Pub Date : 2023-01-25DOI: 10.48550/arXiv.2301.10483
Kung-Hsiang Huang, Kung-Hsiang Huang, Siffi Singh, Xiaofei Ma, Wei Xiao, Wei Xiao, Nicholas Dingwall, William Yang Wang, K. McKeown
Missing information is a common issue of dialogue summarization where some information in the reference summaries is not covered in the generated summaries. To address this issue, we propose to utilize natural language inference (NLI) models to improve coverage while avoiding introducing factual inconsistencies. Specifically, we use NLI to compute fine-grained training signals to encourage the model to generate content in the reference summaries that have not been covered, as well as to distinguish between factually consistent and inconsistent generated sentences. Experiments on the DialogSum and SAMSum datasets confirm the effectiveness of the proposed approach in balancing coverage and faithfulness, validated with automatic metrics and human evaluations. Additionally, we compute the correlation between commonly used automatic metrics with human judgments in terms of three different dimensions regarding coverage and factual consistency to provide insight into the most suitable metric for evaluating dialogue summaries.
{"title":"SWING: Balancing Coverage and Faithfulness for Dialogue Summarization","authors":"Kung-Hsiang Huang, Kung-Hsiang Huang, Siffi Singh, Xiaofei Ma, Wei Xiao, Wei Xiao, Nicholas Dingwall, William Yang Wang, K. McKeown","doi":"10.48550/arXiv.2301.10483","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10483","url":null,"abstract":"Missing information is a common issue of dialogue summarization where some information in the reference summaries is not covered in the generated summaries. To address this issue, we propose to utilize natural language inference (NLI) models to improve coverage while avoiding introducing factual inconsistencies. Specifically, we use NLI to compute fine-grained training signals to encourage the model to generate content in the reference summaries that have not been covered, as well as to distinguish between factually consistent and inconsistent generated sentences. Experiments on the DialogSum and SAMSum datasets confirm the effectiveness of the proposed approach in balancing coverage and faithfulness, validated with automatic metrics and human evaluations. Additionally, we compute the correlation between commonly used automatic metrics with human judgments in terms of three different dimensions regarding coverage and factual consistency to provide insight into the most suitable metric for evaluating dialogue summaries.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":"1 1","pages":"512-525"},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46883582","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Hagen‐Zanker, Jingyan Yu, Susan Hughes, N. Santitissadeekorn
Scenarios of future urban expansion are expected to be plausible: they must be diverse to reflect future uncertainty, yet realistic in their depiction of urban expansion processes. We investigated the plausibility of scenarios derived from a novel data-driven simulation approach. In a Turing-like test, experts completed a quiz in which they were asked to identify the map showing true urban expansion amidst three model-generated scenarios. Across diverse expansion patterns, ranging from compact to dispersed, the experts had no significant ability to identify the true pattern. The results support the hypothesis that the investigated scenarios are plausible and hence that cluster analysis of estimated dynamic models is a viable method for producing scenarios of future urban expansion.
{"title":"A Turing Test of the Plausibility of Model-Generated Urban Expansion Scenarios","authors":"A. Hagen‐Zanker, Jingyan Yu, Susan Hughes, N. Santitissadeekorn","doi":"10.32866/001c.68147","DOIUrl":"https://doi.org/10.32866/001c.68147","url":null,"abstract":"Scenarios of future urban expansion are expected to be plausible: they must be diverse to reflect future uncertainty, yet realistic in their depiction of urban expansion processes. We investigated the plausibility of scenarios derived from a novel data-driven simulation approach. In a Turing-like test, experts completed a quiz in which they were asked to identify the map showing true urban expansion amidst three model-generated scenarios. Across diverse expansion patterns, ranging from compact to dispersed, the experts had no significant ability to identify the true pattern. The results support the hypothesis that the investigated scenarios are plausible and hence that cluster analysis of estimated dynamic models is a viable method for producing scenarios of future urban expansion.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"44917619","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-25DOI: 10.48550/arXiv.2301.10439
Cong Dao Tran, Nhut Huy Pham, Anh-Viêt Nguyên, T. Hy, Tu Vu
This paper presents ViDeBERTa, a new pre-trained monolingual language model for Vietnamese, with three versions - ViDeBERTa_xsmall, ViDeBERTa_base, and ViDeBERTa_large, which are pre-trained on a large-scale corpus of high-quality and diverse Vietnamese texts using DeBERTa architecture. Although many successful pre-trained language models based on Transformer have been widely proposed for the English language, there are still few pre-trained models for Vietnamese, a low-resource language, that perform good results on downstream tasks, especially Question answering. We fine-tune and evaluate our model on three important natural language downstream tasks, Part-of-speech tagging, Named-entity recognition, and Question answering. The empirical results demonstrate that ViDeBERTa with far fewer parameters surpasses the previous state-of-the-art models on multiple Vietnamese-specific natural language understanding tasks. Notably, ViDeBERTa_base with 86M parameters, which is only about 23% of PhoBERT_large with 370M parameters, still performs the same or better results than the previous state-of-the-art model. Our ViDeBERTa models are available at: https://github.com/HySonLab/ViDeBERTa.
