Inter-GPS: Interpretable Geometry Problem Solving with Formal Language and Symbolic Reasoning
8057Authors
nan
Abstract
AbstractGeometry problem solving has attracted much attention in the NLP community recently. The task is challenging as it requires abstract problem understanding and symbolic reasoning with axiomatic knowledge. However, current datasets are either small in scale or not publicly available. Thus, we construct a new large-scale benchmark, Geometry3K, consisting of 3,002 geometry problems with dense annotation in formal language. We further propose a novel geometry solving approach with formal language and symbolic reasoning, called Interpretable Geometry Problem Solver (Inter-GPS). Inter-GPS first parses the problem text and diagram into formal language automatically via rule-based text parsing and neural object detecting, respectively. Unlike implicit learning in existing methods, Inter-GPS incorporates theorem knowledge as conditional rules and performs symbolic reasoning step by step. Also, a theorem predictor is designed to infer the theorem application sequence fed to the symbolic solver for the more efficient and reasonable searching path. Extensive experiments on the Geometry3K and GEOS datasets demonstrate that Inter-GPS achieves significant improvements over existing methods. The project with code and data is available at https://lupantech.github.io/inter-gps.
Keywords
Full Content
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 632–643 August 1–6, 2021. ©2021 Association for Computational Linguistics 632 Modeling Language Usage and Listener Engagement in Podcasts Sravana Reddy Spotify sravana@spotify.com Mariya Lazarova Spotify mimiglazarova@gmail.com Yongze Yu Spotify yongzey@spotify.com Rosie Jones Spotify rjones@spotify.com Abstract While there is an abundance of popular writ- ing targeted to podcast creators on how to speak in ways that engage their listeners, there has been little data-driven analysis of pod- casts that relates linguistic style with listener engagement. In this paper, we investigate how various factors – vocabulary diversity, dis- tinctiveness, emotion, and syntax, among oth- ers – correlate with engagement, based on analysis of the creators’ written descriptions and transcripts of the audio. We build mod- els with different textual representations, and show that the identified features are highly pre- dictive of engagement. Our analysis tests pop- ular wisdom about stylistic elements in high- engagement podcasts, corroborating some as- pects, and adding new perspectives on others. 1 Introduction What makes a particular podcast broadly engaging? As a media form, podcasting is new enough that such questions are only beginning to be understood (Jones et al., 2021). Websites exist with advice on podcast production, including language-related tips such as reducing filler words and disfluencies, or incorporating emotion, but there has been little quantitative research into how aspects of language usage contribute to overall listener engagement. This paper investigates the linguistic factors that correlate with engagement, leveraging the written descriptions of the parent show and episode as well as the transcript of the audio. Our metric of en- gagement is stream rate, which we define as the proportion of first-time listeners – of those who have begun streaming the episode – who listen for at least five minutes. Notably, stream rate is dif- ferent from the metric of popularity as given by the raw number of streams; the latter is inevitably influenced by factors unrelated to the content, such as the host or publisher reputation, publicity, expo- sure in recommendations and search engines, and time of publication, whereas a listener’s decision to continue listening for as long as five minutes is likely to be influenced by the content. We perform a series of descriptive tests to ex- amine differences in language usage between high and low engagement podcasts, and build predictive models. Our tests show that while much of the conventional wisdom on engaging podcasting style (such as to use positive language) bears out in the data, other assumptions (such as to speak slowly) are contradicted and deserve a closer look. We find that stylistic features tend to be more correlated with engagement for podcasts with low absolute numbers of streams than for the most popular pod- casts, suggesting that listeners may be less sensitive to style in podcasts made by well-known creators. We also identify those linguistic factors that corre- late with our engagement metric across the popular- ity spectrum, and …
Content truncated to 500 words
Labeled Text
AbstractGeometry problem solving has attracted much attention in the NLP community recently. The task is challenging as it requires abstract problem understanding and symbolic reasoning with axiomatic knowledge. However, current datasets are either small in scale or not publicly available. Thus, we construct a new large-scale benchmark, Geometry3K, consisting of 3,002 geometry problems with dense annotation in formal language. We further propose a novel geometry solving approach with formal language and symbolic reasoning, called Interpretable Geometry Problem Solver (Inter-GPS). Inter-GPS first parses the problem text and diagram into formal language automatically via rule-based text parsing and neural object detecting, respectively. Unlike implicit learning in existing methods, Inter-GPS incorporates theorem knowledge as conditional rules and performs symbolic reasoning step by step. Also, a theorem predictor is designed to infer the theorem application sequence fed to the symbolic solver for the more efficient and reasonable searching path. Extensive experiments on the Geometry3K and GEOS datasets demonstrate that Inter-GPS achieves significant improvements over existing methods. The project with code and data is available at
Publication Details
- Paper ID:
- 8057
- Year:
- 2021
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan