Data Warehousing Management

This repository contains two Java projects developed for the CSCI 5408 course: Data Management and Warehousing Analytics Projects.

Project 1: MySQL-Java

• Created a lightweight multi-user database system using Java.
• Features multi-user authentication, Query Processor, and Single Transaction Management.

Project 2: NoSQL Data Extraction and Processing

This project involves solving 3 problems related to data extraction and processing. It includes algorithms for cleaning news heading data and performing sentiment analysis using a Bag of Words (BoW) model. The cleaned data is pushed to a MongoDB cloud cluster by Atlas.

Algorithm

Read and Parse XML File

• The function starts by taking the fileName and MongoCollection as parameters.
• It creates a File object using the provided file name and gets its absolute path.
• It uses the Jsoup library to parse the XML file using the XML parser. The result is stored in a org.jsoup.nodes.Document object named document.

Initialize List for MongoDB Documents

• It initializes an empty List named newsDocuments to store the MongoDB documents that will be created from the XML file.

Iterate Over "REUTERS" Elements

• The function iterates over each "REUTERS" element in the parsed XML document.

Extract Title and Body

• For each "REUTERS" element, it extracts the text content of the "TEXT > TITLE" and "TEXT > BODY" elements.
• The extracted text is passed through the cleanText method, which removes non-alphanumeric characters.

Problem 1B: News Data Processing using Spark

Steps followed: UniqueWordCount.py file can be found in the repository inside the Problem1B folder.

Problem 2: Sentiment Analysis using BoW Model

• Successfully implemented the task using files of Positive and Negative words from the same author. This algorithm performs sentiment analysis on news titles stored in a MongoDB collection. The sentiment analysis involves scoring each news title based on the presence of positive and negative words. The results are then inserted into a MySQL table for further analysis.

Algorithm

1. Connect to MongoDB:

   • Established a connection to MongoDB using the provided connection string, database name, and collection name.
   • Fetched the desired collection (newsCollection) from the MongoDB database.

2. Retrieve News Titles:

   • Query the MongoDB collection to retrieve the news titles.
   • Stored the news titles in a list (newsTitles).

3. Read Positive and Negative Words:

   • Read positive and negative words from files (positive-words.txt and negative-words.txt).
   • Stored positive and negative words in separate lists (positiveWords and negativeWords).

4. Sentiment Analysis Loop:

   • Iterated over each news title in the newsTitles list.

5. Build Bag of Words (BoW):

   • Built a Bag of Words (BoW) for each news title.
   • Used a map (bow) to store word frequencies in the news title.

6. Score Calculation:

   • Compared each word in the BoW to positive and negative words.
   • Calculated a sentiment score based on the frequency of positive and negative words.

7. Determine Polarity:

   • Determined the polarity of the news title based on the calculated score.

8. Store Results:

   • Stored the news title, matched words, sentiment score, and polarity in a map (result).
   • Added each result map to a list (results).

9. Insert into MySQL Table:

   • Established a connection to a MySQL database using the JDBC URL, username, and password.
   • Inserted each result into a table named NewsAnalysis with columns: NewsTitle, MatchedWords, Score, and Polarity.
   • Used a prepared statement to efficiently insert multiple records.

10. Close Connections:

    • Closed the MongoDB connection after processing all news titles.
    • Closed the MySQL connection after inserting all results into the table.

Author

• Sumit Savaliya (sumit.savaliya@dal.ca)