{"title":"ViDeBERTa: A powerful pre-trained language model for Vietnamese","authors":"Cong Dao Tran, Nhut Huy Pham, Anh-Viêt Nguyên, T. Hy, Tu Vu","doi":"10.48550/arXiv.2301.10439","DOIUrl":"https://doi.org/10.48550/arXiv.2301.10439","url":null,"abstract":"This paper presents ViDeBERTa, a new pre-trained monolingual language model for Vietnamese, with three versions - ViDeBERTa_xsmall, ViDeBERTa_base, and ViDeBERTa_large, which are pre-trained on a large-scale corpus of high-quality and diverse Vietnamese texts using DeBERTa architecture. Although many successful pre-trained language models based on Transformer have been widely proposed for the English language, there are still few pre-trained models for Vietnamese, a low-resource language, that perform good results on downstream tasks, especially Question answering. We fine-tune and evaluate our model on three important natural language downstream tasks, Part-of-speech tagging, Named-entity recognition, and Question answering. The empirical results demonstrate that ViDeBERTa with far fewer parameters surpasses the previous state-of-the-art models on multiple Vietnamese-specific natural language understanding tasks. Notably, ViDeBERTa_base with 86M parameters, which is only about 23% of PhoBERT_large with 370M parameters, still performs the same or better results than the previous state-of-the-art model. Our ViDeBERTa models are available at: https://github.com/HySonLab/ViDeBERTa.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":"1 1","pages":"1041-1048"},"PeriodicalIF":0.0,"publicationDate":"2023-01-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42611792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-23DOI: 10.48550/arXiv.2301.09759
Yamen Ajjour, Johannes Kiesel, Benno Stein, Martin Potthast
Many computational argumentation tasks, such as stance classification, are topic-dependent: The effectiveness of approaches to these tasks depends largely on whether they are trained with arguments on the same topics as those on which they are tested. The key question is: What are these training topics? To answer this question, we take the first step of mapping the argumentation landscape with The Argument Ontology (TAO). TAO draws on three authoritative sources for argument topics: the World Economic Forum, Wikipedia’s list of controversial topics, and Debatepedia. By comparing the topics in our ontology with those in 59 argument corpora, we perform the first comprehensive assessment of their topic coverage. While TAO already covers most of the corpus topics, the corpus topics barely cover all the topics in TAO. This points to a new goal for corpus construction to achieve a broad topic coverage and thus better generalizability of computational argumentation approaches.
{"title":"Topic Ontologies for Arguments","authors":"Yamen Ajjour, Johannes Kiesel, Benno Stein, Martin Potthast","doi":"10.48550/arXiv.2301.09759","DOIUrl":"https://doi.org/10.48550/arXiv.2301.09759","url":null,"abstract":"Many computational argumentation tasks, such as stance classification, are topic-dependent: The effectiveness of approaches to these tasks depends largely on whether they are trained with arguments on the same topics as those on which they are tested. The key question is: What are these training topics? To answer this question, we take the first step of mapping the argumentation landscape with The Argument Ontology (TAO). TAO draws on three authoritative sources for argument topics: the World Economic Forum, Wikipedia’s list of controversial topics, and Debatepedia. By comparing the topics in our ontology with those in 59 argument corpora, we perform the first comprehensive assessment of their topic coverage. While TAO already covers most of the corpus topics, the corpus topics barely cover all the topics in TAO. This points to a new goal for corpus construction to achieve a broad topic coverage and thus better generalizability of computational argumentation approaches.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":"1 1","pages":"1381-1397"},"PeriodicalIF":0.0,"publicationDate":"2023-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41750336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-01-23DOI: 10.48550/arXiv.2301.09685
Ruoyu Xie, Antonios Anastasopoulos
An ongoing challenge in current natural language processing is how its major advancements tend to disproportionately favor resource-rich languages, leaving a significant number of under-resourced languages behind. Due to the lack of resources required to train and evaluate models, most modern language technologies are either nonexistent or unreliable to process endangered, local, and non-standardized languages. Optical character recognition (OCR) is often used to convert endangered language documents into machine-readable data. However, such OCR output is typically noisy, and most word alignment models are not built to work under such noisy conditions. In this work, we study the existing word-level alignment models under noisy settings and aim to make them more robust to noisy data. Our noise simulation and structural biasing method, tested on multiple language pairs, manages to reduce the alignment error rate on a state-of-the-art neural-based alignment model up to 59.6%.
