ML Models & Algorithms
Metaphone
In shortA phonetic algorithm that encodes words by how they sound in English, so differently spelled names that sound alike can be matched.
Metaphone is a phonetic matching algorithm that turns a word into a short code based on how it sounds when spoken in English. Words that sound alike get the same code even when they are spelled differently, so “Lazarina”, “Lazareena” and “Lazarena” can be recognised as the same name. It was published by Lawrence Philips in 1990 as an improvement on the older Soundex algorithm.
How it works
Metaphone applies a set of pronunciation rules to letter combinations. It keeps the first letter’s sound, drops most vowels, and maps groups such as “PH” to F or a hard “C” to K. Two strings match when their codes match, regardless of spelling. Double Metaphone, published in 2000, returns a primary and a secondary code and handles more spelling patterns from other languages that appear in English, such as Slavic, Germanic or Spanish names.
In Python, the jellyfish library provides Metaphone alongside Soundex and NYSIIS.
How it’s used in SEO and marketing
- Brand monitoring. Catch brand mentions that are misspelled or spelled phonetically in social posts and reviews.
- Author and people names. Group variants of the same person’s name across mentions, citations and CRM records.
- Voice and transcripts. Match terms that speech-to-text tools have spelled the way they sound.
- Data clean-up. Deduplicate customer or company names before analysis.
Example
In the notebook that accompanies MLforSEO’s fuzzy matching guide, “Lazarina”, “Lazareena” and “Lazarena” all collapse to the same Metaphone code. Running a list of social mentions through Metaphone and grouping by code surfaces misspelled brand or author mentions that an exact-match search would miss.
Limitations and things to watch
- Precision is low. Many unrelated words share a code, so phonetic matching returns lots of false positives. MLforSEO recommends pairing it with an edit-distance check such as Levenshtein to rank candidates.
- The original algorithm is built around English pronunciation; use Double Metaphone or a language-specific method for names of other origins.
- Short words collide often because their codes are only a few characters long.
- It matches sound, not meaning, so synonyms are invisible to it.
Learn it on MLforSEO
Where Metaphone shows up on MLforSEO
Learn it in a course
Read about it on the blog
Sources and further reading
Other terms from ML Models & Algorithms
- BERT (Bidirectional Encoder Representations from Transformers)Google's 2018 transformer language model that reads text in both directions to produce context-aware embedding…
- BERTopicA modular topic modelling technique that clusters transformer embeddings and labels each cluster with class-ba…
- BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies)BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies) clusters very large numeric datasets by s…
- BM25A classic ranking function that scores documents by how often query words appear, adjusted for word rarity and…
- Boyer-MooreA fast exact string-search algorithm that compares a pattern from right to left and skips ahead after mismatch…
- c-TF-IDFClass-based TF-IDF: treats each cluster as one document to find the words that best describe it, used by BERTo…
- Centroid-based ClusteringA clustering family that groups data around central points (centroids), each the average of its cluster's memb…
- Convolutional Neural Network (CNN)A deep learning architecture that scans data with small learned filters to detect local patterns; the classic…
- DBSCANDensity-based clustering that groups closely packed points, finds the number of clusters itself and labels iso…