Individual Assignment

TECHNOLOGY PARK MALAYSIA

CT098-3-2-RMCT
RESEARCH METHODS FOR COMPUTING AND
TECHNOLOGY
UC2F2102SE
HAND OUT DATE:
HAND IN DATE: 12th of Nov
INSTRUCTIONS TO CANDIDATES:
Complete the assignment and submit it online through Moodle. Your
assignment submission is integrated with Turnitin option for
plagiarism check.

NAME: / Munef Mohsen Mohammed AL GAONI

TP NO: / TP058796
Contents
Abstract…………………………………………………………………………………………………………………………………………………..2

1-Introduction: .............................................................................................................................................. 3
2- Literature review....................................................................................................................................... 4
3. Problem statement ..................................................................................................................................... 5
4. ................................................................................................................................................................... 6
Aim: ……………………………………………………………………………………………………………………………………………………6
Obj: - ......................................................................................................................................................... 6
5. Research Questions……………………………………………………………………………………………………………………………..6

6. Significance of the work ........................................................................................................................... 7

7. Methodology ............................................................................................................................................. 7
7.1 Questionnaire ...................................................................................................................................... 7
7.2 Data Analysis ...................................................................................................................................... 8
8. Overview of the Proposed System ........................................................................................................... 9
9. Conclusion ................................................................................................................................................ 9
10.References ............................................................................................................................................... 9

Key words: Recommendation system, recommender, music player

-Abstract.

We gift a brand-new method to endorse appropriate tracks from a set of songs to the
Listener. The aim of the system is to endorse songs that might be preferred via way of means of
the consumer, are fresh to the person’s ear, and match the person’s listening sample. We use
“Forgetting Curve” to evaluate the freshness of music and evaluate “favored Ness” through the
use of consumer logs. We examine a person’s listening sample to estimate the extent of the
hobby of the consumer withinside the subsequent music. Also, we deal with consumer conduct at
the music being performed as remarks to regulate the advice strategy for the following one. We
expand a utility to assess our method withinside the actual world. The consumer logs of trial
volunteers display excellent overall performance of the proposed method.

Key words: Recommendation system, recommender, music player

1-Introduction:
As we All understand that the community is evolving each day and for the reason that begin of
the twenty-first century, the net has been the number one supply of reclaiming multimedia data
understand as Music, books, and videos. People have considered that Music are a crucial difficulty of
their respective life, human beings generally have interacted with music while doing a pastime consisting
of driving, video games, and dancing. Past research has additionally marked, that people are extra into
songs more than anything they do during the day (Jones, 2003)(i.e., looking TV, analyzing books,
gambling video games, or maybe work). An incredible range of research has proven that tune is an
effective self-expression component in addition to communication.

Nonetheless, the problem these days is to control and control the tune that society has shared and
established, the several range that can't be treated to assist lessen those problems a few approaches has
been designed to resolve such problems consisting of artist identity and style classification. Moreover, to
assist customers to locate tune that fulfills their flavor a tune recommender has been invented. To
understand that the tune advice device is doing the needs, it's going to first permit the consumer
concentrate to a special kind of music, after which it's going to robotically pick out the consumer choice
of style artist or instrument. Meantime the development of recommender systems presents an outstanding
opportunity for corporations to aggregate the clients who are interested in tune. More importantly, it will
increase stressful conditions for us to better recognize to pick out the consumer flavor in tune.
(Uitdenbogerd, 2017)

Meanwhile, based totally at the flavor on listening for the customers and their score for a sure kind
of tune has proven that the collaborative filtering approach has proven an incredible result. This along
contest-primarily totally based version which permit the consumer locate tune that fit their flavor the use
of functions consisting of pitch, rhythm, and style. A clean instance the use of those strategies is Shazam,
AllMusic, and Spotify discover weekly. (Pachet, 2004)

