cdesiniotis / yelp-faceted-search

Performing faceted search on a Yelp dataset

Yelp Faceted Search

A data analysis application for performing faceted search on Yelp's business review data.

Here is a demo for searching delis in Las Vegas:

Overview

This project is a standalone Java application for running queries on a real Yelp dataset. The dataset used is a subset of the one used in the 2013 "Yelp Dataset Challenge," containing 20,544 businesses, 211,002 users, and 826,190 reviews. Users can search for businesses based on business categories (main & sub-categories), business attributes (e.g. accepts credit cards, has delivery, etc.), location, and days/hours open. Each query returns matched businesses and allows the user to view all the reviews for each business.

The application utilizes faceted search, which is a popular technique in commercial search applications. It is a technique for accessing information organized according to a faceted classification system, allowing users to explore a collection of information by applying multiple filters.

The emphasis for this project is on the database infrastructure, not on the user interface design. This project was built using NetBeans.

Run

Extract the Yelp dataset:

$ cd YelpGUI
$ tar -xvzf YelpDataset.tar.gz

Bring up the Oracle database:

$ docker run --name oracledb --rm -d -p 1521:1521 epiclabs/docker-oracle-xe-11g

Initialize the database with the necessary tables/indices (default username/pw is system/oracle):

$ sqlplus ${user}/${pass} @createdb.sql
$ sqlplus ${user}/${pass} @createIndex.sql

Populate the database with the Yelp dataset (Warning: this may take about 20 mins):

$ javac -cp "./dist/lib/ojdbc.jar:./dist/lib/json-20190722.jar" src/yelp/populate.java
$ java src.yelp.populate

Run the GUI:

$ java -jar dist/YelpGUI.jar

Delete the database:

$ sqlplus ${user}/${pass} @dropdb.sql
$ docker kill oracledb