Terry

FINAL

 * Information and data flow**


 * System Contributors**

Recommender systems are designed to be used in e-commerce websites to allow users to locate their preferred product seamlessly, without having to face the possible situation of information overloads (Chen, and Mcleod). Consumers like us are often the ones who are responsible for contributing data and information to the system. A number of previous studies have indicated that this group of consumers primarily consists of college students (Ant Ozok, Fan and Norcio). With today’s advance technology, these Internet savvy students are most susceptible to using retail e-commerce. Basically, anyone who is interested in purchasing a product from an e-commerce website is responsible for contributing to the recommender system. Whenever one purchases a product online, one would first have to key in specific product details and attributes such as the size, quantity and color. This information plays an important role as it is being collected and transformed into useful data for the recommender system. Using the example of a website selling movie DVDs, consumers can locate a movie title based on their interests, genre, language, duration, theme or cast. The recommender system would subsequently present a list of movies based on the consumer’s preferences. These preferences as stated by the consumers are information for the recommender system. The system is then able to analyze the consumers’ shopping behavior and used it as a predictor for future buying behavior (Schafer, Konstan, and Riedl).


 * Users of the system**

Consumers who are interested in purchasing a product rely on the system to push out a list of recommended products based on their stated preferences. These consumers are usually unsure of their decision on buying that particular product. In some cases, some consumers have little or no knowledge about the products. They then have to rely on this recommender system to determine their purchasing decision. The recommender system acts as a form of shopping advice to the consumer, since it presents out to him with useful information such as the ratings of a particular movie, price comparison (with other e-commerce websites), and reviews from people who have purchased the product before. The recommender system also presents alternative or complementary products to consumers (Ant Ozok, Fan and Norcio). For example, if he selects the movie title “The Day After Tomorrow”, the system might recommend the movie “2012” to him too since they belong to the same genre/theme. At times, the system might also attempt to cross-sell him a movie memorabilia such as an autographed wall poster or t-shirt.


 * Types of decisions made from the system**

The data that is churned out of the recommender system provides consumers with a list of products which matches their preferences and analyzes their shopping behavior. The system might track where the user clicks on a website and from that, determine a list of recommendations that might be of interest to the consumer. There are two types of decisions that can be derived from the recommender system. Firstly, the consumers’ minds are set and have decided on their final purchasing decisions. The system narrows down the wide range of products to a more focused list, which allows the consumer to analytically decide what he wants to buy. In addition, information such as feedback reviews and ratings are presented to aid in the decision making process.

The second type of decision would be consumers walking away from the e-commerce website and deciding not to purchase any product. This could be the result of a poorly designed recommender system, which provided him with a list of products that fail to grab his attention. Often, consumers want to know the transparency factor on how and why the system presents such recommendations to him (Ant Ozok, Fan and Norcio). Another reason could be the usability of the system, specifically user interface issues such as navigation and layout (Ant Ozok, Fan and Norcio). For example, a consumer always has to return to the home page and key in all his preferences again after adding a particular movie title into the checkout cart. He might find the purchasing process troublesome, tedious and non-user friendly. This would subsequently decline his interest and attention level, resulting him in to cancel his purchase. For random visitors (with no intention of buying) to an e-commerce website, the recommender system could possibly help them find products which could be of interest to them and even convince them to buy it.


 * Information flow of recommendation system**

In the recommender system, it consists of two components in which information travels in and out. This is basically the user and the computer. In this context, the computer will be known as the server. In order for the recommender system to push out with a list of recommendations, the user would first have to provide information to the server. The flow chart of the recommender system goes like this: user à server à user. Using the example of movies, the user would have to key in his preferences and select all the relevant attributes on the website. This information is collected and sent to the server. Over here, the recommender system within the server would analyze the consumer’s input using various algorithmic methods, content based and collaborative filtering methods. Upon computing the information, it is returned back to the user where he can use it to make a purchasing decision. The recommender system functions according to the information it receives from the user. There are millions of user data that can vary within a short period of time. For instance, shopping behavior patterns are not fixed as there are constantly new shoppers entering the system at any one time (Chen, and Mcleod).

Chen, Anne Yun-An, and Dennis Mcleod. "Collaborative Filtering for Information Recommendation Systems." Department of Computer Science and Integrated Media System Center n. pag. Web. 29 Mar 2010. 

Ant Ozok, A., Quyin Fan, and Anthony F. Norcio. "Design guidelines for effective recommender system interfaces based on a usability criteria conceptual model: results from a college student population ." Behaviour & Information Technology. 29.1 (2010): 57-83. Print. Schafer, J.Ben, Joseph Konstan, and John Riedl. "Recommender Systems in E-Commerce." GroupLens Research Project Department of Computer Science and Engineering). Print.