Framework For Recommender Systems Computer Science Essay

Published:

In this paper, we combine both content based and collaborative filtering approaches for providing advices to users about items they might wish to purchase or examine for finding approximate results on matching criteria for developing framework for recommender systems.

Using content based and collaborative filtering algorithms will provide results to the user and accepts feedback from him. As per the feedback it does changes in the rating of recommendations and accordingly gives the results.

This software will be useful in many applications for example book library, movie library, audio library, e-shopping products etc.

Keywords

Content based filtering, Collaborative filtering

1. Introduction

In this information age, searching is the key driver of the corporate and economic growth. People want everything to be done on their finger-tips from anywhere, anytime and that too very fast and precisely.

Recommender systems provide advice to users about items they might wish to purchase or examine. Recommendations made by such systems can help users navigate through

Lady using a tablet
Lady using a tablet

Professional

Essay Writers

Lady Using Tablet

Get your grade
or your money back

using our Essay Writing Service!

Essay Writing Service

Large information spaces of product descriptions, news articles or other items

These systems aggregate data about customers' purchasing habits or preferences, and make recommendations to other users based on similarity in overall purchasing patterns.

2. Detailed Problem definition

Basically, our system is a framework for recommender system. It can be used for any application. It will give choice to the administrator to add his applications and give recommendations to his clients. The administrator will be the one who wants this software, will decide his applications and also will provide software developer the whole information required to feed to the software. Mainly there are two algorithms used for recommender system. One is collaborative filtering another is content based filtering.

Collaborative filtering: It explores technique for matching people with similar interest and then making recommendation on this basis. Collaborative filtering works by building a database of preferences for products by consumers. A new consumer is matched against the database to discover neighbors, which are other consumers who have historically had similar taste to new consumer.

Content based filtering: Content based recommenders build on the intuitions "find me things like I have liked in past". It learns preferences through user feedback .Feedback may be of two kinds implicit and explicit.

Using these algorithms, will provide results to the user and accepts feedback from him. As per

the feedback it does changes in the rating of recommendations and accordingly gives the results. This software will be useful in many applications for example book library, movie library, audio library, e-shopping products.

3. Current Market Survey

According to the current market survey, the systems available are very typical. The systems are through and through only recommender systems that is the systems implemented till date are solely based on a particular type of application for example E-shopping, music Shoppe, book library etc. These systems can only suggest to the user the items they possibly can like but they don't have any particular system to take users feedbacks into the account after they have taken the recommended items. For these recommendations either they depend upon the other users or if the user is coming for the nth time they can use his own profile for recommending items. This can be done by using collaborative filtering algorithm or content based filtering algorithm. But there is no such system using both algorithms for giving recommendations.

4. Need of the system

The new millennium is an age of information abundance. The 1990s have seen an explosion of information and entertainment technologies, and thus of choices a person faces. People may choose from dozens to hundreds of television channels, thousands of videos, millions of books, CDs and multimedia, interactive documents on the World Wide Web and seemingly countless other consumer items presented in catalogs or advertisements in one medium or another. The web in particular offers myriad possibilities - in addition to interactive documents, there are conversations to join and items to purchase. Not only is there a vast number of possibilities, but they vary widely in quality. So what can we do? When people have to make a choice without any personal knowledge of the alternatives, a natural course of action is to rely on the experience and opinions of others. We seek recommendations from people who are familiar with the choices we face, who have been helpful in the past, whose perspectives we value, or who are recognized experts. We might turn to friends or colleagues, the owner of a neighborhood bookstore, movie reviews in a newspaper or magazine, or Consumers Union product ratings. And we may find the social process of meeting and conversing with people who share our interests as important as the recommendations we receive. Today increasing numbers of people are turning to computational recommender systems. These systems aim to mediate, support, or automate the everyday process of sharing recommendation.

5. Updations to the previous system

Lady using a tablet
Lady using a tablet

Comprehensive

Writing Services

Lady Using Tablet

Plagiarism-free
Always on Time

Marked to Standard

Order Now

The systems that are available are designed solely for the purpose of entertaining any single application under the recommender system for e.g. E-shopping, etc. But what is provided under the Para-Host Framework for Recommender System is the freedom of choosing any application for which Recommender system can be implemented. All the fields or parameters that are to be introduced into the system, that will be needed for giving recommendations can be dynamically changed or entered by the administrator. Two algorithmic approaches are combined to give the precise results namely collaborative filtering algorithm and content based filtering algorithm. After some recommended items are taken by the user the feedback is taken related to the recommended items, if the feedback is positive then that type of recommendations are continued by the system in the future also but somehow if the recommendations are not appreciated by the user then that type of recommendations are avoided in the future.

