1 Basic Requirements
1. Programming language: Matlab (recommended and template codes provided), or other languages (without template codes)
2. Implement your code under Code folder, and read data from Data folder (one neg folder contains all the negative reviews and one pos folders contains all the positive ones).
3. Final project report under name report.pdf at the project root folder.
Part I: Text Classification with Bag of Words and kNN (50pt) 2.1 Vocabulary (lexicon) creation (10pt)
1. Template file buildVoc.m
2. Function template function voc = buildVoc(folder, voc);
3. Input: a folder path, which contains training data (a bunch of .txt _les from Multimedia App reviews in HW00);
4. Output: matlab cell array voc, which represents the vocabulary of the words shown in the training data, except the stop words (stop words list is embedded in the code template)
5. Implement your code under %PUT YOUR IMPLEMENTATION HERE tag;
6. Useful matlab functions strtok(), lower(), regexprep(), ismember(), any();
2.2 Bag of Words feature extraction (20pt) 1. Template file cse408_ bow.m
2. Function template function feat_vec = cse408 bow(filepath, voc)
3. Input: a file path filepath, which contains one review (one .txt file) and a vocabulary cell array voc
from previous sub-section.
4. Output: one dimentional matlab array feat vec, which represent the bag of words feature vector given the vocabulary voc;
5. Implement your code under %PUT YOUR IMPLEMENTATION HERE tag;
6. Useful matlab functions strtok(), lower(), regexprep(), ismember(), any();
2.3 k-Nearest Neighbor Classification (20pt) 1. Template file cse408_knn.m
2. Function template function pred label = cse408 knn(test feat, train label, train feat, k, Dst- Type)
3. Input: 1) test feature vector test_feat; 2) training set groundtruth label set train_label; 3) training set feature vector set train_feat; 4) Hyperparameter k of knn; 5) Distance computation method DstType, 1 for sum of squared distances (SSD) and 2 for angle between vectors and 3 for Number of words in common;
4. Output: predicted label pred_label of the testing _le. 1 for positive review, 0 for negative review; 5. Implement your code under %PUT YOUR IMPLEMENTATION HERE tag;
6. Useful matlab functions sort();
2.4 Test your implementation
1. After your implementation, you could run P01Part1_test.m to debug and validate your code. It
basically iteratively select one of the training review file as a validation file.
2. Question? Where is the testing data. Write another product review and see if your system be able to classify it correctly?
3 Part II: Text Sentimental Analysis (30pt)
1. Implement a basic sentimental analysis module. Read in a lexicon, in which each word has a sentimental score. Iterate through each review file and sum up the sentimental scores for each word that exists in the sentimental strenghth lexicon;
2. Template file sentimentalAnalysis.m
3. Input: a file path filepath, which contains one review (one .txt _le) and a word with sentimental strenghth file wordwithStrength.txt under Data folder.
4. Output: one sentimental score.
5. Implement your code under %PUT YOUR IMPLEMENTATION HERE tag;
6. Useful matlab functions strtok(), lower(), regexprep(), containers.Map();
3.1 Test your implementation
1. After your implementation, you could run P01Part2_test.m to debug and validate your code. 2. Question? What is the accuracy of the performance of your code?
4 Report Requirements (20pt)
Please include the following analysis in your report.
1. Make sure to explain where the algorithms worked and where they didn’t and why. You are encouraged to use both text and plots to explain your observations.
2. Analyze Hyperparameter K in the KNN part, which K you emprically observed that could achieve
the best performance?
3. Among the three distance metrics (sum of squared distances (SSD), and the angle between vectors and Number of words in common), which one intuitively makes more sense for classifying positive and negative review? Which one you empirically observed it to achieve the best performance?
4. For Text Sentimental Analysis task (Part II), which review in our dataset has the highest positive score but it is a negative review? And, which one has the lowest negative score, but it is a positive review? Which set of words from these reviews confused your sentimental analysis system?