US 12,450,268 B2
Efficient search query auto-completion from unstructured text
Jia Xiang Lim, Singapore (SG); En Zhi Tan, Singapore (SG); Benjamin Yan Han Yap, Singapore (SG); Benjamin Jin Boon Tan, Singapore (SG); Lenald Ng, Singapore (SG); Min Pyae Moe, Singapore (SG); Pern Piao Calvin Soh, Singapore (SG); and Steven Peng Chen Fu, Singapore (SG)
Assigned to SAP SE, Walldorf (DE)
Filed by SAP SE, Walldorf (DE)
Filed on Jul. 26, 2023, as Appl. No. 18/359,150.
Prior Publication US 2025/0036667 A1, Jan. 30, 2025
Int. Cl. G06F 16/332 (2025.01); G06F 40/232 (2020.01); G06F 40/274 (2020.01); G06F 40/295 (2020.01)
CPC G06F 16/3322 (2019.01) [G06F 40/232 (2020.01); G06F 40/274 (2020.01); G06F 40/295 (2020.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for auto-complete of search queries received by an information retrieval (IR) system, the method being executed by one or more processors and comprising:
receiving a set of descriptions provided as unstructured data, the set of descriptions comprising item descriptions for one or more of goods and services available to be searched using the IR system, each description associated with one or more entities in a set of entities that can be queried using the IR system;
providing, from the set of descriptions, a first set of training data by applying rule-based extraction to the set of descriptions using part-of-speech (POS) tags, the first set of training data comprising at least a first set of entities comprising at least a portion of the set of entities;
training a named-entity recognition (NER) model using at least a portion of the first set of training data;
constructing a ranked trie-tree (RTT) using a set of extracted entities determined from the NER model, each extracted entity having an associated confidence score;
receiving, by the IR system, a portion of a search query; and
processing the portion of the search query using the RTT to provide a set of auto-complete suggestions comprising k auto-complete suggestions determined through order-specific traversal of the RTT that is terminated after finding the k auto-complete suggestions based on confidence scores, during processing, changing the confidence score of an extracted entity to a predefined minimum score within the RTT in response to adding the extracted entity to the set of auto-complete suggestions to prevent the extracted entity being selected as having a highest score in subsequent processing; and
after determining the set of auto-complete suggestions, updating the RTT to restore confidence scores of extracted entities included in the set of auto-complete suggestions.