Pakalo

Pakalo is an application that contains recipes and has a chatbot. The chatbot uses speech and text input from its users and understands it to be able to find recipes, go step by step throughout the recipe alongside the user, suggest recipes according to the user?s interests and the limited ingredien

Project Title

Pakalo

Project Area of Specialization

Artificial Intelligence

Project Summary

Pakalo is an application that contains recipes and has a chatbot. The chatbot uses speech and text input from its users and understands it to be able to find recipes, go step by step throughout the recipe alongside the user, suggest recipes according to the user’s interests and the limited ingredients a user has, mention ingredients that are required to make a particular recipe, suggest halal substitute ingredients for recipes with their opposite counterparts and more.

Apart from the chatbot, users will have his or her own account through which users can upload their own recipes. Users can also comment, rate, and share recipes of other users that they liked.

As the user will communicate with the chatbot through either speech or text, the chatbot will use Natural Language Processing to deduce what the user is trying to say. The chatbot will then check the database for helping material and then take action accordingly

Project Objectives

• To provide a step by step guidance at the user’s own pace when cooking recipes they would love.
• To recommend recipes based on a limited set of ingredients, similar interests, and user’s history.
• To have a smart easy to talk to chatbot that the user will be comfortable talking to, and that will aid the user in cooking.
• To make a rich user friendly social experience for cooks, with user generated recipes and ratings.
• To suggest substitution of ingredients in cases where they are not available or undesirable

Project Implementation Method

Pakalo was divided into distinct and significant modules which included scraping and collecting the recipes data, setting up the NoSQL database, training multiple chatbot models and comparing them to figure out which one would work best, establishing the knowledge graph that would work in conjunction with the chatbot, and developing as well as testing the Android application. Naturally the project was divided into iterations, each of which was its own sprint, and a Scrum software development framework was used. Each sprint was given approximately one and a half months on average.

Benefits of the Project

Pakalo, with its chatbot as a core feature, will aim to revolutionize cooking. More and more smart home solutions are coming into fruition and this bot will be a significant addition to the smart kitchen. The application will be user friendly, similar to the Google Assistant, only it will be focused primarily on cooking.
The reason many people find cooking to be a difficult medium to break into is that they lack a proper guide; a culinary teacher that will help them step by step to cook what they want to cook, and help them in any way they need. Even though there exist conventional means of learning how to cook, such as written recipes and video tutorials, these methods have major flaws.
The process of cooking demands attention and focus, and requires the cook to multitask. With this amount of work, it is significantly difficult for a cook to go back and forth from reading recipes, making notes, pausing or rewinding the videos, to actually cooking. In a kitchen environment, where the hands often get dirty, recipe books, notes, smartphones, and tablets are sure to get dirty too as a result of the cook often touching them.
Beit for survival, or for the pursuit of art, many individuals, fascinated by the act of cooking, find it difficult to pursue their interests in it. They are often intimidated by the perceived difficulty of cooking, or because they have previously failed to produced favourable results in their cooking endeavors. Pakalo aims to tear down this formidable wall and guide the up and coming chefs, by taking them step by step through recipes they wish to cook, and aid them in their culinary venture.
Culinary chatbots do exist, but they are not a guide to the user in a way that Pakalo is. A chatbot called Betty Crocker is an app offered for Amazon’s Alexa. This chatbot helps the user by answering culinary questions and recommending substitutes. Another chatbot for the Alexa is Make It Soy, which helps the user find vegan substitutes for ingredients. Seamless is another chatbot that helps its user order food quickly. Several examples of culinary chatbots like these exist, but none of them offer the user guidance, at their own pace, along the recipe. It is obvious that Pakalo fills a massive gap in the culinary market.

Technical Details of Final Deliverable

For the gathering of recipes, several open source website datasets were utilized as well as many recipes were scraped from the internet. Every recipe was pre-processed and formatted into a unified format which was then to be stored in a database. The firebase NoSQL database was set up, and the recipes were stored in it. This database was utilized as it would work well with an Android application, and because the database was beneficial to be on the cloud in the case of this project.
The chatbot was to be the core and most important part of the application. Thus it was important that we compare multiple chatbots so as to see which one would prove most useful in the case of Pakalo. We developed two chatbots side by side, and compared them to determine the final chatbot that would be integrated into the app. One chatbot was a custom made and coded from scratch LSTM based chatbot. The second chatbot utilized the Rasa framework with Rasa NLU and Rasa Core. Unsurprisingly, the Rasa chatbot proved to be better for this application, so it was selected. The training data for the chatbot and accompanying stories for conversations were made purposefully for this chatbot.
The knowledge graph was to be integrated into the chatbot. This knowledge graph would link recipes and ingredients so a user can check if one of the ingredients that they might lack is a recipe on its own and they do have the ingredients to make that first.
The Android application was developed using Android studio, with custom buttons and UI elements made in Photoshop.

Final Deliverable of the Project

Software System

Core Industry

Food

Other Industries

Education

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT)

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com