/Java-Web-Service-Development-Event-Search-and-Ticket-Recommendation-

Primary LanguageHTML

Event Search and Ticket Recommendation [Xiangyu Zeng]

##Summary This is An interactive web application full stack project for users to search events by their geolocations and get recommendations based on their favorite records.

Tech Stack

  • Front end: HTML/CSS/JavaScript
  • Back end: JAVA8(Eclipse JEE)
  • Database: MYSQL MongoDB
  • Java servlet: Apache Tomcat v9.0
  • Test: Apache JMeter

Logic layer

alt text

Recommendation Algorithm(content-based)

In this project, I recommend events based on categories that the user has favorited. By knowing the category of the item the user favorited, I recommend some events belong to this category nearby this user.

Concrete steps are as follow:

  • Fetch all ids of these events this user has visited.
  • Given all these events, fetch the categories of these events.
  • Given these categories, find what are the events that belong to them.
  • Filter events that this user has visited.
  • Sort the recommendation list on ascending order of distance between recommended events's locations and user's location.

RPC

Java servelet could handle the lists RPCs below:

  • login
  • logout
  • register
  • search items
  • recommend items
  • set/unset favorite items

Database

I implement two databases: MySQL and MongoDB. I prefer MySQL in the project at first, but if I need to scale the prject, MongoDB could be a good choice. The image below is the schema for MySQL database of this project.

Schema of data in MySQL database

alt text

Screenshots

login/logout

alt text

register

alt text

search

alt text

favorite

alt text

recommendation

alt text

Deployment

Amazon EC2

  • Launch an ubuntu machine(set port 80)
  • Install Java, MySQL or MongoDB, Apache Tomcat
  • Upload WAR file
  • Start Java servlet(change port to 80)
  • http://YOUR_IP_ADDRESS/Jupiter/
  • You can register or use a dafault account(username: 1111, password:2222)

Test

I use Apache JMeter to test the maximum QPS in Amazon EC2. The peak throughput happens between 1000 and 2000 threads, so I can test more data points between 1000 and 2000 threads. The throughput will be around 150QPS. When I used 2500 threads, I got errors, which means the memory size is my server's bottleneck.

Summarized report

alt text