Per-Starke
commented
1 year ago Suggestion for structure:
Abstract
Table of contents
Table of figures
1 Introduction
1.1 Structure of the thesis
In this section, the structure and organization of the present work will be introduced. An overview of the individual sections will be provided to give the reader a clear orientation.
1.2 Research Question
Name and describe the research question: “Do collaborative filtering or content based methods of recommender systems deliver more satisfying results when being used to recommend recipes matching the taste of the customer and optionally matching required macronutrients?”
1.3 Introduction to nutrition and nutrition coaching
Give a short introduction to what makes nutrition an important topic in everyday-life, and especially for athletes, what athletes have to take care of regarding nutrition, and what nutrition coaches do.
1.4 Introduction to recommendation systems
Give a short introduction to recommendation systems in general and what a nutrition recommendation system could do.
1.4.1 Recommendation algorithms
Name and explain the two used algorithms, why they were chosen and what makes them different from each other.
1.4.1.1 Item-KNN algorithm
1.4.1.2 Content-based algorithm
1.5 Relevance of the developed software
Describe why the software I developed is helpful and relevant / how it can help nutrition coaches and athletes.
1.6 Relevance of the research question
Describe why the research question is of interest / why the answer to it is helpful (for further development in this area)
1.7 How was the research question answered?
Describe briefly how the research was done to answer the research question.
2 “Recipe Recommender” software architecture
2.1 Functionality of the software
Describe everything the software offers for coaches and athletes, how it can be used, possibly show screenshots to demonstrate.
2.2 Programming languages and tools
Name and briefly introduce the programming language I used (Python), as well as all frameworks (Flask, Pandas, CaseRec) and the recipe-API (Spoonacular).
2.3 Software structure
How the software is internally structured, what part of the code and data is stored in what directory/file.
2.4 Recipe data
How I created the recipe database(s) using Spoonacular, and how the data is stored.
2.5 Similarity calculation
How I calculated the similarity between recipes and how the data is stored.
2.6 Getting initial ratings
How another part of the software was added to get initial ratings and how I got people to rate the recipes.
2.7 Storing ratings relevant for answering the research question
(Short part I guess) Explain how ratings relevant for answering the research question are stored separately in another file with the additional information about what algorithm recommended the particular recipe.
2.8 Limitations of the developed software
To limit the scope of the thesis, I needed to make a lot of limitations. I still have a really good software, but many things could be added or improved for “real-world use”, like using a DB instead of csv files or allowing users to choose a username and edit their profile, allowing users to store recipes for later use, … In this part, I want to mention all limitations and why I choose to do these things the way I did.
3 Methodology
Describe in more detail what was briefly mentioned in 1.7: how I did the research with participants and coaches that use the software and what data I need from them in order to answer the research question.
4 Results
State the facts of the data I got from users/coaches while using the software.
5 Discussion
5.1 Interpretation of the results
Interpret the results from (4), and with that, answer the research question.
5.2 Limitations
What limitations does the research have? (Like, low participant numbers, low recipe numbers, cannot know if better rated recipes from one algorithm really means that this algorithm performs better when trying to suit customers taste or if that is due to form of the day, appetite, mood and so on of the participants / users).
6 Conclusion
Briefly present and summarize the most important results and information, without adding new information.
7 Appendix
7.1 Development workflow
Using GitHub repo, GitHub project and issues, GitHub branches
(e.g.: Comparative Analysis of Collaborative Filtering and Content-Based Methods in Recommender Systems for Personalized Recipe Recommendations)