complexity.     

用户个性化推荐系统的设计与实现

为实现个性化服务,理解用户兴趣就成了提供服务的关键任务,因此,提出了隐性采集用户浏览内容、用户浏览时间和用户操作时间的信息方法,通过对网络爬虫程序抓取的网页进行内容清洗提取出主要内容之后,利用VSM建立文档模型,并采用SVM分类方法建立推荐库.基于从客户端采集的用户兴趣信息建模,以及根据该模型和推荐库的相似度,给用户推荐信息.此外,给出了基于该模型的推荐原型系统的实现,使用查准率来评价该系统.试验结果表明,系统较好地实现了基于用户兴趣来推荐阅读的信息.     

基于用户关注度的个性化新闻推荐系统

为满足用户需求,以用户为中心,解决用户关注度不断变化、数据稀疏性、优化时间和空间效率等问题,提出基于用户关注度的个性化新闻推荐系统。推荐系统引入个人兴趣和场景兴趣来描述用户关注度,使用雅克比度量用户相似性,对相似度加权求和预测用户关注度,从而提供给用户经过排序的新闻推荐列表。实验结果表明,推荐系统有效地提高了推荐精准度和覆盖度,改善了系统可扩展性和自动更新能力,具有良好的推荐效果。     

知识表示学习研究进展

新型深度学习推荐模型已广泛应用至现代推荐系统,其独有的特征——包含万亿嵌入参数的嵌入层,带来的大量不规则稀疏访问已成为模型预估的性能瓶颈. 然而,现有的推荐模型预估系统依赖CPU对内存、外存等存储资源上的嵌入参数进行访问,存在着CPU-GPU通信开销大和额外的内存拷贝2个问题,这增加了嵌入层的访存延迟,进而损害模型预估的性能. 提出了一种基于GPU直访存储架构的推荐模型预估系统GDRec.GDRec的核心思想是在嵌入参数的访问路径上移除CPU参与,由GPU通过零拷贝的方式高效直访内外存资源. 对于内存直访,GDRec利用统一计算设备架构（compute unified device architecture,CUDA）提供的统一虚拟地址特性,实现GPU 核心函数（kernel）对主机内存的细粒度访问,并引入访问合并与访问对齐2个机制充分优化访存性能;对于外存直访,GDRec实现了一个轻量的固态硬盘（solid state disk,SSD）驱动程序,允许GPU从SSD中直接读取数据至显存,避免内存上的额外拷贝,GDRec还利用GPU的并行性缩短提交I/O请求的时间. 在3个点击率预估数据集上的实验表明,GDRec在性能上优于高度优化后的基于CPU访存架构的系统NVIDIA HugeCTR,可以提升多达1.9倍的吞吐量.