Maciek+-+Annotated+Bibliography

===Adomavicius, Gediminas, & Alexander Tuzhilin. "Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions". //IEEE Transactions on Knowledge and Data Engineering// 17.6 (2005): 734-749. Print.=== This paper proposes several ways in which current generation recommendation systems can be improved to be more effective in a wider range of applications. The authors provide a detailed survey of methods used by recommendation systems which fall into three categories: content-based, collaborative, and hybrid. A movie recommending approach is used as an example for each method to add clarity. General algorithms for each method are analyzed, their limitations are identified, and representative research examples are provided. The authors argue that many real-life recommendation applications are more complex, which creates a need for more advanced methods, particularly in business settings. Various approaches to extending current recommendation methods to extend their capabilities are proposed. The paper concludes that the intent is to advance discussion about the next generation of recommendation systems.

===Blanco-Fernandez, Yolanda, et. al. "Incentivized provision of metadata, semantic reasoning and time-driven filtering: Making a puzzle of personalized e-commerce". //Expert Systems with Applications// 37.1 (2010): 61-69. Print.=== The authors discuss their e-commerce system in which the users actively participate in providing metadata about resources. Current e-commerce recommending systems lack semantic annotations and have poor time adaptation. They argue that allowing users to annotate resources is the only solution towards having rich semantic metadata. The recommending system design is explained in detail, outlining how user generated annotations are formulated into an ontology, and the time-driven semantically reasoned algorithms used to produce user tailored recommendations. A sample scenario is provided which helps to put all the pieces of the system into perspective. The system is put to experimental evaluation; its method and results are well documented which confirm that a time aware semantically reasoned filter produces high quality recommendations. Future plans to improve the system, and extend it into the digital television field are discussed in the conclusion.

===IM, IL, & Alexander Hars. "Does a One-Size Recommendation System Fit All? The Effectiveness of Collaborative Filtering Based Recommendation Systems Across Different Domains and Search Modes". //ACM Transactions on Information Systems// 26.1 (2007): 4:1-4:30. Print. CITED IN PAPER === Past research on collaborative filtering (CF) has two major limitations. It has been too focused on consumer products, while neglecting other domains. CF needs to be understood in different context in order to be designed better. There is also a lack of studies which address effects from user-side factors. The authors aim to examine how domain and user factors affect recommendation accuracy of CF systems in order to formulate guidelines towards better CF-based recommendation systems. A discussion on the concept of collaborative filtering along with reviews of past research is provided. The authors conducted a study which measured the differences in how users evaluate items in knowledge and product domains, how the search mode (problematic vs scanning) affected these evaluations, and its effects on recommendation accuracy. They conclude that performance of a CF system is domain dependent, and the users search mode affects accuracy. In order to increase accuracy, CF systems’ need to be designed to discriminate between search modes and domains to apply appropriate algorithms.

===Kim, Taek-Hun, & Sung-Bong Yang. "Using attributes to improve prediction quality in collaborative filtering". //E-Commerc and Web Technologies//. Ed. Bauknecht K, Bichler M, & Proll B. Germany: Srpinger-Verlag Berlin, 2004. 1-10. Print. CITED IN PAPER === The authors argue that in order for a collaborative filtering (CF) system to produce more accurate recommendations, it must exploit the attributes of each item, and have more refined neighbor selection models. A brief discussion on CF, and various neighbor selection models are presented. The paper proposes a new prediction formula which considers attributes of items during recommendation. Six separate recommender systems are setup using different neighbor models. The GroupLens Research Group movie dataset is used. Each system is tested with, and without attributes, and in each case the system using attributes produced more accurate predictions. The authors conclude that using attributes in a CF system not only improves accuracy, but can solve scalability issues of large scale systems without affecting the quality of recommendations.

Zhen, Lu, George Q. Huang, & Zuhua Jiang. "An inner-enterprise knowledge recommender system". //Expert Systems with Applications// 37.2 (2010): 1703-1712. Print.
The authors propose a model for an inner-enterprise knowledge recommendation system which focuses on both user and knowledge context. The paper first introduces related works with regard to knowledge management, and recommendation systems in general, providing sufficient background information. The rest of the paper focuses on implementing a knowledge recommendation system in a manufacturing firm in China. It outlines the necessary knowledge resources which should be in place prior to implementing a recommendation system, and its architecture. The design of the system is presented with an analysis of key technologies being used. The authors argue that through flexible recommendation rules engine, context awareness, and semantic reasoning, the system is flexible enough to be adapted into different enterprise contexts.