The integration of Baidu Search and LLMs today is providing users with even greater value. AI has redefined search, transforming Baidu Search from a simple list of web results based on text input to an intelligent search engine that can "listen and see." Its understanding and adaptability to user queries have continuously improved, making its content and services more accurate and diverse, and it has become increasingly user-friendly. In this session, three engineers from Baidu Search will each discuss their explorations into the integration of LLMs, the practical applications of generative question-answering, and the open-source retrieval engine PUCK. We welcome everyone to join our conversation.

Time: 13:30-14:10

Title: The LLM is refactoring the search engine

Abstract : The advent of large language models has opened up vast new possibilities for search engines, significantly enriching the quality of search query responses due to their advanced comprehension and reasoning abilities. In this presentation, we'll take a broad look at how Baidu Search harnesses the power of Large Language Models, with a special emphasis on the LLM, to enhance and reshape its search engine functionality. Additionally, we'll explore the potentially revolutionary shifts and emerging trends this technology may introduce in the future.

Speaker: Haibo currently serves as the Tech Lead of Intelligent Question-Answering Technology for Baidu Search and is responsible for the application of large-scale models in search. He graduated from the EECS program at Peking University. Haibo has an in-depth expertise in machine learning, natural language processing, question answering, as well as in the pre-training and fine-tuning applications of large generative models.

Time: 14:10-14:50

Title: Application of Generative Question Answering in Search Engine

Abstract : This report primarily introduces the generative question-answering application based on search results. With the development of generative LLMs, search engine also embraces significant changes. For the questions posed by users in the search engine, by organizing multiple search results, we can provide accurate, comprehensive, and efficient answers, thereby achieving one-shot satisfaction. However, in the search scenario, there are many challenges, such as the accuracy of search reference results, the credibility of generated results, time constraints in search engine, etc. The report introduces a series of solutions to these challenges.

Speaker: Xiaodong is Baidu senior R & D engineer. He received Ph.D degree in Computer Science from Peking University. His research interests include natural language processing, question answering and dialogue system. He has published articles on AAAI, IJCAI, ACL, etc. Currently，he mainly focuses on large language model and question answering in Baidu Search.

Time: 14:50-15:20

Title: PUCK：Open Source Search Engine

Abstract : In this talk, Spearker will introduce and summarize Approximate nearest neighbors(ANN) search methods. Then, introduce Puck which is an open source search engine library developed by Baidu. Puck has ranked first in multiple billion-scale datasets in the Big-ANN competition and significantly outperforms its competitors on multiple data sets including millions, billions, and billions. Puck is used extensively in Baidu’s services, which supports trillions of index size and massive search requests.

Speaker: Jie is Baidu senior R & D engineer. She is responsible for the optimization and application of Puck.