AI Powered Search Engine

PROJECT REPORT – PHASE I

Submitted in partial fulfillment of the requirements for the award of

Bachelor of Engineering degree in Computer Science and Engineering with
specialization in Artificial Intelligence and Machine Learning

By

Mohamed Hisham (Reg. No – 41611115)

Mrudhula .M (Reg. No –41611119)

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

SCHOOL OF COMPUTING

SATHYABAMA
INSTITUTE OF SCIENCE AND TECHNOLOGY
(DEEMED TO BE UNIVERSITY)
CATEGORY - 1 UNIVERSITY BY UGC
Accredited “A++” by NAAC I Approved by AICTE
JEPPIAAR NAGAR, RAJIV GANDHI SALAI, CHENNAI - 600119

AUGUST - 2024
 DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

BONAFIDE CERTIFICATE

This is to certify that this Project Report is the bonafide work of Mohamed Hisham (41611115) who
carried out the Project entitled “AI Powered Search Engine” under my supervision from June 2024
to August 2024.

Internal Guide
Dr. R.Shalini, M.E.,(Ph.D.,)

Head of the Department

Dr. S. VIGNESHWARI, M.E., Ph.D.,

Submitted for Interdisciplinary Viva Voce Examination held on

Internal Examiner External Examiner

ii
 DECLARATION

I, Mohamed Hisham (Reg. No- 41611115), hereby declare that the Project Report entitled “AI
Powered Search Engine” done by me under the guidance of Dr. R.Shalini, M.E.,(Ph.D.,) is submitted
in partial fulfillment of the requirements for the award of Bachelor of Engineering degree in Computer
Science and Engineering with Specialization in Artificial intelligence and Machine Learning .

DATE:

PLACE: Chennai SIGNATURE OF THE CANDIDATE

iii
 ACKNOWLEDGEMENT

I am pleased to acknowledge my sincere thanks to Board of Management of Sathyabama

Institute of Science and Technology for their kind encouragement in doing this project and
for completing it successfully. I am grateful to them.

I convey my thanks to Dr. T. Sasikala, M.E., Ph. D., Dean, School of Computing, and Dr. S.
Vigneshwari, M.E., Ph.D., Head of the Department of Computer Science and Engineering
with Specialization in Artificial intelligence for providing me necessary support and details at
the right time during the progressive reviews.

I would like to express my sincere and deep sense of gratitude to my Project Guide Mr.
R.Sundar, M.E.,(Ph.D.,)for her valuable guidance, suggestions, and constant encouragement
paved way for the successful completion of my project work.

I wish to express my thanks to all Teaching and Non-teaching staff members of the
Department of Computer Science and Engineering with Specialization in Artificial
intelligence and Machine learning who were helpful in many ways for the completion of the
project.

iv
ABSTRACT

Today’s search engines focus mainly on the static content of the web. Since more and more websites are being
created with dynamic pages, i.e. with server side scripts that queries a database and creates web pages “on the
fly” based on some input criteria, this is a problem that is growing rapidly. This project consists of designing,
developing and evaluating a search engine for dynamic data in the form of second hand products available on
the Internet in Sweden. Another part of the project is to improve the quality of the search results by equipping
the search engine with artificial intelligence. This report describes and compares different search engines and
the technologies behind them. It gives a thorough description of the components in the system and motivates
the choice of solution and platform.

9
CHAPTER 1

INTRODUCTION:

In the realm of academic research, staying abreast of the latest developments and insights within a given field
is essential. Literature reviews are an essential part of academic research, but they can also be time-consuming
and tedious. Conducting comprehensive literature reviews has long been a cornerstone of scholarly inquiry,
but the sheer volume of available research materials today poses a daunting challenge for researchers seeking
to identify relevant sources efficiently. Searching for relevant papers, reading through them, extracting key
information, and synthesizing the findings are all tasks that require a lot of manual effort and attention.

