![2
Argument Mining
[1] Chen et al., “Analyzing Culture-Specific Argument Structures in Learner Essays,” ACL ArgMining 2022
[2] Nguyen et al., “Extracting Argument and Domain Words for Identifying Argument Components in Texts,” ACL ArgMining 2015
• What is Argument Mining?](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-3-320.jpg)
![3
Argument Mining
[1] Stab et al., “Parsing Argumentation Structures in Persuasive Essays” ACL’17
• Argumentation Structures](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-4-320.jpg)
![4
Argument Mining
[1] Sofi et al., “A Robustness Evaluation Framework for Argument Mining” ACL ArgMining 2022
• Argument Mining](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-5-320.jpg)
![5
Argument Mining Workshop
[1] Nguyen et al., “Extracting Argument and Domain Words for Identifying Argument Components in Texts” ACL ArgMining 2015
• Nguyen et al., “Extracting Argument and Domain Words for Identifying
Argument Components in Texts,” ACL ArgMining 2015
LDA : To learn argument and domain words
• Lexical aspect : topic model
development data (described below) to separate argument words (e.g. ‘view’, ‘conclude’, ‘think’)
from domain words (e.g. ‘art’, ‘life’) <-> n-gram
• Structural aspect : subject and main verb
use dependency parses to extract pairs of subject and main verb of sentences, e.g. “I.think”,
“view.be”](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-6-320.jpg)
![6
Argument Mining Workshop
[1] Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022
• Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative
Text” ACL ArgMining 2022](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-7-320.jpg)
![7
Argument Mining Workshop
[1] Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022
• Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative
Text” ACL ArgMining 2022
Reasoning Marker Prediction
Argument Component
Input AC types: “[premise] premise text [RM]
[claim] claim text“
RMs such as “in conclusion" are very
common in the dataset before major claims.](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-8-320.jpg)
![8
Argument Mining 2021 Task
[1] Alshomary et al., “Key Point Analysis via Contrastive Learning and Extractive Argument Summarization” ACL ArgMining 2021
• KPA(Key Point Analysis)
: Key Point Analysis (KPA) is a new NLP task, with strong relations to Computational Argumentation, Opinion
Analysis, and Summarization](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-9-320.jpg)
![9
Workshop on Argument Mining 2021
[1] Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021
• Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned
Parenthood Debate” ACL ArgMining 2021
contribution : It is possible to distinguish between arguments and reasons for short articles
such as Twitter. Public opinion on the SNS becomes clear.
• Support with reason
: support the claim
• Support without reason
: support the claim. But not provide a reason
• No explicit support
:the user has a neutral or unclear stance](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-10-320.jpg)
![10
Workshop on Argument Mining 2021
[1] Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021
• Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned
Parenthood Debate” ACL ArgMining 2021](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-11-320.jpg)
![11
Argument Mining 2022 Task
[1] Heinisch et al., “Data Augmentation for Improving the Prediction of Validity and Novelty of Argumentative Conclusions” ACL ArgMining 2022
• PREDICTING THE VALIDITY AND NOVELTY OF ARGUMENTS](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-12-320.jpg)
![12
Workshop on Argument Mining 2022
[1] Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022
• Ruckdeschel et al. “Boundary Detection and Categorization of Argument
Aspects via Supervised Learning” ArgMining 2022
Contribution : Clearly Argumentation which context unit aspects should be considered for
categorization of arguments (token, chunked, and sentence-level)
• Sequence Tagging
Named Entity Recognition(NER)
• Chunk Normalization
Syntactic Chunker
• Multi-Class Chunk Classification
Entire Chunk, separated with [SEP] token
• Sentence Classification](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-13-320.jpg)
![13
Workshop on Argument Mining 2022
[1] Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022
• Ruckdeschel et al. “Boundary Detection and Categorization of Argument
Aspects via Supervised Learning” ArgMining 2022](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-14-320.jpg)
![14
Argument Mining Workshop
[1] Poudyal et al., “ECHR: Legal Corpus for Argument Mining” ACL ArgMining 2020
• Poudyal et al., “ECHR: Legal Corpus for Argument Mining” ACL ArgMining
2020
Proceed with Argument Mining using BERT
The learning process is the same as before
Provide visualization tools in user text
-> Efficient understanding of Argument](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-15-320.jpg)
![15
Development of Argument Mining
[1] https://argmining-org.github.io/2023/
• Argument Mining 2023 Task](https://image.slidesharecdn.com/argumentminingseminarfinal-230814063245-bcc455f1/85/Argument-Mining-16-320.jpg)
Argument Mining
- 2. Contents • Argument Mining • Workshop on Argument Mining Argument Mining Procedure 2021 Task : Key Point Analysis 2022 Task : Validity-Novelty Prediction Other papars • Development of Argument Mining
- 3. 2 Argument Mining [1] Chen et al., “Analyzing Culture-Specific Argument Structures in Learner Essays,” ACL ArgMining 2022 [2] Nguyen et al., “Extracting Argument and Domain Words for Identifying Argument Components in Texts,” ACL ArgMining 2015 • What is Argument Mining?
