Volume 5, Issue 8, August – 2020 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Intelligent Food Recommendation System Using

Machine Learning
Pratiksha Ashok.Naik
Information Technology
SDMCET
Dharwad, India

Abstract:- The buying behavior of the consumer is the desired product and then some techniques such as data
affected by the suggestions given to the items. cleaning, tokenization, frequency computation, product
Recommendations can be made in the form of a review feature frequency is used to get specific words from the set
or ranking given to a specific product. Calories of reviews. Hence, these techniques provide us with clean
consumed by people contains carbohydrates, fats, data that is required in this project. In the user logged in
proteins, minerals and vitamins, and any malnutrition page, the user is imposed with set of questions to categorize
causes severe health problems. In this paper, we propose them into any nutrition gainers such as salt, protein,
a recommendation system which is trained on the basis sugar, fat etc. Based upon the option marked, the rating is
of the recommendations received by the customer who provided to the specific user with one or more nutrition
has already used the product. Software recommends the gainers having highest rating with it. Hence based on this
product to the customer on the basis of the experience of the product is recommended to the purchasers likeliness.
the consumer using the same product. Each person has This technique provides easiest way to sort the user liking
his or her own eating patterns, based on the preferences products.
and dislikes of the user, indicating that personalized diet
is important to sustain the success and health of the user. Web Crawler implementation is based review
The proposed recommendation method uses a deep collection algorithm which must collect reviews about food
learning algorithm and a genetic algorithm to provide products and store them in this application. Questions posed
the best possible advice. to the user about nutrition remains responsible for
generating the health profile for the particular user. Product
Keywords:- Deep learning algorithm, Genetic algorithm, classification algorithm is responsible for performing data
Optimized Nutrition, Recommendation System, RESTFul cleaning, tokenization, and frequency computation. Genetic
web services, TESCO database, Web crawler. Algorithm must establish the relationship between the
product and user. Product buying module must be
I. INTRODUCTION responsible for purchase of the food products in this
application.
The numbers of e-commerce websites have been
increasing rapidly in today’s world. They have their own II. METHODOLOGY
recommendation system to be provided to the user or buyer.
Many algorithms have been used by different product The figure depicts the methodology for the project i.e.
recommendation system for providing accurate items to the it provides information about various modules involved in
users. But some recommendation system provides access to the project
all people for providing review or rating about the product
irrespective of whether the person has purchased that  Registration: The customer will be able to register by
product or not. This project uses both deep learning and providing the details like first name, last name, user id,
machine learning algorithms. Genetic algorithm plays password, Email, Phone No. After all the validations are
important role in providing recommendation to buyer based performed, if user id does not exist then user is allowed
on their likeliness. This project has an added feature of to register otherwise registration fails.
custom recommendation provided mainly for the purchasers
suffering from any problems such as high blood pressure,  Login: This module asks for username and password
heart attack, diabetes etc. Here the nutrition is divided into allowing the user to access the system.
five types i.e. energy, sugar, proteins, fats and salts. The
genetic information of the user or purchaser is always  Grocery Product Purchase: This module allows viewing
constant but the recommendation of the products is dynamic the grocery product and completing the transactions for
so as to provide an accurate product that matches with user various categories of product namely biscuit, chocolate,
likeliness over time. Here the TESCO dataset is used to yogurt and noodle. For the transaction to be completed,
provide recommendation to the users. The reviews are user will select the product and then provide the
collected from the famous websites such as Amazon, respective details to complete the transaction.
MouthShut.com, and Flipkart for taking the review about

IJISRT20AUG414 www.ijisrt.com 616

Volume 5, Issue 8, August – 2020 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Nutrition based Questioner: This module provides a list the help of genetic algorithm. Finally, the recommendation
of questionnaires to the customer, which is based on the of product is given to the user based upon user health
genetic profile like taste of salt, sugar, fat, protein and profile.
energy. For future analysis the collected profile is saved.
The above diagram illustrates review collection, stop
Collecting offline and online review based on web words and training words that act as input. Offline Review
crawler. This process is responsible for collecting the Collection is responsible for collecting the reviews from the
reviews using a Web Crawler in order to get real time end user. Collection of the reviews from the web URL and
reviews and offline if an offline review has to be submitted. XPath is accomplished by web crawler. Unwanted words are
Word Embedding and Deep Learning for Product removed from the review i.e. stop words and special symbol
Classification Training data is fed to this algorithm. Unique with the help of data cleaning algorithm.
words from the training data is found, then it computes the
occurrence of the words and generates a matrix of P*Q word After data cleaning, tokenization process has to be
vector where P is number of reviews and Q is the number of accomplished. Conversion of statements into list of words is
unique words amongst all the reviews. Once the word vector called as tokenization. Now the redundancy has to be
is found, later it acts as an input for the classification. removed by providing a weight factor for each token. This
process is called as frequency computation. Classification of
 Genetic Algorithm for filtering: The genetic algorithm is the product has to be rendered using word embedded matrix.
applied on the product classification data Hence product classification procedure has to be followed.
 Recommendations provided to the end user: Based on To generate the health profile of a specific user, the user has
the high nutrition sorting the recommendation is to answer the list of questions related to food namely salt,
performed. Hence it provides authentic nutrition details sugar, energy and fat. The relationship between the user
to the customer. profile and nutrition factor is established with the help of
genetic computations. To obtain list of products, the
recommendation are used. Hence this outputs list of
products to the authentic user.

 Java Language:
Java is an object oriented, robust programming
language. (Dot) java is the extension given to the java file.
When compiled it is converted automatically into (Dot)
class format. Compiler than compiles the source code and
then converts it into (Dot) class extension. This file now
consists of byte code that is fed to the Java Virtual Machine
(JVM). As this JVM can run on any machines such as
Linux, Windows, Unix with the byte code format of our
Fig 1:- System Methodology source code. Hence Java language is called as platform
independent i.e. it follows WORA (Write Once Run
III. SYSTEM ARCHITECTURE Anywhere) Architecture.

