Verb Physics: Relative Physical Knowledge of Actions and Objects

Abstract

Learning commonsense knowledge from natural language text is nontrivial due to reporting bias: people rarely state the obvious, e.g., "My house is bigger than me." However, while rarely stated explicitly, this trivial everyday knowledge does influence the way people talk about the world, which provides indirect clues to reason about the world. For example, a statement like, "Tyler entered his house" implies that his house is bigger than Tyler.In this paper, we present an approach to infer relative physical knowledge of actions and objects along five dimensions (e.g., size, weight, and strength) from unstructured natural language text. We frame knowledge acquisition as joint inference over two closely related problems: learning (1) relative physical knowledge of object pairs and (2) physical implications of actions when applied to those object pairs. Empirical results demonstrate that it is possible to extract knowledge of actions and objects from language and that joint inference over different types of knowledge improves performance.

Do NLP Models Know Numbers? Probing Numeracy in Embeddings

“…More generally, numeracy is one type of emergent knowledge. For instance, embeddings may capture the size of objects (Forbes and Choi, 2017), speed of vehicles, and many other "commonsense" phenomena (Yang et al, 2018). Vendrov et al (2016) introduce methods to encode the order of such phenomena into embeddings for concepts such as hypernymy; our work and Yang et al (2018) show that a relative ordering naturally emerges for certain concepts.…”

Modeling Semantic Plausibility by Injecting World Knowledge

“…relative sizes in man-swallowpaintball/desk). Previously, Forbes and Choi (2017) proposed a three level (3-LEVEL) featurization scheme, where an object-pair can take 3 values for, e.g. relative size: {−1, 0, 1} (i.e.…”

“…Physical world knowledge modeling appears frequently in more closely related work. Bagherinezhad et al (2016) combine computer vision and text-based information extraction to learn the relative sizes of objects; Forbes and Choi (2017) crowdsource physical knowledge along specified dimensions and employ belief propagation to learn relative physical attributes of object pairs. Wang et al (2017) propose a multimodal LDA to learn the definitional properties (e.g.…”

Commonsense mining as knowledge base completion? A study on the impact of novelty

“…Knowledge extraction from text corpora is a vast research area (Banko et al, 2007;Mitchell et al, 2015), yet works that specifically target commonsense knowledge are comparatively rare (Gordon, 2014). Our focus is on the specific approach to mining commonsense knowledge by casting it as a KBC task, as in Li et al (2016) and Forbes and Choi (2017).…”