SYSTEM FOR TURKISH CUISINE Supervisor Assist. Prof. Dr. Engin DEMR - PowerPoint PPT Presentation

RECIPE RECOMMENDATION SYSTEM FOR TURKISH CUISINE Supervisor Assist. Prof. Dr. Engin DEMİR Prepared by 201112031 Damla Pınar GÜVENER 201211001 Esin AÇIK 201211042 Hivda ÖZATLI

2 Contents Definition of the Problem Literature Review Software Requirement Analysis Software Design Future Plan Conclusion References

3 What is the Recommendation System?

4 Can we call for recommendation?  yes  to know what we are looking for

5 Definition of the Problem  hardness of making decision between lots of food recipes  health (calories), time constraints  hardness of food selection in group organization  hardness of food selection for people who allergic to smth.

6 Definition of the Problem Motivation  beneficial for Turkish people  to generate accurate recommendation for all types of user  to provide making weekly meal plan

7 Literature Review  Recommendation System  Challenges of this Project - cold-start problem - changing preferences  Recommendation Techniques User/Recipe R1 R2 R3 R4  Algorithms U1 4 5 2 U2 3 1  User’ Rating U3 2 2 5

8 Literature Review Recommendation Techniques  Collaborative Filtering

9 Literature Review Recommendation Techniques  Content-Based Filtering

10 Literature Review Recommendation Techniques  Hybrid Filtering CF Based Input Recommender Recommendation Combiner Content Based Input Recommender

11 Literature Review Algorithms  Memory-Based Algorithm  Model-Based Algorithm

12 Literature Review Existing Recommendation Systems

13 Literature Review Our Differences  A New System for Turkish Cuisine  Document Database - fast search, flexible in terms of relationship  Comprehensive System - calories, nutrition, allergies, diets, cost, time likes/dislikes

14 Software Requirement Analysis Functional Requirement Specifications  User Use-Case  Admin Use-Case  System Use-Case (Recommendation Engine)

15 Software Requirement Analysis Non-functional Requirements  accessability  usability  performance  security

16 Software Requirement Analysis Flow Description  need about 1000 recipes  user registers then log in to the system  RE recommends 5 different types of recipes  user selects recommended recipe or search a recipe  user rates selected recipe

17 Software Design  Waterfall Development Methodology  Scrum Development Methodology  Deployment Diagram  Whole System Sequence Diagram  Interface Design

18 Software Design Waterfall Development Methodology

19 Software Design Scrum Development Methodology  fast and motivational  efficient work in a short time

20 Software Design Deployment Diagram

21 Software Design Whole System Sequence Diagram

22 Software Design Interface Design Home Page

23 Software Design Interface Design Personal Information Page

24 Software Design Interface Design Preferences Page

25 Software Design Interface Design Search Page

26 Software Design Interface Design Recommendation Page

27 Software Design Interface Design Recipe Page

28 Software Design Interface Design Recipe Page

29 Future Plan  MongoDB (Document database)  VPN (Web Service)  Hybrid Filtering  WEKA (Machine Learning Algorithm)  Data gathering from http://www.nefisyemektarifleri.com https://www.lezzet.com.tr etc.

30 Conclusion What We Have Done Until Now?  Problem Definition and Project Plan  Literature Review - on Recommender System - on Machine Learning and Data Source  Requirement Analysis for System Design  Design of System - Deployment Diagram - Sequence Diagram - Interface Design

31 References  J. Leskovec, A. Rajaraman and J. D. Ullman, "Ch 9 - Recommendation Systems," in Mining of Massive Datasets, 2014, pp. 307-343.  By Michael D. Ekstrand, John T. Riedl and Joseph A. Konstan, "Collaborative Filtering Recommender Systems," 2011.  D. Bianchini, V. D. Antonellis, N. D. Franceschi and M. Melchiori, "PREFer: a Prescription-based Food recommender," Computer Standards & Interfaces, 2016.  longqi Yang; Cheng-Kang Hsieh; Serge Belongie; Nicola Dell; Hongjian Yang; Deborah Estrin, "Yum-me: Personalized Healthy Meal Recommender System," 2016.  M. N. Moreno, S. Segrera, V. López, M. D. Muñoz and Á. L. Sánchez, "Web mining based framework for solving usual problems in recommender systems. A case study for movies' recommendation," Neurocomputing- Elseiver, pp. 72-80, 2016.

32

33 an any qu ques esti tions ons