9
Sometimes it is not the dearth of materials that pose a problem but the myriad materials available which can
lead to information overload, thereby preventing a researcher from knowing where to start the search from,
what to choose from the available options which are in millions and what is actually relevant to one’s
research. Conducting a literature review the traditional way could sometimes take weeks if not months to
accomplish. However, with the advent of AI-powered search engines, researchers can now access and analyze
millions of peer reviewed papers in a matter of seconds (Baviskar et al, 2021). These tools use natural
language processing and machine learning to understand the user's query, find the most relevant papers, and
provide evidence-based answers. They can also generate summaries, analyses, and citations for the papers,
making the literature review process more efficient and accurate (Anton et al., 2021). This is where the
revolutionary capabilities of AI-powered search engines come into play. As seasoned researchers, we are
witnessing a paradigm shift in how we approach literature reviews, with AI-driven tools offering the potential
to streamline the process, unearth hidden gems of knowledge, and provide deeper insights than ever before
(Heidt, 2023). The time for search is cut short and more relevant literature is right on a researcher’s fingertips.
This makes the researcher much more efficient and effective in carrying out research rather than having to
spend a lot of time on searching for relevant literature. In this exploration, this article delves into the
transformative impact of leveraging AI-powered search engines for literature review, examining how these
cutting-edge technologies can enhance the precision, comprehensiveness, and efficiency of our information
discovery and synthesis endeavors in the academic world.

9
CHAPTER 2

LITERATURE SURVEY
1. Omobolanle Seri Fasola Ajayi Crowther University, Oyo, Nigeria, “HARNESSING
ARTIFICIAL INTELLIGENCE-POWERED SEARCH ENGINES FOR THE LITERATURE
REVIEW PROCESS”
Artificial intelligence-powered search engines are specialist information retrieval tools made for academic
and research use. In order to assist researchers and academics in quickly exploring and discovering
scholarly papers, academic articles, and pertinent content within the academic domain, these engines make
use of artificial intelligence and machine learning techniques (Wagner et al., 2022). They are designed to
recognize context, comprehend the subtleties of academic search queries, and deliver exact, contextually
appropriate results from enormous academic literature libraries. In order to facilitate more efficient and
thorough literature reviews in academic and research contexts, AI-powered search engines for literature
reviews optimize the research process by suggesting pertinent sources, summarizing key findings, and
helping to organize research materials.

2. Rajab Ssemwogerere, Assadig Abdelrhman Sajo, Nambobi Mutwalibi, Asha Khamis Mzee
“A Survey About the Application of Artificial Intelligence in Search Engines”
Artificial intelligence (AI) mimics or stimulates human behaviors or thinking to solve specific
problems. It has been applied in the analysis of huge datasets and provides reliable outputs without
human supervision in various online platforms, for example, information retrieval in search engines,
digital assistants, voice assistants, digital marketing, personalized learning, social media, etc. This
technology has provided many opportunities and challenges in line with strengthening the authenticity
of the information provided via different search engines.
3. Yodhi Yuniarthe Application of Artificial Intelligence (AI) in Search Engine “Application of AI
in Search Engine”
AI development at present, although it has been quite advanced, but it is considered by most observers
are still far from the goal to replace humans. In this effort, proposed a general theory of artificial
intelligence . The general theory of artificial intelligence developed is more oriented to the classifier
models and statistical learning. This theory consists of a number of axioms, among others: analogous to
the nervous system of man, the basic functional unit of the nervous system of intelligent systems is the
artificial neurons (neural-like units), all data-free neural-like elements are novel neural-like units, all
neural-like elements that carry certain pieces of information are equivalent neural-like elements, at the
lack of information on the receptors of the novel neural-like elements they continue in the mode of
light arbitrary background excitation, and background excitation is a fluctuating arbitrary excitation
value of the neural-like element. Claims that the exploration into this theory will allow the study of
artificial intelligence bring in various imitation of human behavior, including motivation, purposeful
behavior, thought, awareness, personality, and 9
so on.
 4. Manas Ranjan Mohapatra , Saismita Mohapatra , Jyoti Ranjan Mohanty “Artificial Intelligence
(AI)’s Role in Search Engine Optimization”

Now-a-days Artificial intelligence has been applied into various areas of human daily life today. It is used in
education, health, investment , court purpose, cyber security and transportation. Artificial intelligence has been
useful for people to obtain reliable and their required information on the Internet through optimization on
search engines. Results of SEO Proficiency in Marketing Survey showed that 96% SEO experts states that
SEO strategy can benefit more if associated with AI, 73% agreed that AI can analyze the data for SEO needs,
61% said that AI can help locate anomalies in their SEO strategy and fix it, 58% believe that AI allows their
SEO strategy to beat its competitors, 49% even think that AI can beat Google RankBrain technology, and 41%
considered that AI eliminates the need for exact phrases as keywords.

