The Background, Approach, and Literature Review draft must conform unerringly to the University College Format and Style Requirements . This draft follows all capstone instructions. The draft is submitted online as an assignment in either a .doc or .docx file format. The draft includes a Title Page, Abstract Heading, and References. The following is the order of this draft's elements. Each numbered item, except for the Title Page, is a major section indicated by a level-one heading. Subsections follow the level-two and level-three heading guidelines. The abstract page should only contain the header for now. You will write the full 120-word abstract once you have completed writing your entire draft.
Order of Contents
1. Title Page
2. Abstract (heading only)
3. Table of Contents
4. Background - 750 words
5. Approach - 250 words
6. Literature Review - 3,000 words
This material may consist of step-by-step explanations on how to solve a problem or examples of proper writing, including the use of citations, references, bibliographies, and formatting. This material is made available for the sole purpose of studying and learning - misuse is strictly forbidden.The impacts of clustering algorithms on the accuracy of the recommender systems
The emergence of the Internet as a concept in the 1950s and its actualization in the 1980s, as discussed by Cohen-Almagor (2011), is perhaps the most revolutionary event in the history of humanity. This is primarily so because it has sufficed unique opportunities for individuals and organizations. One of the breakthroughs facilitated by the Internet is e-commerce. Fundamentally, e-commerce is the practice of buying and selling items and services over digital media, primarily the Internet. Researchers such as Dahiya (2017) contend that e-commerce has remained as a major threat to offline or brick-and-mortar firms. Indeed, e-commerce has become commonplace, owing to the proliferation of its use. Thanks to its e-commerce business model, Amazon, a prominent retail company based in the U.S., currently boasts of 310 million customers drawn across the world (Amazon, 2020). The nearest physical competitor is Wal-Mart, which has a customer base of 256 million (Walmart, 2020). The winning formula for e-commerce organizations, according to Dahiya (2017), is primarily because of their convenience often delivered by easy accessibility and low prices. The majority of e-commerce organizations disseminate large volumes of information through their online platforms. However, a key problem in the shape of overwhelming the customers with excessive information such that they are unable to digest it has sufficed.
Luckily, the most progressive e-commerce organizations have integrated recommender systems to make the dissemination as well as consumption of online information not only more targeted but also convenient (Schafer, Konstan, & Riedl, 2001). Broadly, a recommender system is a subclass of information filtering system commonly applied to predict the preference or rating that a user will potentially give to an item (Aggarwal, 2016). This information is then used to make the experience of the customer more satisfied. For example, the recommender systems assist the customers or the consumers in finding their preferred products and services easily and quickly in an increasingly complex e-commerce domain by suggesting items to them (Bhatnagar, 2014). YouTube is a benchmark recommender system, given that it applies a sophisticated infrastructure to suggest videos to the users founded on their viewing history. Pierce (2016) explains that users often upload at least 400 hours of videos each minute on YouTube. YouTube recommender system enables the users to single out the content that relates to the best rather than the consumer having to navigate the entire range of videos. Amazon also follows this procedure, given that the company’s recommender system often suggests newly published items based on the purchase history of the users.
Important to say is that the quality of the recommender system is often based on the accuracy of its recommendation, which is usually expressed through the following formula;(#Correct recommendations)/(#All Recommendations). Problematically, low-quality recommender systems usually make it difficult for an entity to secure new customers. At the same time, insufficient recommender systems make life unproductive for the existing customers owing to the continuous junk advert emails that appear threatening than inviting. This, therefore, prompts them to log out of the system. Unfortunately, as documented by Madadipouya and Chelliah (2017), it appears that even the most successful firms have not found a breakthrough in carving out accurate recommender systems. Indeed, commentators assert that the degree of accuracy portrayed by YouTube and Amazon recommender systems stands at only about 60% each (Trouvus, 2015 and Grajales, 2017). The case for other recommender systems is even more worrying, given that their accuracy levels fall way below the 60% level. The accuracy problem is triggered by the inadequacy of algorithms and data sets on which the current recommender systems are based.
As shown by Schafer, Frankowski, Herlocker, and Sen (2007), Mohamed, Hussien, Khafagy, and Ibrahim (2015), and Al-Bashiri, Abdulgabber, Romli, and Kahtan (2018)...