Quantity Knowledge Extraction and Search

Abstract

Vinh Thinh Ho is an applied scientist at Amazon Development Center, working in Alexa AI-NLU team. He completed his PhD at Max Planck Institute for Informatics, under the supervision of Prof. Gerhard Weikum. His research broadly covers the area of semantic web, with main focuses on knowledge bases, quantity search, information extraction and retrieval, rule mining and NLP. This research conducted by Vinh Thinh Ho developed new methods for the extraction and search of quantity knowledge over web contents. Quantities are more than mere numbers. They represent measurements of various entities in the world, such as the heights of buildings, the timings of athletes, the energy efficiency of car models, or the energy production of power plants, all expressed in numbers with associated units. While modern search engines effectively support entity-centric searches and question answering, they struggle when the queries involve quantity filters, such as searching for athletes who ran 200m under 20 seconds or companies with quarterly revenue above $2 Billion. These systems fail to understand the quantities, including the condition (less than, above, etc.), the unit of interest (seconds, dollar, etc.), and the context of the quantity (200m race, quarterly revenue, etc.). QA systems based on structured knowledge bases also fail as quantities are poorly covered by state-of-the-art KBs. We developed new methods to overcome these limitations and advance the state-of-the-art on quantity knowledge extraction and search.