What algorithms are used by Delpha

Explore the advanced duplicate detection algorithms Delpha uses for string, email, phone, and address fields. Learn how Delpha ensures accurate, AI-powered record matching.

What Algorithms Are Used by Delpha to Detect Duplicates?

Delpha uses a set of advanced matching algorithms and normalization techniques to identify duplicate records across different field types in Salesforce. The logic determines how values are compared between records to calculate similarity and duplication scores.

Matching Algorithms & Normalization by Field Type

Field Type	Matching Algorithm	Normalization
String	Exact, Jaro-Winkler, QGram Tokenizer, TF-IDF	Lowercase, ASCII table
Boolean	Exact	None
Email	Exact, Partial segmentation, Domain name average	Lowercase, Validity checks
Phone	Exact	ASCII table, e164 normalization
Number	Exact	Rounded
URL	Exact	ASCII & percent-encoded triplet, Path ending normalization ("/"), Protocol normalization
Address	Partial, Euclidian distance	Coordinate normalization
Coordinates	–	Coordinate normalization

Why It Matters

Delpha’s duplicate detection engine uses both exact and fuzzy matching algorithms to catch near-duplicates often missed by basic tools. For example:

• Emails are split and scored by domain and format.

• URLs are normalized to avoid false negatives caused by minor syntax differences.

• Addresses use Euclidian distance to detect close matches even with typos or format variations.