Phrase Indexing and Matching Overview Example

📂 1️⃣ Load the Data

Load the reference product names and the query product names used in the overview example.
The reference table will be indexed, while the query table will be searched against that phrase index.

🧱 2️⃣ Build Index

Build a searchable phrase index from the reference product names.
The Phrase Indexer tokenizes multi-word strings and stores them in an Index Object for efficient phrase retrieval.

🔐 How to Get Your License

Use this node to request and register your exorbyte matchmaker license before running any toolbox nodes.

1. Choose Demo (30 days) or Production.

2. Enter your email (and Customer Token if production).

3. Execute the node — it sends a secure request to exorbyte team.

4. If offline, manually email the request file toknime-node-license@exorbyte.com.

5. When you receive the .lic file, reopen the node → Use available license fileand run the node → run License Activator.

⚠️ Each KNIME installation or Hub environment needs its own license.

👉 See full workflow guide: How to license exorbyte Extension

Configurations

• Select column to index: chooses the phrase column that should be turned into a searchable index.

• Delimiter: defines how the phrase is split into tokens, usually by blank space.

• Index type: controls whether downstream results use the original phrase form or a normalized representation.

• Character Mapping Object: optionally normalizes the input before indexing to improve consistency.

• Alias Object: optionally expands or canonicalizes terms to improve matching across known variants.

🔎 3️⃣ Match

Search the indexed product names using the query phrases and compare the returned matches.

Configurations

• Max. Quality Spread: Maximum difference to best Match Value

• Min. Match Value: Minimum Match Value of indexed terms. Lower the value, higher the tolerance

🔁 Improve Search with Aliases

This section shows how an Alias Object can improve phrase search by expanding known term variants before matching.
Using the Alias Creator, users can define mappings such as "FE" → "Fan Edition" to make product searches more robust and improve hit rates for abbreviated or alternative phrases.

Search for:

• "Samsung S23 Fan Edition"

• "Bose QC Ultra"

Bidirectional Alias Mapping:

• "FE" ↔ "Fan Edition"

• "QC" ↔ "QuietComfort"

🌍 Improve Search with Character Mapping & Aliases

This section shows how Character Mapper and Alias Generator can improve phrase search for multilingual names.

Bidirectional Alias Mapping:

• "Liz" ↔ "Elizabeth"

• "Bob" ↔ "Robert"

• "Bill" ↔ "William"

Character Mapping

• Deaccentuate: enabled
  removes accents such as José → Jose

• Any to Latin: enabled
  transliterates non-Latin characters into Latin equivalents

• Map Spaces: enabled
  normalizes inconsistent whitespace