Various end points are shown in the above figure. The  Architecture of Java Language :
web container is maintained by Apache Tomcat Server. Here Main components of Java Programming Language are
the web.xml allows only the valid request into web Java Runtime environment (JRE), Java Virtual
container. The user interface is designed with the help of Machine(JVM) and Java Development Kit(JDK). JDK
JSP, Angular JS and Ext JS. To implement business services consists of javac i.e. Java compiler etc. JDK can be used to
and the entire algorithm RESTFul Services are used. only develop or write the code but it cannot be used to
execute the code. JDK and JRE are both physical entities.
The inputs given are review collection, stop words and JRE is used to execute the code as it contains JVM in it.
keywords. The procedure involves of product buying, word JVM is a logical entity.
embedding algorithm, product classification, Genetic
algorithm for computation and recommendation provided to  Spring FrameWork:
the user about the product Spring Framework is used for generating the java
application. Concepts of Spring Framework are (MVC)
Input variables consist of stop words, review collection Spring Model View Controller,(JDBC) Spring Database
and keywords. The procedure involves Processing of review Connectivity etc. Instead of manually creating the object,
collection that collects the reviews in offline mode or online We can use Spring IOC to automatically generate the object.
mode with the help of web crawler. For performing the
classification, the product has to be found, this is done with  Spring MVC Framework:
the help of word embedding algorithm and product Spring MVC Framework is a model that represents
classification algorithm. The relationship has o be business logic operation and the data which is responsible
established between user and the product, this is done with for representing presentation logic. Controller acts as an

IJISRT20AUG414 www.ijisrt.com 617

Volume 5, Issue 8, August – 2020 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
intermediate between model and view. Controller receives The above snapshot depicts the registration page
the input and it demands model and view to change which can be used by new users. The registration page asks
accordingly. the user to enter first name, last name, password, email id,
country, state and city. Finally, the user is registered with
the application and can use the application for knowing the
likeliness of products.

Fig 2:- System Architecture

IV. RESULTS
Fig 4:- Sample List of questions
 The following are the steps followed:
The request is taken from the web browser. The The above snapshot depicts the nutrition questionnaire
request will be sent to web container and validated against that is shown to the user whenever the user wants to check
the web browser. The request is then forwarded to his/her likeliness of products.
dispatcher servlet. Taking the input from dispatcher servlet
and performing the basic validations is done by controller Based on the questions the user answers he/she will be
class. If the basic validations are successful, then error given some score. These scores will be used to detect the
messages are sent back else the control moves to step5.The feature of content present in the product like sugar, salt etc.
delegate class is responsible for taking input from controller
and gives it to service and then take results and sends it to
controller. The Service class will execute the algorithms and
the business logic and takes help of either single DAO class
or multiple DAO classes to complete the job. The DAO is
responsible for communicating with MYSQL. Fig 5:- Feature Points

The obtained features are used to recommend the

products to the user.

Fig 3:- Registration

Fig 6:- Products recommended

IJISRT20AUG414 www.ijisrt.com 618

Volume 5, Issue 8, August – 2020 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
V. CONCLUSION

In this project, by including web crawler algorithm, the

admin will be able to collect the reviews online. The
sequence of deep learning algorithms such as data cleaning,
tokenization, frequency computation, product classification
is used for classification of products. The customer is
allowed to log in with valid credentials and based on
questions asked about nutrition intake, using genetic
algorithm, the recommendations are provided. Hence, it
becomes easy for the user to know about the current intake
of nutrients and based upon it, the user is recommended with
certain products. Hence, this system is responsible of
providing valid products to the user. The recommendations
are provided based on user likeliness.

VI. FUTURE SCOPE

The system for now provides recommendations to the

user based upon their likeliness and also the system provides
custom recommendation for the user having any genetic
issues such a heart problem, diabetes, hypertension etc.
Further this project can be extended to add more categories
of products so as to provide solutions about
recommendation to variety of users liking products with
unique nutrition gainers. This project can also be
amalgamated with social media to collect the reviews
related to the product.

REFERENCES

[1]. Anand Shanker Tewari ; Abhay Kumar ; Asim Gopal

Barman,"Book recommendation system based on
combine features of content based filtering,
collaborative filtering and association rule
mining",2014 IEEE International Advance Computing
Conference (IACC),Year: 2014,Page s: 500 – 503
[2]. JunBo Xia,"E-Commerce Product Recommendation
Method Based on Collaborative Filtering
Technology",2016 International Conference on Smart
Grid and Electrical Automation (ICSGEA),11-12 Aug.
2016
[3]. Yi-Jing Wu ; Wei-Guang Teng,"An enhanced
recommendation scheme for online grocery
shopping",2011 IEEE 15th International Symposium on
Consumer Electronics (ISCE),Year: 2011,Page s: 410 –
415
[4]. Priyanka A. Nandagawali ; Jaikumar M. Patil
,"Community based recommendation system based on
products",2014 International Conference on Power,
Automation and Communication (INPAC),6-8 Oct.
2014
[5]. S. Linko, “Expert systems what can they do for the food
industry?” Trends Food Sci. Technol., vol. 9, no. 1, pp.
3–12, 1998.
[6]. A.McAfee, E. Brynjolfsson, T. H. Davenport, D. Patil,
andD.Barton, “Big data,” Manage. Revolution Harvard
Bus. Rev., vol. 90, no. 10, pp. 61–67,20

IJISRT20AUG414 www.ijisrt.com 619