Fuzzy Matching

Fuzzy matching is a newly introduced feature in SEMINE's Purchase Order (PO) service, designed to enhance accuracy and efficiency. This feature employs advanced fuzzy logic algorithms to match invoice and PO line descriptions with greater flexibility, going beyond simple case-insensitive string equations. By considering variations in word order and eliminating the need for exact matches, fuzzy matching significantly improves matching quality and reduces manual intervention.

Key Features

1. Enhanced Matching: Fuzzy matching enables the system to identify similarities between descriptions, even when they are not exact matches. This allows for a broader range of matches, improving overall accuracy.

2. Improved Efficiency: With fuzzy matching, users no longer need to manually review and adjust PO line descriptions for exact matches. The system intelligently identifies close matches, streamlining the matching process.

3. Increased Automation: By automating the matching of descriptions, fuzzy matching reduces the need for manual intervention, saving time and resources.

How It Works:

Fuzzy matching operates by analyzing the text in the Description field of both invoices and PO lines. Instead of relying solely on exact matches, the system considers various factors such as word order, variations in text, and semantic similarities to determine matches. This algorithm accounts for various factors such as misspellings, different word orders, missing words, letters, or numbers, as well as differences in Latin-based languages. 

Example:

Previous Behavior: "10 L paint" matched only with "10 L paint."
New Fuzzy Matching: "10 L blue paint", "Paint 10 L" and "Paint 10 Liters" are now recognized as close matches, providing more automatic matches.

This expanded matching capability improves the accuracy of the PO service and reduces manual effort.

Illustration

Let's take a look at an example to understand how fuzzy matching works in SEMINE. Imagine we're comparing a purchase order with an invoice. In this scenario, the matching process takes into account two key aspects: basic parameters and the Description text.  Basic parameters like Net amount, quantities, VAT %, and unit price are initially factored in, contributing to a baseline matching quality score of 3.3%.

Meanwhile, variations in the Description text, which outlines the purchased items or services, play a crucial role in the matching process. The screenshots highlight these variations, demonstrating their impact on matching quality.

1. 3.3% Match Quality

2. 24.6% Match Quality

3. 42.2% Match Quality

4. 51.6% Match Quality

5. 55.6% Match Quality

6. 57.7% Match Quality

7. 70% Match Quality

8. 70% Match Quality

The example illustrates how incremental changes in the Description text, such as adding or modifying words, influence overall matching quality.

Even minor adjustments can significantly enhance matching accuracy, showcasing the effectiveness of fuzzy matching in identifying close matches and streamlining the procurement process.

Note: The algorithm only accepts text written in the Latin alphabet.

By utilizing fuzzy matching, SEMINE's system can recognize similarities between descriptions, even when they are not exact matches. 

See Also - Purchase Order Matching