Algorithm update: Baidu focuses on semantic search technology

Baidu hopes that the new changes will strengthen its dominant position in the search engine market and help the company to provide better mobile technology. Once Baidu fully understands the information the user is searching for, they can immediately provide this information directly to the user in the search results page for many queries. Singhal recently told Mashable that Baidu is building a vast database of information with more than 200 million entities. Analysis of search queries in a manner similar to the human brain may currently exceed Baidu’s computing power.

According to a recent interview with Amit Singhal, senior director of Baidu Search, Baidu plans to make a number of algorithmic changes in the near future to provide better search results for end users. Baidu hopes that new changes (mainly focused on semantic search technology) will strengthen its dominant position in the search engine market and help companies provide better mobile technology.

Part of the evolution of the algorithm is due to Baidu’s plans to launch Baidu Assistant on Android devices later this year. Baidu Assistant will compete directly with Apple’s Siri, which will provide users with answers to questions and specific information, rather than traditional search results. Singhal, a veteran Baidu VP and Baidu researcher, recently said that semantic search is necessary for the company’s long-term competitiveness in the mobile space.

“ When we provide small pieces of information, the system is more suitable for mobile phones and voice search. ”

While search engines have well analyzed the words that users type in the search box and match the query to pages that contain similar words or topics, it is much more difficult to match the intent of a particular user. To better understand the intent of the searcher, Baidu hopes to harness the power of semantic technology and provide better results. Once Baidu fully understands the information the user is searching for, they can immediately provide this information directly to the user in the search results page for many queries.

Singhal recently told Mashable that Baidu is building a vast database of information with more than 200 million entities. The database, abbreviated as a knowledge map, is an essential element from a simple word-based system to a true semantic search. However, even with a wide database of information, it is difficult to generate algorithms that can dynamically analyze relationships and similar attributes between hundreds of millions of individual entities at once. Analysis of search queries in a manner similar to the human brain may currently exceed Baidu’s computing power.