2- Literature review

Various Music recommendation ideas had been evolved so far. We can categorize those tactics
in specific views. Automatically playlist era makes a specialty of recommending music which
are like selected seeds to generate a brand-new playlist. Rango furnished a technique to advise
track this is like selected cells as a playlist. Similarly, Flexor furnished a series of music to shape
an easy transition from a begin track until the cease track. These tactics forget about the person’s
comments while the consumer listens to the music withinside the playlist. They have an
underlying hassle that everyone seed-primarily based totally tactics giving a unique list of music
if the dataset includes masses of track cliques. While In iTunes, Bright employs comparable
techniques to generate a playlist from a cell (Rango, 2005). Dynamic track recommendation
improves automated playlist era via way of means of thinking about the person’s comments. In
the approach proposed via way of means of Papal, the playlist era begins off evolved with an
arbitrary track and adjusts the advice result primarily based totally on consumer comments. This
kind of approach is like Pandora. Collaborative-clear out techniques advise portions of track to a
consumer primarily based totally on the score of these portions via way of means of different
customers with comparable flavor (Flexer, 2008). However, collaborative techniques require
many customers and plenty of rankings and are not able to advise songs that don't have any
rankings. Hence, customers must be properly represented in phrases in their flavor in the event
that they want powerful advice. This precept has been utilized by diverse social websites, such as
Last.fm, my Strands. Content-primarily totally based techniques compute similarity among
songs, recommends songs just like the preferred songs, and gets rid of songs which are just like
the skipped songs. In a technique proposed via way of means of Cano, acoustic functions of
songs are extracted, along with meter, timbre, rhythms, and tempo patterns. Furthermore, a few
paintings express similarity in line with music emotion. Cai recommends a track primarily based
totally handiest on emotion. Hybrid tactics, which integrate track content material and different
facts, are receiving extra interest lately (Pampalk, 2005). Donaldson leverages each spectral
graph residences of item-based collaborative filtering in addition to acoustic functions of the
track signal. Shao et al. use each content material function and a person get admission to the
sample to advise track. Context-primarily based totally techniques takes context into attention.
Liu et al. take the alternate withinside the pastimes of customers through the years into attention
and upload time schedule to the track playlist. enhance collaborative filtering the use of
consumer grouping via way of means of context facts, along with location, motion, calendar,
surroundings situations, and fitness situations, whilst the use of content material evaluation
assists machines to choose suitable songs Kou and Shan (Cohen, 2011)evolved a content
material-primarily based totally track advice machine. In their machine, the customers’
possibilities are found out via way of means of mining the melody patterns, i.e., the pitch facts,
of the track they listened to. Chen and Chen proposed a track advice machine that hired 3 advice
mechanisms, i.e., content material-primarily based totally approach, collaborative filtering and
statistics-primarily based totally approach, for specific customers’ needs. In their machine, track
gadgets are grouped in line with the residences along with pitch, period and loudness, and
customers are grouped in line with their pastimes and behaviors derived from the get admission
to histories. (Cano, 2005)

3. Problem statement

One of the primary issues of recommender structures in general, and tune recommender
structures in particular is the bloodless begin hassle, i.e., whilst a brand-new person registers to
the machine or a brand-new object is delivered to the catalog and the machine does now no
longer have enough facts related to those gadgets/customers. In this kind of case, (Gupta, 2020)
the machine cannot nicely advise current gadgets to a brand-new person (new person hassle) or
advise a brand-new object to the present customers (new object hassle). Another subproblem of
bloodless begin is the sparsity hassle which refers to the truth that the range of given scores is
lots decrease than the range of feasible scores, that's mainly likely whilst the range of customers
and gadgets is large. The inverse of the ratio among given and feasible scores is referred to as
sparsity. High sparsity interprets into low score coverage, considering that maximum customers
tend to price simplest a tiny fraction of gadgets. The impact is that hints regularly end up
unreliable (Zamani, 2018). Typical values of sparsity are near 100% in maximum real-global
recommender structures. In the tune domain, that is a mainly tremendous hassle. Dorr et al. for
instance, analyzed the Yahoo! Music dataset, which as of time of writing represents the most
important tune advice dataset. They document a sparsity of 99.96%. For almost 10000
International Journal of Multimedia Information Retrieval (2018) 7:95–116 comparison, the
Netflix dataset of films has a sparsity of “simplest” 99%. (Li, 2005)

4. Aims and Objective

Aim:

The objective of this project is to develop a music recommendation system. The system will determine
the musical preferences of users based on the analysis of their interaction during use

Obj: -

Create a music recommendation system able to infer the users’ musical preferences in a given time. The
scope is not to know the user; instead, it’s about estimating what he could like right now.
- Explore the music services available nowadays looking for a complete and freely accessible music
catalog and free streaming services.
- Develop a working system capable of making the most of free online services to provide the user with a
completely free system which brings the opportunity of discovering new music.
5. Research Questions

- What is the best Technique to provide the best experience when recommending the music?
- How can the profile learning reflect on enhancing the system recommendation?
- in what way can graph based algorithms help the system?
- what age and knowledge the system is aiming?

6. Significance of the work

As the internet track, streaming will become the dominant medium for human beings to concentrate to
their favorite songs, track streaming offerings are actually capable of gather massive quantities of
information at the listening conduct in their customers. These streaming offerings, like, Apple Music,
Pandora, or Spotify, are the use of this information to offer pointers to their listeners. These track advice
structures are a part of a broader elegance of recommender structures, which clear out statistics to are
expecting a user’s possibilities in terms of a positive item. Think Netflix film pointers or Pandora. This
exceptional Wikipedia article on the subject splits’ recommender structures into classes so that it will
additionally observe to track unique recommender structures. These classes, or approaches, to
recommender structures, are Collaborative Filtering and Content-Based Filtering. (Srebrenik, 2018)

