🔎Customer Review Classification with Fuzzy Matching

This workflow demonstrates how exorybte's Approximate String Matcher node can be used for fuzzy classification of online store product reviews. This simulates a typical scenario where topics have to be extracted from reviews with spelling mistakes.

🛠️ Steps in the Workflow

Step 1: Loading Data

Load a reference dictionary, containing keywords and their corresponding categories (e.g. "cheap", "expensive", "affordable" etc. → price)
Load a table of product reviews, assigning a unique identifier to each review.

Step 2: Preprocessing

Process the reviews by removing punctuation, tokenizing and ungrouping.
Create a table where each row represents one word, along with the full review text and its identifier.
This serves as the basis for matching

Step 3: Matching

Use the Approximate String Matcher (Levenshtein distance) to compare each review word against the reference dictionary.
This step captures both exact matches (e.g. "premium") and approximate matches (e.g. "solidd", "Esy").
Only close matches (distance ≤ 1) are kept in the resulting table.

Step 4: Postprocessing

Rejoin the matched results with the original review table so that all reviews appear, even those without any matched words.
Add the category labels from the dictionary.

Step 5: Results

Create a dashboard summarizing the fuzzy classification results and showing example reviews.
Build a summary table with one row per review and a list of all detected categories, ready for further analysis..

	Agents
	Address Normalization & Matching
	Bank Customer Name Validation with Single-Field Indexer
	Customer Review Classification with Fuzzy Matching
	Frequency-Aware Anomaly Detection with Single-Field Indexer
	Multi-Field Indexing & Dynamic Matching using Single-Field Indexer
	PDF Contract Entity Extraction & Matching
	S&P500 Company Lookup with Approximate Phrase Matcher
	String Indexing and Matching in KNIME

Public