Deep parsing in Watson

Abstract

Two deep parsing components, an English Slot Grammar (ESG) parser and a predicate-argument structure (PAS) builder, provide core linguistic analyses of both the questions and the text content used by IBM Watsoni to find and hypothesize answers. Specifically, these components are fundamental in question analysis, candidate generation, and analysis of passage evidence. As part of the Watson project, ESG was enhanced, and its performance on Jeopardy!i questions and on established reference data was improved. PAS was built on top of ESG to support higher-level analytics. In this paper, we describe these components and illustrate how they are used in a pattern-based relation extraction component of Watson. We also provide quantitative results of evaluating the component-level performance of ESG parsing.

Strong Cognitive Symbiosis: Cognitive Computing for Humans

“…Every leading technology company has declared that "it's superior AI" are key to its continued success 11 . Russia's president Vladimir Putin has publicly declared that whoever masters AI will "rule the world" 12 .…”

“…[11]) rather than statistical models. In fact, even Watson uses a structured lexicon in question analysis and candidate generation [12].…”

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)

“…The Predicate Lexicon (PL) was constructed by parsing manually-detected claims (Aharoni et al, 2014) using the Watson ESG parser (McCord et al, 2012), and considering those which have exactly one verb. Then the verb and a concatenation of its right-modifiers, termed here the predicate, were extracted from each claim and added to the PL if they contained at least one sentiment word from the sentiment lexicon of .…”

“…We determine the topic's plurality using the Watson parser (McCord et al, 2012), and do the surface realization with SimpleNLG (Gatt and Reiter, 2009) …”

A Tool for the Semantic Analysis and Recommendation of videos in e-learning

“…The Watson algorithms can mine both structured and unstructured data, extracting information and originating new knowledge, which results in the emergence of multiple answers that are weighted through a confidence estimation. The key feature required by the competition was mastering the language [14] because the game proposes clues full of irony, subtle meanings, and other complexities allowed by a malicious use of natural language [15]. For what concerns the underlying technology, we report that Watson owes its smartness to the DeepQA Project [16] a massively parallel probabilistic evidence-based architecture, which can also be adapted to different business applications and additional exploratory challenge problems including medicine, enterprise search, and gaming [17].…”