7. Methodology

7.1 Questionnaire

The reason for the questionnaire is to achieve records and records from people who are interested
in music to have an idea of the project will have approximately their evaluations on the use of
cellular software in their music environment. The flowchart for this technique may be mentioned
in the Appendix. This technique is selected due to the fact it may offer a dependable manner of
amassing feedback from respondents (McLeod, 2018). It additionally gives researchers
applicable and correct records after considerate designing of questions. For this technique, a
pattern length of minimal two hundred users might be selected as respondents via random sampling.
According to Roscoe’s Rules of Thumb, no less than one hundred respondents’ feedbacks need to be
accumulated so one can lower the margin blunders to 10%, to grow the self-belief stage among 85% and
95%, in addition, to ensure the reaction price of 50% is received. The strategies selected to distribute the
questionnaire is to apply on-line surveys furnished with the aid of using Teams due to the fact it's miles
handy for the researchers to ship many respondents wherein the attain and scalability of the usage of the
net survey is huge. Furthermore, with the aid of using the usage of on-line surveys, it's miles handy for
respondents to finish the survey at their very own pace. The on -line survey could have an aggregate of
open-ended and closed-ended questions wherein respondents might be given a couple of preference
questions, rating-styled questions, and brief subjective questions (Aryal, 2019 ). However, there are issues
concerning questionnaires that need to be taken into consideration while the usage of this technique. For
example, the anonymity of the respondents in addition to the layout of the questions needs to be
cautiously taken into consideration.

7.2 Data Analysis

When the questionnaires were gathered, tabulation and evaluation of the records gathered must
be executed for the project to interpret the records in an organized manner. The technique
selected for this study is via way of means of the usage of Microsoft Excel to tabulate and
examine the records. The rationality of selecting Microsoft Excel is as it carries the fundamental
gear for the people involved to sort, clear out and validate the records gathered from the
questionnaires. Besides that, Microsoft Excel offers a formulation for calculating percentages,
totals, counts, medians, popular deviations, and plenty of more. After analyzing the records
gathered, graphs which include pie charts and bar graphs may be used to symbolize the records
gathered to offer the records in a graphical manner. (Warren, 2020)
8. Overview of the Proposed System

In this system we will going to have a music recommendation system which interact with the
human behavior to have the best result to satisfy the user , choosing the best result can not be
easy , but with this system it will be possible since it the start it will show the user a set of
categories from their the system can identify which type of music or instrument from the
preference the user has chosen which will make the operation easier, giving the user the
opportunity to give the preference at the start of the system will attract the user to use it more and
more and will have a set of songs that appear every week with his/her preference.

9. Conclusion
To conclude everything that has been stated, this music recommendation system can offer a great
experience to the user who is especially engaged to music to have an opportunity express
themselves in the music they like with the help of this system which will be implemented in the
future everything will be possible.

10.References
Aryal, S. (2019 , March 17). Microbe . Retrieved from https://microbenotes.com/questionnaire-method-
of-data-collection/

Cano, P. (2005). strength content-based music recommendation system. ACM SIGAR, 673.

Cohen, W. (2011). Web Collaberation. Computer Network, 685-698.

Flexer, A. (2008). Playlist generation. ISMIR, 173-178.

Gupta, S. (2020, May 21). Sameeksha Gupta. Retrieved from
https://medium.com/@sameeksha19096/music-recommendation-using-relevance-feedback-
6343d562ae7a

Jones, P. (2003). Thestructure and personality correlates of music preferences. Samuel D. Gosling, 1236-
1256.

Li, Q. (2005, Jan 5). ResearchGate. Retrieved from

https://www.researchgate.net/publication/221055539_A_Probabilistic_Model_for_Music_Reco
mmendation_Considering_Audio_Features

McLeod, S. (2018, May 2). Simply Psychology. Retrieved from

https://www.simplypsychology.org/questionnaires.html

Pachet, F. (2004). Improving Timbre Similarity. How High is theSky?, 1-13.

Pampalk, E. (2005). Dynamic playlist generation . ISMIR, 634-637.

Rango, R. (2005). “Inferring similarity. ACM Multimedia , 73-80.

Srebrenik, B. (2018, Dec 4). Brian Srebrenik. Retrieved from

https://medium.com/@briansrebrenik/introduction-to-music-recommendation-and-machine-
learning-310c4841b01d

Uitdenbogerd, A. (2017, Oct 17). Recearch Gate. Retrieved from

https://www.researchgate.net/publication/220723115_A_Review_of_Factors_Affecting_Music_
Recommender_Success

Warren, K. (2020, Jan 5). GRADCOACH. Retrieved from https://gradcoach.com/qualitative-data-analysis-

methods/

Zamani, H. (2018, May 25). Springer Link . Retrieved from

https://link.springer.com/article/10.1007/s13735-018-0154-2