Bing’s latest ‘Deep Search’ functionality delivers more extensive responses to intricate search inquiries.
Microsoft Bing is introducing a novel “Deep Search” feature, powered by OpenAI’s GPT-4. This innovative feature aims to provide users with more pertinent and extensive responses to intricate search queries. It’s important to note that Microsoft positions Deep Search as a supplement to Bing’s existing web search functionality, rather than a replacement, enabling users to delve deeper into web exploration.
In a blog post, Microsoft elucidates that this new feature builds upon Bing’s current web index and ranking system, augmenting them with the capabilities of GPT-4. GPT-4 takes the user’s search query and transforms it into a more comprehensive description of the desired search results.
For instance, consider a user seeking information on loyalty programs in various countries and entering the query: “how do points systems work in Japan.” Deep Search would expand this query into the following:
- Provide an in-depth explanation of how different loyalty card programs operate in Japan, encompassing their benefits, prerequisites, and constraints. Include examples of well-known loyalty cards from diverse categories such as convenience stores, supermarkets, and restaurants.
- Present a comparison of the pros and cons of using loyalty cards versus alternative payment methods in Japan, including current rewards and advantages.
- Highlight the most popular services and participating merchants related to loyalty programs in Japan.
This expanded description empowers users to convey their intent more comprehensively compared to using just a few words.
In situations where a search query is ambiguous, Deep Search identifies all potential intents and crafts a comprehensive description for each of them. These multiple intents are then presented to the user, allowing them to select the most appropriate one.
Once the expanded description is created, Bing retrieves pertinent results that may not typically appear in standard search outcomes. Deep Search identifies pages that potentially align with the expanded query, reformulates the query, and subsequently explores these variations as well.
Continuing with the loyalty points query example mentioned earlier, Deep Search might also search for terms like “loyalty card programs Japan,” “best loyalty cards for travelers in Japan,” “comparison of loyalty programs by category in Japan,” “redeeming loyalty cards in Japan,” and “managing loyalty points with phone apps.”
“In doing so, Deep Search can unearth results that cover different facets of the query, even if they don’t explicitly contain the original keywords,” as highlighted in Microsoft’s blog post. While regular Bing searches already consider millions of web pages for each query, Deep Search expands this scope tenfold, aiming to deliver results that are more informative and specific than those typically ranked higher in conventional search results.
After collecting web pages, Deep Search ranks them based on their alignment with the expanded description. When evaluating relevance, Deep Search takes into account factors such as the topic’s correspondence to the query, the depth of information provided, the page’s credibility, and its recency and popularity.
Microsoft emphasizes that Deep Search is an optional feature and may take up to 30 seconds to complete. Therefore, it is not intended for every query or user. Users who prefer quick and concise answers can still receive regular Bing search results instantaneously.
