- October 21, 2026
- By Diya Sharma M.S. ’27
A team of computer scientists at the University of Maryland and Google Research has developed an algorithm able to build a search index for 1 billion items in under 20 minutes on a single computer, cutting the time it takes to build this crucial AI infrastructure from hours or even days.
Faster construction of the search index — a backend map that allows the AI system to quickly find relevant information — could lower computing costs, improve energy efficiency and help systems keep pace with constantly changing information.
“If your dataset changes every hour, and it takes you more than an hour to build the index, you literally can’t use that index,” said Tobias Rubel, who developed the algorithm known as PiPNN with fellow doctoral student Richard Wen and adviser Laxman Dhulipala.
Dhulipala is an assistant professor of computer science with an appointment in the University of Maryland Institute for Advanced Computer Studies (UMIACS) and is also a research scientist with the graph mining team at Google Research.
PiPNN addresses a challenge known as “approximate nearest-neighbor search.” When a user asks an AI to find related documents, recommend a video or answer a question, the system can find relevant information by treating data as points on a map and searching for nearby matches.
But an AI doesn’t just look at two dimensions like latitude and longitude. It maps data across hundreds or thousands of dimensions that can capture characteristics such as an article’s topic, tone and vocabulary. To find the “closest” matches quickly, the system must first build a web of connections linking similar items. While these digital roadmaps enable fast searches, building them is a notoriously slow process.
PiPNN divides the massive dataset into overlapping local neighborhoods, which the computer can process in efficient batches. To prevent too many overlapping connections in dense neighborhoods from slowing the system down, the team developed a memory-saving method called HashPrune, which processes incoming connections and removes redundant ones, retaining useful, diverse paths without overloading the machine’s memory.
Tests on a powerful cloud computing machine demonstrated PiPNN’s ability to handle large datasets; average construction speedups ranged from sixfold to tenfold, the research team found.
PiPNN won the Best Research Paper Award at the ACM SIGKDD 2026 conference, and the researchers secured first place in the 2026 SISAP Indexing Challenge. Follow-up work also won a best paper award at the Vector Database workshop at the 2026 International Conference on Very Large Data Bases.