{"title":"Noisy Parallel Data Alignment","authors":"Ruoyu Xie, Antonios Anastasopoulos","doi":"10.48550/arXiv.2301.09685","DOIUrl":"https://doi.org/10.48550/arXiv.2301.09685","url":null,"abstract":"An ongoing challenge in current natural language processing is how its major advancements tend to disproportionately favor resource-rich languages, leaving a significant number of under-resourced languages behind. Due to the lack of resources required to train and evaluate models, most modern language technologies are either nonexistent or unreliable to process endangered, local, and non-standardized languages. Optical character recognition (OCR) is often used to convert endangered language documents into machine-readable data. However, such OCR output is typically noisy, and most word alignment models are not built to work under such noisy conditions. In this work, we study the existing word-level alignment models under noisy settings and aim to make them more robust to noisy data. Our noise simulation and structural biasing method, tested on multiple language pairs, manages to reduce the alignment error rate on a state-of-the-art neural-based alignment model up to 59.6%.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":"1 1","pages":"1471-1483"},"PeriodicalIF":0.0,"publicationDate":"2023-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43632699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper examines changes in e-bike awareness and consideration among commuters to the University of California, Davis campus using data from an annual travel survey. The analysis shows that awareness of e-bikes increased among commuters while consideration declined between 2019 and 2021. Awareness significantly increased among staff and undergraduate students and also increased among those who feel safe biking to campus. Consideration declined significantly among undergraduate students and commuters who bike to campus or use other modes.
{"title":"Changes in E-bike Awareness and Consideration for Commute","authors":"Aakansha Jain, S. Handy","doi":"10.32866/001c.67840","DOIUrl":"https://doi.org/10.32866/001c.67840","url":null,"abstract":"This paper examines changes in e-bike awareness and consideration among commuters to the University of California, Davis campus using data from an annual travel survey. The analysis shows that awareness of e-bikes increased among commuters while consideration declined between 2019 and 2021. Awareness significantly increased among staff and undergraduate students and also increased among those who feel safe biking to campus. Consideration declined significantly among undergraduate students and commuters who bike to campus or use other modes.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49250340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ashima Sharma, Jay Patrikar, Brady G. Moon, S. Scherer, C. Samaras
We quantify and analyze the potential number of flyable hours for an advanced air mobility (AAM) vehicle over the contiguous United States. We use Meteorological Aerodrome Reports (METARs) from 2019, covering 91 airports in the US. By filtering the METARs based on Federal Aviation Administration mandated flight conditions and the vehicle’s physical capabilities, our analysis shows nearly double the amount of annual acceptable flying time between the most flyable and least flyable locations in the country and identifies the largest cause of non-flyable hours as cloud cover. Our work can be used to understand the viability of AAM vehicles in a geographic location.