- 4. 3 Argument Mining [1] Stab et al., “Parsing Argumentation Structures in Persuasive Essays” ACL’17 • Argumentation Structures
- 5. 4 Argument Mining [1] Sofi et al., “A Robustness Evaluation Framework for Argument Mining” ACL ArgMining 2022 • Argument Mining
- 6. 5 Argument Mining Workshop [1] Nguyen et al., “Extracting Argument and Domain Words for Identifying Argument Components in Texts” ACL ArgMining 2015 • Nguyen et al., “Extracting Argument and Domain Words for Identifying Argument Components in Texts,” ACL ArgMining 2015 LDA : To learn argument and domain words • Lexical aspect : topic model development data (described below) to separate argument words (e.g. ‘view’, ‘conclude’, ‘think’) from domain words (e.g. ‘art’, ‘life’) <-> n-gram • Structural aspect : subject and main verb use dependency parses to extract pairs of subject and main verb of sentences, e.g. “I.think”, “view.be”
- 7. 6 Argument Mining Workshop [1] Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022 • Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022
- 8. 7 Argument Mining Workshop [1] Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022 • Clayton et al., “Predicting the Presence of Reasoning Markers in Argumentative Text” ACL ArgMining 2022 Reasoning Marker Prediction Argument Component Input AC types: “[premise] premise text [RM] [claim] claim text“ RMs such as “in conclusion" are very common in the dataset before major claims.
- 9. 8 Argument Mining 2021 Task [1] Alshomary et al., “Key Point Analysis via Contrastive Learning and Extractive Argument Summarization” ACL ArgMining 2021 • KPA(Key Point Analysis) : Key Point Analysis (KPA) is a new NLP task, with strong relations to Computational Argumentation, Opinion Analysis, and Summarization
- 10. 9 Workshop on Argument Mining 2021 [1] Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021 • Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021 contribution : It is possible to distinguish between arguments and reasons for short articles such as Twitter. Public opinion on the SNS becomes clear. • Support with reason : support the claim • Support without reason : support the claim. But not provide a reason • No explicit support :the user has a neutral or unclear stance
- 11. 10 Workshop on Argument Mining 2021 [1] Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021 • Bhatti et al., “Argument Mining on Twitter: A Case Study on the Planned Parenthood Debate” ACL ArgMining 2021
- 12. 11 Argument Mining 2022 Task [1] Heinisch et al., “Data Augmentation for Improving the Prediction of Validity and Novelty of Argumentative Conclusions” ACL ArgMining 2022 • PREDICTING THE VALIDITY AND NOVELTY OF ARGUMENTS
- 13. 12 Workshop on Argument Mining 2022 [1] Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022 • Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022 Contribution : Clearly Argumentation which context unit aspects should be considered for categorization of arguments (token, chunked, and sentence-level) • Sequence Tagging Named Entity Recognition(NER) • Chunk Normalization Syntactic Chunker • Multi-Class Chunk Classification Entire Chunk, separated with [SEP] token • Sentence Classification
- 14. 13 Workshop on Argument Mining 2022 [1] Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022 • Ruckdeschel et al. “Boundary Detection and Categorization of Argument Aspects via Supervised Learning” ArgMining 2022
- 15. 14 Argument Mining Workshop [1] Poudyal et al., “ECHR: Legal Corpus for Argument Mining” ACL ArgMining 2020 • Poudyal et al., “ECHR: Legal Corpus for Argument Mining” ACL ArgMining 2020 Proceed with Argument Mining using BERT The learning process is the same as before Provide visualization tools in user text -> Efficient understanding of Argument
- 16. 15 Development of Argument Mining [1] https://argmining-org.github.io/2023/ • Argument Mining 2023 Task