https://drive.google.com/file/d/1t-h0blhJePqqwrOg7O5L3AV84yFnSRBv/view?usp=sharing

Due to time constraint, I've react-search-ui, a powerful search engine powered by lucene, with an intuitive front-end. react-search-ui has a lot of additional features such as fitering and sorting. To keep the code size small, I've stripped away a lot of the additional features.

In my initial write-ups i suggested using MySQL for storing the NGram similarity. After some study I realized that it is not very effective in storing big matrix. Using an infrastructure such as Hadoop with HBase might be a better choice.

(altered according to time constraint)

• Start to suggest uppon entering one character

Must have:

• Suggestion on each character - done
• Search in both titles and description text - done
• Test Coverage - not started

Nice to have:

• analytics on search pattern

Not in scope:

• add or update database

• FORK this repo
• cd backend
• npm install & npm run dev
• open localhost:3000
• enjoy

(in addition to scope)

• benchmark different indexing stragies: Edge-Ngram vs Prefix Query vs Completion Suggester.