5. Mr. Markus Anttila “RESEARCH ON SEARCH ENGINE STRATEGY “

With the rapid development of Internet search engine technology, it not only changes the way how search
engines develop their algorithms, but it also changes how people are able to gather the information they need
online. The digital marketing field is constantly evolving with new trends coming and going. Voice search is
one of the most promising trends surfacing the digital marketing landscape within the last 10 years. These days
most of the search engine queries are done by typing in the search using keywords. However, no matter how
advanced technology gets, the best communication tool for humans will always be their voice. Humans are
born with the skill to communicate with words produced with their voices

6. Joan Ofulue1 · Morad Benyoucef “ Data monetization: insights from a technology-enabled AI”

Data is one of the most valuable resources in the world. The new data economy has led to several data use
cases and one of them is monetization. Given how recent the concept of data monetization is, this study aims
to investigate it, better understand its meaning, implications and issues in the academic literature and provide
guidance for practitioners as well as direction for further research. A systematic literature review is conducted
on English literature from 2013 to 2022. 54 articles were identified from where a topic and sub-topic
categorization and a conceptual framework is developed. The paper makes contributions in three areas:
providing a holistic understanding of data monetization models through the extension of a framework from the
literature, categorizing of the key topics and trends in data monetization, and using a systematic approach to
identify managerial implications and a future research agenda.

9
 7. Ajay Ashok Mahajan “ EVOLUTION OF SEARCH ENGINES AI ”

The data collection process will involve an extensive search of relevant scholarly articles, conference papers,
reports, and other authoritative sources published between 2015 and 2022. The search will encompass
electronic databases and academic search engines, as well as manual searches on the websites of major search
engine companies, industry reports, and relevant government publications. The selected sources will be
critically reviewed, and pertinent information related to the research objectives will be extracted for analysis.

8. Kalyani Wagaj , Prof.Subhash Pingale “ Keyword Query Processing in Web Search Engines”

Search engine algorithms take the key elements of a web page, including the page title, content and keyword
density, and come up with a ranking for where to place the results on the pages. Each search engine’s
algorithm is unique, so a top ranking on Yahoo! does not guarantee a prominent ranking on Google, and vice
versa. The number of potential results may increase exponentially with the number of sources and links
between them. Yet, most of the results may be not necessary especially when they are not relevant to the user.
The routing problem, we need to compute results capturing specific elements at the data level. Routing
keywords return all the source which may or may not be the relevant sources.

9. Aldo Faisal Umam “ Impact of Artificial Intelligence on Search Engine Strategies”

This study examines the impact of artificial intelligence (AI) on search engine optimization (SEO) strategies.
In recent years, search engines have undergone significant changes in the use of AI technology to improve the
quality of search results. Although the implementation of AI technology in SEO is still relatively new, research
has shown that the use of this technology can help improve the quality of search results and provide a better
search experience for users. This article discusses how AI technology can help improve the quality of search
results through content optimization, the development of more sophisticated search algorithms, and the use of
machine learning and natural language processing techniques to understand the search context and provide
more relevant results. In addition, this article also discusses the implications of using AI technology in SEO
and the challenges that may be faced in implementing this technology. In this study, the research method used
is a literature review, by collecting and analyzing data from academic sources and related publications. By
examining the literature, this study highlights the potential benefits and limitations of using AI technology in
SEO, and provides insights into the future direction of search engine optimization strategies. Overall, this
article contributes to the growing body of knowledge on the role of AI technology in improving search engine
optimization strategies, and provides valuable insights for researchers, practitioners, and industry professionals
in this field.

