Step 1: Reading Textual Data

1. Use the Table Reader node to import data from:
   TripadvisorReviews-SanFranciscoRestaurants-ItalianChineseFood.table

2. Filter (remove) rows with missing titles.

3. Convert strings to documents with the Strings To Documents node.

4. Filter (remove) all columns except the column containing the "Documents".

Step 2: Enrichment I

1. Assign POS tag with the POS Tagger node.

2. View tagged documents using the Tagged Document Viewer node.

Step 3: Pre-processing I

1. Filter numbers, punctuation marks, stop words with the Number Filter, Punctuation Erasure and Stop Word Filter node, respectively.

2. Convert texts to lower case using the Case Converter node.

3. Perform stemming with the Snowball Stemmer node.

4. Keep only tokens tagged as nouns, verbs, and adjectives using the Tag Filter node.

Step 3: Pre-processing II

1. Create Bag of Words using the Bag Of Words Creator node.

2. Use the Term To String node to extract terms as strings from a bag of words.

3. Use the GroupBy node to group terms and count how many times they appear in pre-processed documents.

4. Keep those terms that appear 5 or more times.

5. Use the Reference Row Filter node to filter bag of words and keep only terms that occur in at least 5 documents.

Step 4: Transformation

1. Compute relative term frequency with the TF node.

2. Create document vectors using the Document Vector node.

3. Extract class label/category for prediction with the Document Data Extractor node.

Step 5: Text Classification

1. Append color information based on class labels using the Color Manager node.

2. Filter out Document column.

3. Split data into training (80%) and test (20%) set with stratified sampling on the target column using the Partitioning node.

4. Train Decision Tree classifier on the training set with the Decision Tree Leaner node.

5. Apply the trained model to the test set with the Decision Tree Predictor node.

6. Score model with the Scorer node.

Lab 7 & 8 - Text Mining

Learning objective: In this exercise, you will practice performing classification using Machine Learning on your transformed text data.