Improving the ranking algorithm of App evaluation
mechanismAssignment 1
Ziyu zhang
ABSTRACT
There are certain problems in the App evaluation mechanism
of major App distribution platforms, such as Apple App Store
and Android App Store, Based on such problems, I plans to
improve the App evaluation mechanism of App store based
on analytic hierarchy process and sorting algorithm. By setting a set of standards, the effectiveness of the App evaluation
algorithm can be ranked. On this basis, I try to improve the
existing algorithm and verify its validity through the actual
effect.
Author Keywords
App evaluation; evaluation algorithms; App platform; rating
INTRODUCTION
Currently, some major App distribution platforms, such as
Apple App Store and Android App Store ratings of software
based on downloads, frequency of interaction with the app,
number and quality of ratings and comments, keyword density and relevance and Uninstalls. Although it basically satisfies the perspective of user experience, there are still many
drawbacks. It leads to the fact that many apps full of violence
and pornography are still released. Therefore, I plan to improve the App evaluation mechanism of App store based on
analytic hierarchy process and sorting algorithm of the comprehensive analysis rating method.
LITERATURE REVIEW
The comprehensive analysis rating method is a mathematical
model to calculate the comprehensive rating score according to the actual statistical data of the evaluated object. A
series of indicators are set to rank certain characteristics of
things, including weighted scoring method, efficiency coefficient method, membership function rating method and so
on. The general practice is to give or set the standard weight
of each specific indicator according to its different status in
the overall goal of rating, and at the same time determine the
standard value of each specific indicator, and then compare
the actual value of the indicator with the standard value to
obtain the score value of the level indicator, and finally summarize the score value of the indicator to obtain the weighted
evaluation total score. With the application of more mathematical techniques, Markov chains and stochastic models are
applied to the evaluation mechanism. Kadohira, a Japanese
Improving the ranking algorithm of App evaluation mechanism
researcher, needed to develop an iterative risk assessment
strategy for zoonotic diseases.[3] He used a combination of
risk analysis and analytic hierarchy process, which was completed by semi-quantitative analysis of existing public health
data. AHP data was collected by issuing questionnaires to
four stakeholder groups. The ranking of doctors, citizens,
researchers and government officials. App store App rating
mechanism is also based on this rating method, but each platform is different, which leads to each application platform has
its advantages and disadvantages in the rating mechanism of
App.
PROBLEM STATEMENT AND MOTIVATION
The scoring algorithm is to quantitatively evaluate a certain
type of subject by analyzing and calculating the existing data,
so as to realize the evaluation, assessment and management
of this type of subject. The practical significance of the scoring mechanism is not only a kind of ranking mechanism, but
more importantly the criteria behind the ranking mechanism,
namely, what criteria are used to make a reasonable evaluation on a series of things. In the current Internet, products
connect users, goods, content, services and service providers.
Each type of subject has a large amount of data, and the score
facilitates the unified assessment of these subjects by other
roles. At the same time, the use of ratings to quantify and
visualize the performance of multiple roles within the Internet is also conducive to the effective management of these
roles, since they are already automatically assigned. As consumers, ratings influence our spending choices all the time,
and our ratings become a source of data for the internal scoring systems of commercial companies. The App platform
also needs to evaluate the apps listed on the platform based on
certain standards, so as to judge whether their legitimacy and
rationality meet the requirements of public release. However,
based on the current observation and summary, there are certain problems in the App evaluation mechanism of major App
distribution platforms, such as Apple App Store and Android
App Store, which leads to the fact that many apps full of violence and pornography are still released due to the omission
of the evaluation mechanism.[5] This phenomenon will seriously damage the security of the application software market
and make the application platform itself lack of credibility.
[4] Based on such problems, I plans to improve the App evaluation mechanism of App store based on analytic hierarchy
process and sorting algorithm.
RESEARCH METHODOLOGY
I investigated the current mainstream application software
rating mechanism in the market, and took Apple, Android and
Amazon platforms as examples to analyze their rating mechanism for App legitimacy. Then, through the data collection
of App users’ real use feedback, it can be compared with the
evaluation mechanism based on this platform, so as to judge
the rationality of different evaluation algorithms. The evaluation algorithms of different platforms are compared, so as to
compare the efficiency of App rating of different platforms.
The evaluation algorithm of the mainstream App platform
was analyzed through consulting professional technicians and
literature review. On this basis, improvements were made
to obtain more accurate and reasonable evaluation algorithm.
APP software has characteristics that general software does
not have, so it is not possible to use general software evaluation method to evaluate APP software. APP user comments
are short and tendentious, but also random and free, which
brings difficulties to APP evaluation.[1] In order to solve the
above problems, this study attempts to propose a method to
filter APP user comments and extract words describing certain features of APP or expressions of users’ emotions. Meanwhile, an APP evaluation method is proposed based on the
extracted words. Finally, APP evaluation is carried out based
on the filtered user comments.[2]
Expected Outcomes and Contributions
The main purpose of this study is to study the rationality of
the rating mechanism of various application platforms.I will
try to find enough samples as experimental objects.Although
I will try to collect enough data and information and consult
enough literatures, some technical problems still cannot be
avoided. Because I cannot get the most in-depth information
of the algorithm engineers of these platforms, and because of
my lack of technical ability, I cannot fully understand the existing evaluation algorithms. This study mainly hopes to get
a comparison and analysis of the reasonable degree of evaluation algorithms for different App platforms. By setting a
set of standards, the effectiveness of the App evaluation algorithm can be ranked. On this basis, I try to improve the
existing algorithm and verify its validity through the actual
effect.
There are certain risks in this study, mainly intellectual property rights and information security risks. Because I try to investigate the application of equal internal information, it may
lead to unnecessary misunderstanding about the theft of confidential technical information. In order to avoid this problem, I will hold a reasonable degree when consulting relevant
algorithm engineers.
REFERENCES
1. Ning Chen, Jialiu Lin, Steven CH Hoi, Xiaokui Xiao,
and Boshen Zhang. 2014. AR-miner: mining
informative reviews for developers from mobile app
marketplace. In Proceedings of the 36th international
conference on software engineering. 767–778.
2. Kate Goddard and Jamie Erskine. 2019. OP58
Developing An Evaluation Based Taxonomy For
mHealth Apps. International Journal of Technology
Assessment in Health Care 35, S1 (2019), 14–15.
3. M Kadohira, G Hill, R Yoshizaki, S Ota, and Y
Yoshikawa. 2015. Stakeholder prioritization of zoonoses
in Japan with analytic hierarchy process method.
Epidemiology & Infection 143, 7 (2015), 1477–1485.
4. Hao Peng, Chris Gates, Bhaskar Sarma, Ninghui Li,
Yuan Qi, Rahul Potharaju, Cristina Nita-Rotaru, and Ian
Molloy. 2012. Using probabilistic generative models for
ranking risks of android apps. In Proceedings of the
2012 ACM conference on Computer and
communications security. 241–252.
5. Mohammed Talal, AA Zaidan, BB Zaidan, OS Albahri,
MA Alsalem, AS Albahri, AH Alamoodi, Miss
Laiha Mat Kiah, FM Jumaah, and Mussab Alaa. 2019.
Comprehensive review and analysis of anti-malware
apps for smartphones. Telecommunication Systems 72, 2
(2019), 285–337.