9
10. BALAMURUGAN S, SURIYAVARMAN D, CHELLAIYA A, “ A Study on Contemporary Issues
in Ai Search Engine

AI search engines are a type of search engine that uses artificial intelligence techniques such as machine
learning, natural language processing, and predictive analytics to provide more accurate and relevant search
results. These search engines are designed to understand the user's intent behind a search query and provide
results that match that intent, even if the query is phrased in natural language rather than using specific
keywords. One of the primary advantages of AI search engines is their ability to provide more personalized
results. By analyzing user behavior and feedback, they can learn about the user's preferences and provide
search results that are tailored to their interests. However, there are also some challenges associated with AI
search engines. They require a significant amount of data to be effective, and biases in the data can lead to
biased search results. There are also concerns about privacy and the use of personal data to train the algorithms
used by AI search engines

Existing System & Proposed System

 EXISTING SYSTEM:

1. Crawling: Web crawlers scan and index web pages to create a massive database of web content.

2. Indexing: Indexed data is stored in a massive database, making it searchable.

3. Query Processing: User search queries are processed using algorithms to match keywords and retrieve
relevant results.

4. Ranking: Results are ranked based on relevance, authority, and other factors to display the most accurate
results.

Limitations of Existing System:

1. Keyword-based search: Relies heavily on exact keyword matches, which can lead to irrelevant results.

2. Lack of context understanding: Fails to understand user intent, context, and nuances of natural language.

3. Information overload: Returns a vast number of results, making it difficult for users to find relevant
information.

4. Outdated information: May return outdated or irrelevant results due to slow indexing and updating
processes.

9
PROPOSED SYSTEM

1. Advanced Crawling: AI-driven crawlers scan and index web pages, identifying relevant content, and
updating the index in real-time.

2. NLP-based Query Processing: User search queries are analyzed using NLP to identify intent, context, and
nuances, enabling more accurate results.

3. Machine Learning-based Ranking: ML algorithms rank results based on relevance, authority, user
behavior, and context, providing personalized results.

4. Knowledge Graph Integration: The search engine utilizes a vast knowledge graph to connect entities,
providing more accurate and relevant results.

5. Entity Recognition and Disambiguation: AI-powered entity recognition identifies and extracts specific
entities, disambiguating them to provide accurate results.

6. Context-Aware Search: The search engine understands user context, including location, search history, and
preferences, to provide personalized results.

7. Real-time Indexing and Updating: The index is updated in real-time, ensuring users receive the most
recent and relevant results.

8. User Feedback and Adaptation: The search engine adapts to user behavior and feedback, continually
improving results and user experience.

Advantages:

1. Improved Accuracy: More accurate results due to NLP, ML, and knowledge graph technologies.
2. Personalized Search: Context-aware search provides personalized results, improving user experience.
3. Real-time Relevance: Real-time indexing and updating ensure users receive the most recent results.
4. Enhanced User Experience: User feedback and adaptation enable continuous improvement

9
System Architecture:

URL Crawlers Store

Server Server

Indexer Repositor
Databa y
se

Searcher
System Flow Diagram:

Robots Store
Servlet Databa
se

Agents Cach Searcher

e

User Web Servlets

Server
Class Diagram
 Activity Diagram:

HashList

The main container for all ads needs to be as optimized as possible because it will be the most
frequent used container. I developed an own class for this called ‘HashList’. It has the features of
both a hash table and a linked list. A hash table is fast when you want to look up a specific object,
but slow for iterating through its objects, and vice versa for a linked list. Also, if you iterate
through a hash table, there is no guarantee in which order the objects will appear, while iteration
through a linked list will return the objects exactly as they were inserted into the list.

The HashList contains a built in HashMap and a special designed linked list. The linked list
contains objects (ads) with references to the next and previous object in the list. I did some
benchmarks where I compared the HashList with the native Java classes ‘HashMap’ and
‘LinkedList’. In this benchmark I created 500.000 objects and iterated through them. The
following table shows the allocated memory needed for the classes to hold the objects (excluding
the memory allocated for the objects themselves), and the time taken for a complete iteration
through the objects.

