A Weak supervision with Syntactic Cues for Reference Resolution
In recipes, contextual understanding of instructions depends on temporal interpretation of the entities because of their spatio-temporal changes. Accordingly, we propose the use of reference resolution to find the origin action of entities, provided that the entity is an output from a previous action, instead of being a raw ingredient. Here, we introduce a weak supervision method that supports the syntactic features for producing latent links between entities and their origin actions. The results show that our weak supervision outperforms the previous unsupervised studies with \%8 F1. In particular, our approach indicates \%82 resolution performance for pronouns and \%85 for null entities.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
A Weak Self-supervision with Transition-Based Modeling for Reference Resolution
The reference resolution is a task to find the link between an entity and its source action in the same recipe. In this study, we introduce a weak self-supervision method with a transition-based model for reference resol…
Entity ResolutionFinding "It": Weakly-Supervised Reference-Aware Visual Grounding in Instructional Videos
Grounding textual phrases in visual content with standalone image-sentence pairs is a challenging task. When we consider grounding in instructional videos, this problem becomes profoundly more complex: the latent tempora…
Multiple Instance LearningSentenceVisual GroundingRGAT: A Deeper Look into Syntactic Dependency Information for Coreference Resolution
Although syntactic information is beneficial for many NLP tasks, combining it with contextual information between words to solve the coreference resolution problem needs to be further explored. In this paper, we propose …
coreference-resolutionCoreference ResolutionGraph AttentionA Simple Unsupervised Approach for Coreference Resolution using Rule-based Weak Supervision
Labeled data for the task of Coreference Resolution is a scarce resource, requiring significant human effort. While state-of-the-art coreference models rely on such data, we propose an approach that leverages an end-to-e…
coreference-resolutionCoreference ResolutionA Simple Unsupervised Approach for Coreference Resolution using Rule-based Weak Supervision
Labeled data for the task of Coreference Resolution is a scarce resource, requiring significant human effort. While state-of-the-art coreference models rely on such data, we propose an approach that leverages an end-to-e…
coreference-resolutionCoreference Resolution