Skip to content

Team HTTP501 from Georgia Tech presents a personalization solution to Yelp; as part of the Yelp Dataset Challenge 2014

License

Notifications You must be signed in to change notification settings

sagz/Your-Very-Own-Yelp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Your Very Own Yelp®

Team HTTP501 from Georgia Tech presents a personalization solution to Yelp; as part of the Yelp Dataset Challenge 2014. This is a semester wide project built during the course of Data and Visual Analytics, CSE 6242 Spring 2014 at Georgia Tech.
By - Ameya Vilankar, Sagar Savla


There are three major parts to this project-

  1. Review Summarization
  2. User Ratings Prediction
  3. Visualization

There is a DOC folder that has additional resources:

  1. Project Final Report
  2. Project Final Presentation Please refer to these resources for a detailed explanation of our project

How To Run Each Part of this Project

Some parts of this project might require the source dataset which can be found at https://www.yelp.com/dataset_challenge/dataset

  1. Review Summarization
    Instructions on Compiling and Executing this part is in a separate Readme.txt inside the subfolder Review Summarization

  2. User Ratings Prediction
    Instructions on Compiling and Executing this part is in a separate Readme.txt inside the subfolder User Ratings Predictions

  3. Visualization
    Our static visualization can be run with any web browser which supports most of the HTML5/CSS3 spec as of April 2014. We have tested it with Mozilla Firefox v30.

  4. To run the viz, please open the file index.html in the web browser.

  5. The main files are all located within the subfolder Visualization This folder is portable.

  6. A brief description of the main files: * index.html
    The primary login screen. Being a static visualization mock-up, the interface is not connected to a backend login server. To pass the login screen, please enter any credentials in the Username and Password field and click on the login arrow to proceed

* `map.html`  
  Our primary interface to show the user's personal rating
  predictions. Please hover on any of the markers to know the name of
  the place. The colour signifies how well we think the user would
  like it. Clicking on any of the markers shows you the feature cloud
  for the place.

* `cloud.html`  
  This is the interface to showcase our review summarization features.
  The cloud represents a collection of salient features of the restaurant
  (or vice versa)  
  The colour of the feature words signifies the sentiment of the feature
  among the reviewers. Green means positive and Red tends to the negative.
  The size of the word signifies how prominent this feature is among all
  the reviews of the place.  
  Hovering on the word yields detailed information about the sentiment
  analysis and opinion count  
  Clicking on the feature makes another word cloud surrounding that
  feature/restaurant. The user can navigate back-and-forth with these
  links.

Some snazzy screenshots!

  1. Our Data Flow Diagram Our Data Flow Diagram
  2. The static mock-up login screen ![The static mock-up login screen](DOC/yelp index.png)
  3. The interactive map of restaurants near a user, colored (ranked) based on likeness ![The interactive map of restaurants near a user, colored (ranked) based on likeness](DOC/yelp map.png)
  4. The feature wordcloud of one selected restaurant from step 3. The feature wordcloud of one selected restaurant from step 3.
  5. The reverse restaurant wordcloud of a selected feature from step 4. The reverse restaurant wordcloud of a selected feature from step 4.

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

About

Team HTTP501 from Georgia Tech presents a personalization solution to Yelp; as part of the Yelp Dataset Challenge 2014

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published