Class Memory (MB) Time (ms)

HashList 26,1 63
LinkedList 18,2 125
HashMap 13,8 156

The total memory if you combine the two native classes will be 32 MB. Since the HashList
contains both a hash table and a linked list it outperforms the two native classes in both speed and
memory consumption. The reason for this is that I could avoid much of the overhead in the native
classes by building my own class very slim; I only implemented the things needed in a way that
was optimal for the usage in this project.

Another very big issue with the HashList is that I managed to make the iteration of it safe and
unsynchronized even in a multi threaded environment. This is impossible for the native classes
since they will automatically throw a ConcurrentModificationException if a write operation
occurs while another thread iterates through the list if it isn’t synchronized. I utilized the garbage
collector for this. deleted. By changing the references for a removed object in such a way that the
object keeps its references and all references to the object are removed, the object will remain in
memory as long as some thread have a reference to it, and thus, if a thread should be in the
middle of an iteration at the same location as this object, the object will not be deleted since the
iterating thread will have a reference to it, and the iterating thread will be able to continue its
iteration since the object has kept its references to the next object in the list. The write
operations of the HashList need to be synchronized though, since two or more write operations
cannot occur simultaneously because that could make the references point in an undetermined
way.
Reduce the size of the result page

No matter how good your hardware is or how much you have optimized the search engine, in the
end what you will send to the end user is a page with the search results. If the end user has low
bandwidth, e.g. a telephone modem, every kilobyte you can reduce from the result page will have
a great impact on the time it will take to download the page. For instance, I managed to reduce
the result page from 50kB to just 7kB, which means that the load time for a user with a 56k
modem was reduced from seven seconds to just one second.

I did this by storing as much static data as possible on the client side and only send the actual
dynamic data. I stored the static data in a JavaScript file that is downloaded the first time a user
visits the page, and then it will be cached on the user’s hard drive. I use the same JavaScript to
build up dynamic HTML pages from the data which is sent from the server. For example, the
result page will just contain calls to a JavaScript function which will produce the HTML code
necessary to build up HTML table structures, etc. This requires some extra processing on the
client’s machine, but that delay is insignificant compared to the download time, especially if the
user has a somewhat modern computer, then this delay is not even noticeable.
Analysis

Evaluation testbed

For the tests I used the configuration that is described in section 3.2. The system held about
214.000 ads at the time when the tests were performed. I tested the functionality, the performance
and accuracy of the search engine and how it behaved under heavy stress. I performed four
different tests to evaluate the different aspects of the search engine:

Functionality test
This was the actual “bug” test which checked that everything works as it should. The test
consisted of creating, changing and removing ads and agents, monitor that the robots collect new
ads correct, that the AI removes duplicates correct, that the agents give correct hits, and that the
search engine gives correct results.

Performance test
This test checks how fast the search engine can produce different search results. I conducted
different types of searches and measured the time between receiving a search query by the search
engine and producing a complete result set.

Accuracy test
This test checks how well the search engine gives results for advanced Boolean searches. I also
compared searches on this search engine with corresponding searches on other search engines
whose ads are indexed by this search engine.

Stress test
In this test I checked how the search engine behaved under heavy stress. I simulated online users
with threads that executed searches.
Results of evaluation

Functionality test

The ads and agents were created/changed/removed correctly. To check the AI I let the robots run
for 24 hours, which produced 5272 new ads. The AI classified 583 of those ads as duplicates, i.e.
currently indexed in the system from some other site. The AI writes the URL of every new ad that
is classified as a duplicate into a log file, along with the URL of the ad that was found to be its
duplicate. I visited each of these URLs manually and compared the ads, and I’m proud to say that
every one was correctly classified as a duplicate.

Performance test

The database held 211.448 ads when this test was performed. The JVM measures time in units of
tens of milliseconds, i.e. time 0 means that the search took less than ten milliseconds to complete
and time 10 means that the search took between: 10 <= time < 20 ms.

