AI SEO Weekly

Multimodal Search

By @ValeIzzo86 & @CyberAndy

In traditional keyword search, while we might be able to recognize plurals, variations and stemming every query must still account for every term and permutation.

How keyword search works 🔎

So when I search for “white sneakers”  (and I don’t have a product with that name)  I will get this 😱

results on a WooCommerce website using WordPress internal search.

Here comes Semantic Search 🧠

Semantic (or Neural) Search helps us surface the most relevant results based on search intent and not just keywords.

Let's try again “white sneakers”  (same WooCommerce store ⚡️)

results using Neural Search by

A multimodal search engine retrieves relevant content using similarity vectors trained on different modalities such as text, images, audio and video.

..what about Multimodal Search?

Let's try now with an image  (same WooCommerce store ⚡️)

results using Neural Search by

Multimodal Search, much like humans do, is able to encompass vision, auditory, and language understanding simultaneously When we give the photo of a pair of sunglasses and we say “in pink” it will find what we are really looking for by combining two modalities. 

Why Multimodal Search Is Important

Can we enable semantic search across multiple languages?

Multimodal Search Multilingual


We worked to validate the results of a multilingual and multimodal  search engine.