{"title":"Quantifying the Effect of Weather on Advanced Air Mobility Operations","authors":"Ashima Sharma, Jay Patrikar, Brady G. Moon, S. Scherer, C. Samaras","doi":"10.32866/001c.66207","DOIUrl":"https://doi.org/10.32866/001c.66207","url":null,"abstract":"We quantify and analyze the potential number of flyable hours for an advanced air mobility (AAM) vehicle over the contiguous United States. We use Meteorological Aerodrome Reports (METARs) from 2019, covering 91 airports in the US. By filtering the METARs based on Federal Aviation Administration mandated flight conditions and the vehicle’s physical capabilities, our analysis shows nearly double the amount of annual acceptable flying time between the most flyable and least flyable locations in the country and identifies the largest cause of non-flyable hours as cloud cover. Our work can be used to understand the viability of AAM vehicles in a geographic location.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45071779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The use of the general transit feed specification (GTFS) data standard has spread rapidly since its introduction in 2007, although it is still not universal in the United States. To explain which transit agencies are likely to have been early adopters of GTFS, we estimate a logistic regression model predicting GTFS adoption based on service area and agency characteristics. We find that agencies with higher ridership and those providing lower shares of a region’s total vehicle revenue kilometers have tended to adopt GTFS earlier.
{"title":"Predictors of Early Adoption of the General Transit Feed Specification","authors":"C. Voulgaris, Charuvi Begwani","doi":"10.32866/001c.57722","DOIUrl":"https://doi.org/10.32866/001c.57722","url":null,"abstract":"The use of the general transit feed specification (GTFS) data standard has spread rapidly since its introduction in 2007, although it is still not universal in the United States. To explain which transit agencies are likely to have been early adopters of GTFS, we estimate a logistic regression model predicting GTFS adoption based on service area and agency characteristics. We find that agencies with higher ridership and those providing lower shares of a region’s total vehicle revenue kilometers have tended to adopt GTFS earlier.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47488550","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Joshua H. Davidson, Ilil Feiglin, Megan S. Ryerson
Much remains to be known about the geographic areas in regions that are well (or poorly) served by one-seat transit – direct service where users do not need to conduct transfers. We describe geographic access to one-seat service, by advancing the framework of a spatial coverage ratio for transit when accounting for commuter flows as reflected in administrative origin-destination data. Our methodology integrates relatively simple spatial approaches with open data, allowing transit providers to modulate thresholds for one-seat service. In doing so, operators can develop new priority areas for intervention and line adaptation.
{"title":"A One-seat Ride Coverage Ratio using Administrative Origin Destination Data","authors":"Joshua H. Davidson, Ilil Feiglin, Megan S. Ryerson","doi":"10.32866/001c.57771","DOIUrl":"https://doi.org/10.32866/001c.57771","url":null,"abstract":"Much remains to be known about the geographic areas in regions that are well (or poorly) served by one-seat transit – direct service where users do not need to conduct transfers. We describe geographic access to one-seat service, by advancing the framework of a spatial coverage ratio for transit when accounting for commuter flows as reflected in administrative origin-destination data. Our methodology integrates relatively simple spatial approaches with open data, allowing transit providers to modulate thresholds for one-seat service. In doing so, operators can develop new priority areas for intervention and line adaptation.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2023-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47796336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hossain Mohiuddin, Md. Hamidur Rahman, Fajle Rabbi Ashik, M. Bhuiya
This study explores the modality and trip chaining patterns of individuals in Dhaka, Bangladesh. We use household-level trip data for a day collected from randomly selected respondents of the Dhaka Metropolitan Development Plan area. We found that walking and rickshaw are the dominant modes of travel. The majority of individuals are unimodal and mostly depend on either walking or on rickshaws. Individuals generally walk for the first and last-mile connections to public transit. Cars are used more for non-home-based business trips. Personalized modes such as cars, cycles, and motorcycles are present at a higher proportion in the super complex trip chain than other types of chains.
{"title":"Analysis of Modality and Trip Chaining Patterns in Dhaka","authors":"Hossain Mohiuddin, Md. Hamidur Rahman, Fajle Rabbi Ashik, M. Bhuiya","doi":"10.32866/001c.56911","DOIUrl":"https://doi.org/10.32866/001c.56911","url":null,"abstract":"This study explores the modality and trip chaining patterns of individuals in Dhaka, Bangladesh. We use household-level trip data for a day collected from randomly selected respondents of the Dhaka Metropolitan Development Plan area. We found that walking and rickshaw are the dominant modes of travel. The majority of individuals are unimodal and mostly depend on either walking or on rickshaws. Individuals generally walk for the first and last-mile connections to public transit. Cars are used more for non-home-based business trips. Personalized modes such as cars, cycles, and motorcycles are present at a higher proportion in the super complex trip chain than other types of chains.","PeriodicalId":73025,"journal":{"name":"Findings (Sydney (N.S.W.)","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2022-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42351520","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：