Rapid response translingual information access in the alias-i interface
Abstract
Our alias-i interface provides a limited class of translingual information access for English speakers in very short period of time. Alias-i consists of three components: An interface for bootstrapping recognition algorithms, a backend indexing and annotation algorithm that processes data, and an information access interface for end users. The system is built upon the linguistic phenomenon of coreference. Essentially, coreference abstracts over superficial differences in how a person, place, concept or event is represented in text, e.g. coreference recognizes that 'Mulla Omar', 'Omar', 'he', his name in Somali all refer to the same person. Coreference is a cognitively natural organizing principle which is an advancement over traditional information access approaches like keyword indexed data (i.e. search engines).
Related Papers
- → A deep neural network model for speakers coreference resolution in legal texts(2020)37 cited
- Learning the Information Status of Noun Phrases in Spoken Dialogues(2011)
- → Gesture Features for Coreference Resolution(2006)4 cited
- → Visual Coreference Resolution in Visual Dialog using Neural Module Networks(2018)