Keywords Criteria Hits Time (ms)

Saab - 4.421 0
Saab AND 9000 AND Turbo - 269 0
Saab NOT 9000 NOT Turbo - 2.239 10
Saab OR 9000 OR Turbo - 4.421 20
Volvo - 11.374 0
Volvo AND 850 AND GLT - 220 0
Volvo NOT 850 NOT GLT - 9.490 30
Volvo OR 850 OR GLT - 11.374 50
- Stockholm 50.789 90
- Computers in Stockholm 1.183 90
 - Cars for sale in Stockholm with 4.139 90
picture
Ford AND Escort Cars for sale in Stockholm with 48 0
picture

As seen by the results, keyword searches are extremely fast. They’re also not affected by the
number of keywords. Keywords with the AND operator are the fastest because they will reduce
the result set very rapidly. Keywords with the NOT operator are a bit slower because they will not
reduce the result set as significant as the AND operator, and thus, the time will depend much on
the size of the starting result set, i.e. the set before the NOT words are removed from the set. The
Boolean operator OR is the slowest. The reason for this is that such a search result will not only
be sorted by date but also by best possible hit, i.e. the ads that contain most OR words will be at
the highest position in the search result list. For this reason, the time will depend on the size of
the result set.

When a criteria search is executed, the whole list is iterated and each ad is checked if it fulfills the
criteria. This takes about 90ms to complete regardless of how many criteria that are submitted.
If a keyword is added to a criteria search, the search engine will first execute a keyword search
and then remove the ads that doesn’t fit the criteria from the result, and hence, the time will be the
same as a pure keyword search.

Accuracy test

Keywords Criteria Hits Sites

Nissan AND 200sx Located in Stockholm 16 4

This test gave 16 hits distributed among 4 different sites. The first site gave 7 hits, the second 1
hit, the third 7 hits and the fourth 1 hit. A search on the first site only gave 3 hits, which all were
hits among these 7. It turned out that this site’s search engine missed the other 4 ads because they
had a space between 200 and sx. The second site gave 1 hit according to the one that was found.
The third site gave 6 hits, which all were among the 7 found. It missed the 7 th ad for the same
reason as the first site. The fourth site gave 1 hit according to the one that was found.

Result: 100% correct hits with even better accuracy than the original sites; it found 5 ads that the
original search engines had missed.
Keywords Criteria Hits Sites
T610 Phones for sale in Gothenburg 23 7
area
This test gave 23 hits distributed among 7 different sites. The first site was this local site and gave 4 hits, the
second 14 hits and the other 5 sites had one hit each. No need to examine the local hits. The second site gave
11 hits, where 10 of these were according to the ones that were found. The 3 other ads were missed by the
original search engine because the ads had text like T610i and T- 610. The ad that was missed was about 2
months old. According to the logs, it had been replaced by a duplicate ad located on another site. This ad had
had later been removed from that site, hence the ad didn’t turn up in the search result. The other sites gave hits
according to the ones that were found.

Result: 96% correct hits. It also found 3 ads that the original search engines had missed. The ad that was
missed was due to a duplicate that had been removed from the original site, which is a scenario that can
happen, and thus no bug.

Keywords Criteria Hits Sites

Porsche AND 911 Cars 43 6

This test gave 43 hits distributed among 6 different sites. The first site gave 29 hits, the second gave 8 hits, the
third gave 2 hits, the forth and fifth gave 1 hit and the sixth gave 2 hits. The first site gave 26 hits, all
corresponding to the ones that were found. The ones that the original search engine missed had text like
‘911/964’, ‘911T’ and ‘911sc’. The second gave all correct hits. The third didn’t support the AND operator (it
defaulted to OR), but from the 6 hits that were found with an OR search, the 2 ads that were found were also
the ones that had both the words Porsche and 911.
The remaining search engines gave correct hits according to the search result.

Result: 100% correct hits. It also found 3 ads that one of the original search engines had missed.

PROJECTIMPLEMENTATION

Project Implementation Technology

The Project is designed and developed in Django Framework. We used Django Framework for
coding of the project. Created and maintained all databases into MySQL Server, in that we create tables,
write query for store data or record of project.

