Dany+-+Annotated+Bibliography

1 - Neumann, Andreas W. "A Survey of Reccomender Systems at Major STI Providers." //Recommender Systems for Information Providers: Designing Customer Centric Paths to Information//. Heidelberg: Physica-Verlag, 2009. Print.

Andrea Neuman’s chapter on “a survey of recommender systems at major STI providers” analyzes how recommendation systems are used for scientific and technical information as well as e-commerce. The author identifies how a recommendation system can be used to support users of electronic libraries to find related information on articles and books. Neuman uses the examples of recommendation systems for Scientific Libraries (ACM portal, IEEE Xplore, Citeseer and Google scholar), Scientific projects (Techlense and the Melvyl Recommender Project), e-commerce (amazon.com), and social tagging (BibSonomy and CiteULike, and LibraryThing) to illustrate how different formats of website require varying strategies to recommend things to consumers.

Even though the descriptions in the chapter are overly technical at some points, it brings relevant information on how challenging it is to implement recommender systems, especially in websites that contain scholarly articles. Furthermore, it shows how recommendation systems can be used for a plethora of web-based activities, not just e-commerce.

A particularly interesting argument based on Amazon.com system, is that these systems are vulnerable to corporate interests. Through anonymous ratings, businesses can make their products more likely to be recommended to consumers.

2 – Garfinkle, Robert, Ram Gopal, Bhavik Pathak, and Fang Yin. "Shopbot 2.0: Integrating Recommendations and Promotions with Comparison Shopping." //Elsevier - Decision Support Systems// 46 (2008): 61-69. Print.

This article assesses how recommender systems are implemented based on a study of Amazon.com. The authors argue that recommendation systems paired with shopbots add more value to the costumer’s online shopping experience by reducing search costs. They also trace a compelling parallel between recommendation systems and word of the mouth advertisement, concluding that the former is more objective and efficient.

The Amazon.com case study that Garfinkel et Al. conducted proved that, by using specific algorithms, amazon.com applies its recommendation system to its best interests as a business such as “inventory clearance through cross-selling” (Garfinkel et AL. 2008). This conclusion makes the reader wonder to what extent are recommendation systems beneficial to consumers or just a business strategy.

3 – Manouselis, Nikos. "Deploying and Evaluating Multiattribute Product Recommendation in E-markets." //International Journal of Management and Decision Making// 9.1 (2008): 43-61. Print.

In this article, Manousellis conducts an experiment to determine which recommender system algorithm yields a more effective e-commerce strategy. Manouselis provides a brief summary of different techniques used in composing recommendation systems. He bases his study on the MultiAttribute Utility Theory which is “the most common approach to represent buyer preferences and drive recommendation process” (Manousellis, 2008) accounting for the various criteria buyers consider when making a purchase decision. The experiment results and applications of algorithms on real scenarios showed that the Pearson Similarity Weighing algorithm is the most effective in providing useful recommendations to buyers. Based in this conclusion, the author states that even though algorithms are important, there is more to recommender systems, such as the user interface, that is important to consider when implementing a recommendation strategy. Yet, Manousellis acknowledges that an experimental study is extremely limited because the real process is highly interactive and subjective and therefore hard to mimic under experimental conditions.

4- Fleder, Daniel, and Kartik Hosanagar. "Blockbuster Culture's Next Rise or Fall: The Impact of Recommender Systems on Slaes Diversity." //Management Science// 55.5 (2009): 697-712. Print.

In their article, Daniel Fleder and Kartik Hosonagar refute the assumption that recommender systems create variety in products’ offer and argue that, instead, they just help install already popular ones by conducting a market simulation. In their experiment, they use collaborative filter type of recommender systems, which “recommend what similar costumers liked or bought” (Fleder & Hosanagar, 2009), to conclude that such systems standardize consumption patterns between consumers. According to his results, recommender systems also facilitate business transactions for firms by making it easier to control inventory.

Even though the article provides a relevant insight on how recommendation systems affect the market, the authors are very vague in assuming that recommender systems are believed to increase variety of offers to consumers. Fleder and Hosanagar do not state where they derive this assumption from.

5- Tveit, Amund. "Peer-to-peer Based Recommendations for Mobile Commerce." //Proceedings of the 1st International Workshop on Mobile Commerce//. International Workshop on Mobile Commerce, Rome, Italy. New York: ACM, 2001. 26-29. Print.

Amund Tveit differentiates information filtering from collaborative filtering recommender systems. He claims that collaborative filtering is more effective in terms of reaching customers and proposes a method to scale it to mobile devices software. His model is based on the //gnutella// algorithm and is especially concerned with the user’s privacy. This article is interesting because it raises the issue of scalability, and how recommender systems algorithms depend on the device being used.