YelpDataSetChallenge

Class Project for Information Retrieval. Two tasks are done as part of challenge.

Task 1: Assigne Categories to different business in Yelp Data Set Task 2: Recommend Liked Dishes and disliked Dishes using reviews and tips from Yelp DataSet

@AUTHORS: Bipra De, Nihar Khetan, Anand Sharma, Satvik Shetty @COLLABORATOR: Professor Xiazhong Liu

Semester Project For ILS Z 534 - Information Retreival
Indiana University Bloomington

Description

Usage and details

JavaClasses and their functionality:

CreateTrainingAndTestCollections.java - Reads Data from given Yelp Dataset and Created two collections (training and test) in MongoDB generateIndex.java - Reads data from MongoDB and creates Training and Test Lucene index FeatureSetExtractor - Reads data from Lucene Training Index and extract top features for a category. It also dumps them to MongoDB CategorySimilatityComparer - finds out similar categories on the basis of a threshold. For example: given category1 and category2 with some feature set and threshold as 70%, category1 is supposed to contain category2 is featureset(category1) and featureset(category2) are 70% similar AssignCategories.java - Reads data from Lucene Test Index and assign categories to them. It is alos capable of assignning multiple categories to a business
MeasurePerformance.java - Reads computed resuls from MongoDB and output the results to a file using evaluation metrics

MongoDB collections:

test_set: Dump of test data training_set: Dump of training data feature_set: Categories and their top features categories_assigned_from_code: Businesses which are assigned new categories by code

Project Page

Work Under Progress:

Ping us at [email protected] if you wish to appreciate/criticize/contribute to the project

Bipra De - Satvik Shetty - Anand Sharma -