Background
In this project I will develop a website that implements the recommender system. Over the course of internet history, it has developed to become a more user-friendly area to give an easier and a more efficient way of surfing the internet to users. The expanding significance of the Web as a mechanism for electronic and business interactions has filled in as a driving force for the advancement of recommender systems. An important facilitator in such manner is the ease of which the Web empowers users to give criticism about their preferences or aversions.
Recommender systems have played a major role in the development of a more user- friendly internet. “A recommender system is an information filtering system that pursues to predict the “rating” or “preference” a user would give to an item.” Recommender Systems are used in different area’s such Music, Movies, Social media advertising Recommender systems are vital for large businesses like Amazon, Netflix, Spotify, YouTube etc. there are 2 common recommendation systems which are Collaborative filtering and Content based filtering. Recommendations are an important measure of the tailored user experience for any business that uses recommendation algorithms on their website. The most common businesses are Netflix, Amazon, Netflix and Spotify have shown how the use and development of recommendation systems have changed the online experience of its users. Amazon, for instance, have calculated an estimation of 35% of its sales being directly contributed by their recommendation system.
In December 2016 Netflix announced its first global recommendation engine. The new recommendation engine takes a number of different algorithms into account and matches the user with similar other users in 190 different countries where the Netflix service is available to its users. Netflix has estimated that only 20% of its subscribers’ users video choices comes from the search bar whilst 80% comes from recommendations to its users. This shows how essential recommendation engines have become in recent internet history. Carlos Gomez-Uribe who is the vice president of product innovation at Netflix has said “If one member in this tiny island expresses an interest in anime, then we’re able to map that person to the global anime community,” (Nathan McAlone, 2016). Netflix priced their recommendation system at $1 billion per year. Netflix’s Chief Product Officer Neil Hunt has said that “the combined effect of personalization and recommendations save us more than $1B per year.” (Nathan McAlone, 2016). Netflix’s reason of pricing its recommendation system at $1 billion is because it minimizes the chances of subscribers from cancelling their subscription. Netflix has said “Consumer research suggests that a typical Netflix member loses interest after perhaps 60 to 90 seconds of choosing,” (Nathan McAlone, 2016). They have also said “The user either finds something of interest or the risk of the user abandoning our service increases substantially.” (Nathan McAlone, 2016).
Related
Collaborative filtering methods are used to collect and analyse a large amount of information of users such as their preferences and predicting what users will like based on the similarities compared to other users. For example, if another user purchased a product the collaborative filtering system examines the considerations of other people who have a similar interest in a product or service and recommends it to the user. Another method of collaborative filtering is Item based collaborative filtering. this is a model-based calculation for recommender Systems. In item based collaborative filtering resemblances between items are based upon these similarities. An example of this method is Amazon collaborative filtering system. Amazon uses recommendation systems as an advertising method to target its customers to purchase products based on the item based collaborative filtering recommendation system. Amazon uses recommendation algorithms to distinguish to each user’s interests. When the user is about to purchase a product there are 2 columns which are ‘frequently bought together’ and ‘customers who bought this item and also bought’ . The ‘frequently bought together’ section recommends other products that were bought together for example, if the user was interested in a games console they would also be recommended to purchase an extra controller and a game. The second section is ‘customers who bought this item and also bought’. This will recommend products to the user that other users also purchased.
I will be able to use this method in my project to filter out the recommendations of a user to suit their profile based on similarities with other users. This would make the recommendation system more efficient and practical.
‘
Another method of a recommendation system is content-based filtering. Content-based filtering methods are based on an account of an item and a profile of the user’s preferences. recommender works with data that the user provides by either giving an item a rating or by clicking a link. This produces a user profile which is then used to offer recommendations of similar products or services. An example of this method being used is the Netflix recommender system. As the user gives more sources of info by watching films or gives a rating on what they have watched the more accurate the recommendations become to suit the users profile. Netflix have different sections such as ‘recently added’, ‘because you liked’ and ‘because you watched’. These sections are used to provide a more efficient experience for the user. The ‘recently added’ section is used to allow the user to browse through new movies and series that have been added to Netflix. The ‘because you liked’ shows similar movies or series that the user has liked. The recommendation algorithm takes in this information from the user and recommends similar movies and series to suit the user’s profile. The ‘because you watched’ section liked’ shows similar movies or series that the user has watched. The recommendation algorithm takes in this information from the user and recommends similar movies and series to suit the user’s profile.
Webpage
This first figure shows the homepage of my website. This is the welcoming page for a new user. The user will be able to search for. New music, view new music releases, search by genres, view personalised recommended music and the overview of the website. The overview of the website
This figure shows the genre the user will be able to select. Examples of genres the user will be able to select are pop, hip-hop, rock, soul etc.
This figure shows the genre the user has selected from the genre page. This will then be stored to the user profile by the recommendation system in order to create a personalised system for the user based on their search history. The recommendation engine will build a profile for the user to give a more user friendly and personalised experience for the user
This figure shows the search of the user for an artist and it also shows the recommendations of similar artists so that they will be able to discover new music and artists that suits their personalised profile.
This last figure shows the personalised page of the user. This page will show what artists are recommended to the user based on what they have recently listened to. The recommendation engine will build a profile for the user to give a more user friendly and personalised experience for the user. I will be able to use this method in my project to filter out the recommendations of a user to suit their profile based on similarities with other users. This would make the recommendation system more efficient and practical. In my project I will be using Content-based filtering. Content-based filtering methods are based on an account of an item and a profile of the user’s preferences. recommender works with data that the user provides by either giving an item a rating or by clicking a link. I will be able to use this method in my project to filter out the recommendations of a user to suit their profile based on what they have previously viewed. The recommendation System will filter to suit the user’s preferences it to create a more personalised profile.
Ethical and social issues
recommendation systems come with ethical and social issues.
One issue within recommendation systems is privacy. If the user shares personal information with a recommendation system, it results to better recommendation that fits the users profile users usually do not feel comfortable. Sharing out personal information over the web. On May the 25th a new data protection act came into fruition. This new law was instated in order to modernise data protection and offer more control to the public. The Act provides exemptions to the rights individuals have over their personal data under the GDPR ‘General data protection regulation’ (Kate Brimsted, Tom Evans, 2018). This means that the user would not have to produce personal information in order to access a website.
Before GDPR was introduced the previous data protection rules across Europe were first created during the 1990s and had struggled to keep pace with rapid technological changes. GDPR alters how businesses and public sector organisations can handle the information of their customers. It also boosts the rights of individuals and gives them more control over their information.
This causes major problems to websites that use recommendation systems for as it means that the user can decline the request to share personal information that makes it harder to recommend services that would fit the user’s profile. For example, Collaborative filtering methods are used to collect and analyse a large amount of information of users such as their preferences and predicting what users will like based on the similarities compared to other users. If the user doesn.t want to share data, it limits the effect of the recommendation algorithm as it will not be able to determine the user’s preferences and predictions.
One social issue with recommendation systems is that users may not be recommended products or services that fits the user’s profile. An example of this would be if a child was on Netflix and they were recommended an inappropriate movie or series that wasn’t suitable for their age. This could cause major issues for Netflix as it may lead to subscription cancellation as the parents might find that it is not an appropriate streaming service for their children. This could lead to Netflix losing out to a large sum of profits. To resolve this issue have included parental control on accounts which allows parents to be able to control the services that are available to their children. Parents will be able to use a 4 digit pin code that must be entered to play selected movies and TV shows that are above a selected age balance.