 Hardware Requirement:
  Processor–Corei3

 HardDisk–160GB

 Memory–4GB RAM

 Monitor

 Software Requirement:

 Windows7or higher

 Python

 Djangoframework

 MySQL database
 OVERVIEWOFTECHNOLOGIESUSED

INTRODUCTION

Python is a powerful multi-purpose programming language created by Guido van Rossum. It has
simple easy-to-use syntax, making it the perfect language for someone trying to learn computer
programming for the first time. This is a comprehensive guide on how to get started in Python, why you
should learn it and how you can learn it. However, if your knowledge of other programming languages and
want to quickly get started with Python. Python is a general-purpose language. It has wide range of
applications from Web development (like: Django and Bottle), scientific and mathematical computing
(Orange, SymPy,NumPy) to desktop graphical user Interfaces(Pygame, Panda3D). The syntax of the
language is clean and length of the code is relatively short. It's fun to work in Python because it allows you
to think about the problem rather than focusing on the syntax.

TESTING

As the project is on bit large scale, we always need testing to make it successful. If each component
work properly in all respect and gives desired output for all kind of inputs then project is said to be
successful. So the conclusion is-to make the project successful, it needs to be tested.

The testing done here was System Testing checking whether the user requirements were satisfied.
The code for the new system has been written completely using python as the coding language, Django as
the interface for front-end designing. The new system has been tested well with the help of the users and all
the applications have been verified from every nook and corner of the user.
Although some applications were found to be erroneous these applications have been corrected before being
implemented. The flow of the forms has been found to be very much in accordance with the actual flow of
data.

Levels of Testing

In order to uncover the errors, present in different phases we have the concept of levels of testing. The
basic levels of testing are:

Client Needs Acceptance Testing

Requirements System Testing

Design Integration Testing

Code Unit Testing

A series of testing is done for the proposed system before the system is ready for the user
acceptance testing.
 The steps involved in Testing are:

 Unit Testing

Unit testing focuses verification efforts on the smallest unit of the software design, the module.
This is also known as “Module Testing”. The modules are tested separately. This testing carried out during
programming stage itself. In this testing each module is found to be working satisfactorily as regards to the
expected output from the module.

 Integration Testing

Data can be grossed across an interface; one module can have adverse efforts on another.
Integration testing is systematic testing for construction the program structure while at the same time
conducting tests to uncover errors associated with in the interface. The objective is to take unit tested
modules and build a program structure. All the modules are combined and tested as a whole. Here
correction is difficult because the isolation of cause is complicate by the vast expense of the entire program.
Thus, in the integration testing stop, all the errors uncovered are corrected for the text testing steps.

 Validation Testing

At the conclusion of integration testing software is completely assembled as a package,

interfacing errors have been uncovered and corrected and a final series of software tests begins , validation
test begins. Validation test can be defined in many ways. But the simple definition is that validation
succeeds when the software function in a manner that can reasonably expected by the customer. After
validation test has been conducted one of two possible conditions exists.

One is the function or performance characteristics confirm to specifications and are accepted
and the other is deviation from specification is uncovered and a deficiency list is created. Proposed system
under consideration has been tested by using validation testing and found to be working satisfactorily.
  Output Testing

After performing validation testing, the next step is output testing of the proposed system since
no system could be useful if it does not produce the required output in the specified format. Asking the
users about the format required by them tests the outputs generated by the system under consideration.
Here the output format is considered in two ways, one is on the screen and other is the printed format. The
output format on the screen is found to be correct as the format was designed in the system designed phase
according to the user needs.

Conclusion:

n conclusion, we have explored various machine learning and AI models. This study indicates that we can
exploit
the machine learning/ AI models for designing more generic material search engines. From this study we
found
that ensemble of bootstrapping method like Random forest, Statistical method such as Naïve Bays and
biologically
inspired artificial Neural Network provides more generic and accurate prediction of materials from their d-
values with
about 100% accuracy and highest F-score. In future, we are going to study the computation and time
complexity of
these models. Currently, we are developing an explainable material search engine using Graph Neural
Network. This
may provide more insight while proposing a synthetic material. Further, the limitation of the data may be
handle by
incorporating data augmentation techniques or by incorporating more number of materials in future
The field of artificial intelligence gives the ability to the machines to think analytically, using concepts.
Artificial Intelligence will continue to play an increasingly important role in the various fields. We conclude
that further research in this area can be done as there are very promising and profitable results that are
obtainable from such techniques, while scientists have not yet realized the full potential and ability of
artificial intelligence. This technology and its applications will likely have far-reaching effects on human life
in the years to come. This review has not attempted to detail all the literature in the area but to report mainly
the most recent work.