Purpose

Decision-making is an integral part of everyday life. When faced with a dilemma, most of us are likely to gather some relevant information before making an informed decision. Recommender systems provide one way of circumventing this problem. The main goal is to identify challenges and suggest new opportunities. People may choose from dozens to hundreds of television channels, thousands of videos , millions of books , CDs , and multimedia , interactive documents, and seemingly countless other consumer items presented in catalogs or advertisements in one medium or another.

Product Scope

Generally, content based filtering and collaborative filtering algorithms which we are going to use in our system, are used widely in many applications. In future, expansion of project can be done by implementing more algorithms such as particle swarm optimization, this will help for comparative study of algorithms and this framework can possibly be further used in more specialized applications to give much more desirable and precise recommendations according to the nature of the application and the personality of the user.

Overview

Requirements specification is organized in two major sections - Overall descriptions and Specific requirements .The first section describes the general factors that affect the product and its requirements. This section does not state specific requirements. Instead, it provides a background for requirements. This section of the SRS contains all of the software requirements to a level of detail sufficient to enable designers to design a system to satisfy those requirements, and testers to test that the system satisfies those requirements.

Overall Description

Basically, our system is a framework for recommender system. It can be used for any application. It will give choice to the administrator to add his applications and give recommendations to his clients. The administrator will be the one who wants this software, will decide his applications and also will provide software developer the whole information required to feed to the software. Mainly there are two algorithms used for recommender system. One is collaborative filtering another is content based filtering.

Product Properties

Existing system

Fig.1: Existing Recommender System

New system

Fig. 2: New Framework for Recommender System

Product Functions

The framework for recommender system will provide the following:-

Administrator can implement the framework for any random application.

Administrator can ask for the relevant information from the user as needed by the recommender system to recommend and also decides other specifications and give attributes.

On the basis of this information recommendation is given to the user.

After that the feedback regarding recommendations is taken from the user so as to check the satisfaction of the user and the precision of the given results to improve the future recommendations.

User Characteristics

There are two types of user's administrator and recommendation seeker. The administrator will enter the metadata and the master data along with the questions related to the personal and basic information that is to be filled by the recommendation seeker. The users would login and they will be asked various types of questions in order to help the system to understand the personality and the exact requirements of the user. System would be user friendly and will not need any typical or extra knowledge about computers to use the system.

External Interface Requirements

Lady using a tablet
Lady using a tablet

This Essay is

a Student's Work

Lady Using Tablet

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Examples of our work

The external interfaces requirements provide a description of all inputs and outputs associated with the framework of recommender system.

User Interfaces

The user interface is the mean of communication between the user and the software system. It has to be simple and easy to use and learn.

Software Interfaces

The software interfaces are nothing but the flow of screens of forms that are occurring one after the other when user is using the system.

System Features

Feedback is the main feature that will be provided in the system as an additional improvement. Feedback will be taken from the user based on his/her previous experiences which would decide the actual percentage of success achieved by the system. If the feedback is positive that means the recommender system is providing the correct suggestions, otherwise it will have to change the strategy and try to give more precise recommendations using the feedback obtained from the user, which in turn increases the efficiency of the system.

6. RESULTS

The framework for recommender is basically designed for the purpose of providing recommendations to the user among the large quantity of options .The framework is designed and developed in such a way that it can be implemented for any kind of application .All the parameters that are used to determine the recommendations are decided by the administrator himself and for providing such facility all the fields are kept dynamic i.e. it can be decided or changed by the administrator. And after deciding the application the master data (information of items related to the application), metadata (parameters related to the items), and training set is to be added to the administrator module. After this, the user that is using this system has to enter their profiles and then the recommendations will be provided on the basis of the algorithm namely, collaborative filtering and content based filtering. Collaborative filtering gives the recommendations on the basis of matching of neighboring profiles with the user who needs recommendations. The content based filtering algorithm is used only after the feedback of the user contains at least ten entries. Altogether the results of this system are the recommendations, the detailed display of liked and disliked items. Framework for recommender system contains various modules. On administrator side metadata module contains the format for master data. Master data is actual data accepting module. Questioner is module for accepting questions from administrator these questions will be then displayed to the user. Training set is module provided for giving training profiles to recommender system. Status part shows the information about current users present in the system.