A FUZZY BASED APPROACH FOR MODELLING PREFERENCES OF USERS IN MULTI-CRITERIA RECOMMENDER SYSTEMS
Need help with a related project topic or New topic? Send Us Your Topic
DOWNLOAD THE COMPLETE PROJECT MATERIAL
A FUZZY BASED APPROACH FOR MODELLING PREFERENCES OF USERS IN MULTI-CRITERIA RECOMMENDER SYSTEMS
Chapter one
INTRODUCTION
This chapter provides a general introduction to the Recommender System, including the suggested system frameworks, issue description, study aim and objectives, research questions, and thesis format.
1.1 Background of the Study
The rapid growth of the Internet of Things (IoT) and the quick development of e-commerce websites have necessitated the urgent need for a recommendation system. Given the vast number of objects (products and services) available on these websites, users found it challenging to make the best option.
The tremendous increase and variety of information available on the Internet, as well as the quick development of new e-business services (selling items, product comparison, auctions, and so on), can overload users, causing them to make poor decisions.
As a result, the availability of options began to reduce users’ well-being rather than increase it. While choice is wonderful, extra choice is not always the best because it leads to information overload
which confuses the system’s user on the proper choice to make from the rising number of possibilities accessible, necessitating the implementation of a recommender system (RS).
Recently, RS has proven to be a beneficial tool for dealing with the information overload problem. Ultimately, an RS solves this phenomena by directing a user to new, untested items that may be relevant to the user’s current work.
1.2 Recommender System (RS)
Recommender Systems (RS) are software tools and strategies that suggest items that users are likely to be interested in. Recommendation is about predicting taste patterns and discovering new and attractive stuff.
A recommender system is a type of information filtering approach that aims to provide users with information regarding objects of interest (movies, music, books, news, web pages, and so on) (Meier, Pedrycz, & Portmann, 2013).
RS developed as a distinct research topic in the mid-1990s; it is primarily utilised in e-commerce websites as a technique to make recommendations to customers who lack proficiency in selecting a few goods from a large number of items on a certain website.
Amazon, Netflix, YouTube, Spotify, LinkedIn, and Facebook are some of the websites that use recommender systems. RSs are primarily intended for persons who lack the necessary personal expertise or aptitude to evaluate the potentially overwhelming amount of alternative things that a website may provide.
A nice example is a movie recommendation system that helps users choose a movie to watch. Netflix, a renowned movie recommendation service, uses an RS to personalise the online store for each user.
RSs can be personalised or impersonal. Non-personalized recommendations are common in periodicals and newspapers, although they are rarely addressed in RS research.
However, because recommendations are typically personalised, various users or user groups benefit from diverse, specialised ideas. Personalised suggestions are presented as sorted (ranked) lists of items provided by the user.
As a result, in order to perform ranking, RS tries to predict what the most suitable products and services are based on the user’s preferences and constraints, and to complete such a computational task
RS collects information from a user’s preferences, which may be explicitly expressed by the user through their ranking or browse history, or implicitly through simple site navigation.
The concept of RS arose from a basic observation that people typically rely on suggestions from others when making daily decisions. For example, it is normal to rely on what one’s friends recommend when choosing a book to read; similarly, school administrators rely on recommendation letters sent to students during the admissions process.
Furthermore, when deciding what movie to see, many read and rely on movie reviews and reviewers. To mimic this real-life scenario, the first RS was implemented using the Collaborative Filtering technique, which states that if an active user previously agreed with certain users, the other recommendations from these similar users should be relevant to the other active users. Tobergte and Curtis (2013) define data in RS as three types of objects: things, users, and transactions.
1. Items: An RS recommends items (movies, books, music, places of interest, and services) to a user.
2. Users: Users are those to whom products are directed. They are defined by their interactions with the RS.
3. Transactions: This is commonly referred to as a recorded interaction between a user and the RS, consisting of log-like data that stores essential information generated during the human-computer interaction and is useful to the system’s suggestion creation algorithm.
It could take the form of explicit or implicit user feedback, such as a rating for the selected item. Ratings are the most common type of transaction data that an RS collects. There are two types of RS rating systems: traditional (single rating) and multi-criteria RS. Single rating
Most Recommender Systems on the market use a single number rating to reflect user opinions on an item. The typical RS functions in a two-dimensional space of users and objects.
The utility of objects to consumers is typically represented by a completely ordered collection of ratings R0. Ratings can take various forms (Tobergte & Curtis, 2013), including numerical ratings (1-5).
Ordinal ratings, such as “strongly agree” or “strongly disagree,” encourage users to select the term that best represents their view on an issue.
Need help with a related project topic or New topic? Send Us Your Topic