In conclusion, we have explored various machine learning and AI models. This study indicates that we can
exploit
the machine learning/ AI models for designing more generic material search engines. From this study we
found
that ensemble of bootstrapping method like Random forest, Statistical method such as Naïve Bays and
biologically
inspired artificial Neural Network provides more generic and accurate prediction of materials from their d-
values with
about 100% accuracy and highest F-score. In future, we are going to study the computation and time
complexity of
these models. Currently, we are developing an explainable material search engine using Graph Neural
Network. This
may provide more insight while proposing a synthetic material. Further, the limitation of the data may be
handle by
incorporating data augmentation techniques or by incorporating more number of materials in future.

In conclusion, we have explored various machine learning and AI models. This study indicates that we can
exploit
the machine learning/ AI models for designing more generic material search engines. From this study we
found
that ensemble of bootstrapping method like Random forest, Statistical method such as Naïve Bays and
biologically
inspired artificial Neural Network provides more generic and accurate prediction of materials from their d-
values with
about 100% accuracy and highest F-score. In future, we are going to study the computation and time
complexity of
these models. Currently, we are developing an explainable material search engine using Graph Neural
Network. This
may provide more insight while proposing a synthetic material. Further, the limitation of the data may be
handle by
incorporating data augmentation techniques or by incorporating more number of materials in future.
REFERENCES

1. Adams C.E, Polzmacher S, and Wolff A (2013) Systematic reviews: work that needs
to be done and not to be done. Journal of Evidence-Based Medicine 6(4): 232–235.
2. Adeyemi, A T. (2018). Clinical diagnosis criteria for periodontal disease: an update.
https://scite.ai/reports/10.15406/jdhodt.2018.09.00409
3. Antons, D., Breidbach, C. F., Joshi, A. M., & Salge, T. O. (2023). Computational
Literature Reviews: Method, Algorithms, and Roadmap. Organizational Research
Methods, 26(1), 107-138. https://doi.org/10.1177/1094428121991230
4. Babor, T F., Stenius, K., Savva, S., & O'Reilly, J J. (2017). Publishing Addiction
Science: A Guide for the Perplexed. https://scite.ai/reports/10.5334/bbd
5. Baviskar, D., Ahirrao, S., Potdar, V., & Kotecha, K. (2021). Efficient Automated
Processing of the Unstructured Documents Using Artificial Intelligence: A Systematic
Literature Review and Future Directions. IEEE Access, 9, 72894-72936.
https://doi.org/10.1109/ACCESS.2021.3072900.
6. Bhosale, U. (2022). How an AI-based Tool Eases the Literature Review Process.
https://www.enago.com/academy/ai-based-tool-can-ease-literature-review-process/
7. Datafoundary (2023) AI-powered Literature Review for Life Sciences and Health
Care. https://datafoundry.ai/m-literature/.
8. Khedkar, S. (2023) Using AI-powered tools effectively for academic research.
Editage Insights, https://www.editage.com/insights/using-ai-powered-tools-
effectively-for-academicresearch
9. Snyder, H. (2019) Literature review as a research methodology: An overview and
guidelines. Journal of Business Research, Volume 104, , Pages 333-339 SpringerLink
(2021) Systematic Literature Review
10. Stapleton, A. (2023) The best AI tools for research papers and academic research
(Literature review, grants, PDFs and more). Academia Insider,
https://academiainsider.com/aitools-for-research-papers-and-academic-research/
11. The Semantic Scholar (2020) We Review ‘Semantic Scholar’: An AI-Powered
Literature Searching Tool. https://thesislink.aut.ac.nz/?p=8408
12. Wagner, G., Lukyanenko, R. and Pare, G. (2022) Artificial intelligence and the
conduct of literature reviews. Journal of Information Technology, Vol. 37(2) 209–226