OneCat Development Notes: 0. Project Introduction and Some Rambling Thoughts in the Preface

Preface

I happened to remember this project during the Spring Festival holiday

This year's Spring Festival holiday I spent more time at home because of the rain, so while messing around on the computer I suddenly thought of the OneCat project

After tinkering with it a bit and getting it to run again, I found it's pretty decent just as an aggregated reading tool hhh

Then I recalled the development experience of the OneCat project: the first time trying to develop a distributed system, the first time trying NLP with deep learning methods, the first time encountering recommendation systems, the first time trying Docker deployment...

As the last project of my university years, I wanted to take it seriously, so that maybe when I look back on these four years I won't feel like I wasted my time~

Although I ran into quite a few difficulties during development, I also learned a lot. After the defense I barely looked after this App and never released it. To see it through from beginning to end, I decided to write a series of blog posts introducing the process of developing the OneCat project from scratch, and to restart development and add more features

What Is OneCat

First, let's address the question of what it is.

OneCat, Chinese name "壹喵", is my undergraduate graduation project. It is an App that combines aggregated reading, a recommendation system, aggregated search, a community, and other features

Origin of the name: it came to me by chance. "壹" means you only need one App; "喵" is because I rather like cats, so I chose a cat as the App's logo. (Naming really is a difficult thing)

App Screenshots

I'll just put up the App pictures. Nobody can see all that stuff in the backend; only the frontend is what users perceive most directly.

Limited by space, I'll just post a nine-grid~

Project Architecture

OneCat is mainly divided into two parts: the backend programs running on the server side, and the application running on the mobile client.

The server side is divided into modules such as data services, trending across the web, the recommendation system, and aggregated search.

Data services include:

• Web information collection (crawlers),
• Data caching (MongoDB database),
• Data cleaning (NLP with a deep-learning BERT model, Chinese text classification),
• Data persistence and Restful APIs for accessing data (DjangoRestFramework), among other modules;

Trending across the web is based on data from the data services, providing a trending list similar to "hot searches" so you can quickly learn about currently popular events;

Recommendation system uses multiple recommendation methods including clustering-based recommendation, machine-learning algorithms (collaborative filtering), and a latent factor model based on association-rule recommendation (deep learning);

Aggregated search includes a metasearch engine and the corresponding data APIs.