NAVER Search Unveils “Multimodal Document Search” based on Multimodal AI Model
NAVER Search Unveils “Multimodal Document Search” based on Multimodal AI Model
NAVER Search Unveils “Multimodal Document Search” based on Multimodal AI Model
- Applied to sneakers search first, sneaker product name, reviews, and styling information all ready with just a few image clicks
- The “Smart thumbnail” technology enhances search convenience by making users understand the search results at a glance
December 2, 2022
NAVER Search is launching a new search service called “multimodal document search,” which will integrate “OmniSearch,” a multimodal AI model developed by NAVER Search. NAVER developed “OmniSearch” in March to deliver search results closely matching the user-entered keywords by learning big data from its services used by millions of people, such as NAVER Blog, Café, Shopping, Knowledge iN, and News.
Multimodal document search utilizes an AI model that combines image and text inputs to quickly analyze what the user is searching for. This analysis helps to provide personalized search results by sorting relevant documents. Multimodal AI is different from traditional search methods that only rely on one source of input such as text, image, or voice. By combining both image and text, multimodal AI increases the convenience of customized searches and is considered the key technology for future search methods.
Multimodal document search was first applied to the sneakers category. Even if the user doesn’t know the exact product name, they can still find the name of a product with just an image. Additionally, they can view reviews of the product from other users and receive styling information. From the vast amount of user content stored in NAVER, the service filters and presents quality documents that match the search target image with the “reviews and styles searched by image” block.
In addition, the “smart thumbnail” technology was implemented to enhance the visibility of the documents provided in search results. This technology displays the most relevant image that corresponds to the user's search intent as the thumbnail image for each document. This allows the user to quickly get an idea of the content without having to check every document individually.
The newly launched search service is available within NAVER’s image search service “Smart Lens.” For instance, users can click on the “Smart Lens” icon located at the bottom of the image in NAVER’s image search results or upload an image of sneakers and click on “View integrated search results” at the bottom of the image. Afterward, they can explore additional information by viewing the “reviews and styles searched by image” search results provided through multimodal document search.
After introducing Smart Lens with the multimodal AI capability in April, NAVER has enhanced its usability to come up with multimodal document search this time. Going forward, NAVER plans to apply the multimodal AI model to other NAVER search services, such as shopping, and create a distinctive search environment.
NAVER Search CIC leader Kang In-ho said, “Multimodal AI, which is built upon the data accumulated on NAVER, is providing users with a new search experience. Multimodal document search is a newly introduced feature that has been developed based on the positive user response to the earlier introduction of multimodal-based search service. Initially, it was applied to the sneakers category with plans to improve its usability and gradually expand its use to other categories.”
-END-