From Sneaker Photos to Product Names and Outfit Info... Naver Applies 'Multimodal AI' to Search

Naver has introduced a new search service called "Multimodal Document Search" utilizing its self-developed multimodal artificial intelligence (AI) "OmniSearch."
[Photo by Naver]

[Asia Economy Reporter Yuri Choi] Naver announced on the 2nd that it will launch a new search service called ‘Multimodal Document Search’ utilizing its self-developed multimodal artificial intelligence (AI) ‘OmniSearch’.

‘Multimodal Document Search’ quickly analyzes the user’s search intent input through multimodal AI using images, text, and more. Based on the analysis results, it classifies documents with high matching relevance and provides them as personalized search results. Unlike the existing method of searching with a single means such as text, image, or voice, it combines images and text to enhance user-customized search convenience.

The new search feature was first applied to the sneaker product category. By using Multimodal Document Search, even if users do not know the exact product name of the sneakers they are looking for, they can find the precise product name using only images. Users can also check reviews left by others and styling information using the product. Among the vast user content accumulated on Naver, high-quality documents with high relevance to the searched image are selected and provided in the ‘Reviews & Coordination Found by Image’ block.

The ‘Smart Thumbnail’ technology, which enhances the visibility of documents provided as search results, was also applied. It extracts the most suitable image for the user’s search intent from the images included in the target document and displays it as the representative image. This allows users to gauge the content by referring only to the thumbnail without checking all documents.

Naver first unveiled a search function incorporating multimodal AI in its image search service Smart Lens last April, and now it has additionally introduced Multimodal Document Search. In the future, it plans to expand the use of multimodal technology across overall search, including shopping search.

Kang In-ho, Head of Naver Search CIC, said, “Multimodal Document Search is a newly introduced feature following the favorable user response to the previously unveiled multimodal-based search service. We first introduced it to the sneaker product category to enhance usability and plan to gradually expand its application range.”

This content was produced with the assistance of AI translation services.