Skip to content
beginnersurvey

Multilingual News Article Similarity

Rating the similarity between pairs of news articles across seven dimensions (geography, entities, time, narrative, overall, style, tone), each on a 4-point scale, to assess whether they cover the same story across multiple languages. Based on SemEval-2022 Task 8 (Chen et al.).

Q1: Rate your experience12345Q2: Primary use case?ResearchIndustryEducationQ3: Additional feedback

Configuration Fileconfig.yaml

This Potato config reproduces the annotation task. Save it as config.yaml and run potato start config.yaml to try it.

yaml
# Multilingual News Article Similarity
# Based on Chen et al., SemEval 2022
# Paper: https://aclanthology.org/2022.semeval-1.155/
# Dataset: https://github.com/euagendas/semeval_8_2022_ia_downloader
#
# This task asks annotators to rate the similarity between two news
# articles across seven dimensions of similarity, each on a 4-point
# Likert scale. Articles may be in different languages.
#
# Following the paper's codebook, each dimension is answered on a
# four-point scale where 1 = Very Similar, 2 = Somewhat Similar,
# 3 = Somewhat Dissimilar, and 4 = Very Dissimilar.
#
# Dimensions (codebook questions):
#   GEO      - Geographic focus (places, cities, countries)
#   ENT      - Named entities (people, companies, organizations, products)
#   TIME     - Time periods described / relevance
#   NARRATIVE- Narrative schemas presented
#   OVERALL  - Whether the articles cover the same substantive news story
#   STYLE    - Writing style
#   TONE     - Tone

annotation_task_name: "Multilingual News Article Similarity"
task_dir: "."

data_files:
  - sample-data.json

item_properties:
  id_key: "id"
  text_key: "text"

output_annotation_dir: "annotation_output/"
output_annotation_format: "json"

port: 8000
server_name: localhost

annotation_schemes:
  - annotation_type: likert
    name: geography
    description: "GEO: How similar is the geographic focus (places, cities, countries, etc.) of the two articles?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: entities
    description: "ENT: How similar are the named entities (people, companies, organizations, products, named living beings), excluding locations already considered above?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: time
    description: "TIME: Are the two articles relevant to similar time periods or describing similar time periods?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: narrative
    description: "NAR: How similar are the narrative schemas presented in the two articles?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: overall
    description: "OVERALL: Overall, are the two articles covering the same substantive news story? (excluding style, framing, and tone)"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: style
    description: "STYLE: Do the articles have similar writing styles?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4
  - annotation_type: likert
    name: tone
    description: "TONE: Do the articles have similar tones?"
    min_label: "Very Similar"
    max_label: "Very Dissimilar"
    size: 4

annotation_instructions: |
  You will see two news articles (possibly in different languages) and the language pair.
  Rate the similarity of the two articles on seven dimensions. Each dimension uses a
  four-point scale where 1 = Very Similar, 2 = Somewhat Similar,
  3 = Somewhat Dissimilar, and 4 = Very Dissimilar.
  - Geography: similarity of the geographic focus (places, cities, countries).
  - Entities: similarity of the named entities (people, companies, organizations, products).
  - Time: whether the articles describe or are relevant to similar time periods.
  - Narrative: similarity of the narrative schemas presented.
  - Overall: whether the two articles cover the same substantive news story (excluding style, framing, and tone).
  - Style: whether the articles have similar writing styles.
  - Tone: whether the articles have similar tones.

html_layout: |
  <div style="padding: 15px; max-width: 800px; margin: auto;">
    <div style="background: #ecfdf5; border: 1px solid #a7f3d0; border-radius: 8px; padding: 12px; margin-bottom: 12px;">
      <strong style="color: #065f46;">Language Pair:</strong>
      <span style="font-size: 15px; margin-left: 8px;">{{language}}</span>
    </div>
    <div style="background: #f0f9ff; border: 1px solid #bae6fd; border-radius: 8px; padding: 16px; margin-bottom: 16px;">
      <strong style="color: #0369a1;">Article 1:</strong>
      <p style="font-size: 16px; line-height: 1.7; margin: 8px 0 0 0;">{{text}}</p>
    </div>
    <div style="background: #fefce8; border: 1px solid #fde68a; border-radius: 8px; padding: 16px; margin-bottom: 16px;">
      <strong style="color: #a16207;">Article 2:</strong>
      <p style="font-size: 16px; line-height: 1.7; margin: 8px 0 0 0;">{{article_2}}</p>
    </div>
  </div>

allow_all_users: true
instances_per_annotator: 50
annotation_per_instance: 3
allow_skip: true
skip_reason_required: false

Sample Datasample-data.json

json
[
  {
    "id": "news_sim_001",
    "text": "The European Central Bank raised interest rates by 0.5 percentage points on Thursday, marking the largest increase in over a decade as the institution battles soaring inflation across the eurozone.",
    "article_2": "La Banque centrale europeenne a releve ses taux d'interet de 0,5 point de pourcentage jeudi, soit la plus forte hausse depuis plus de dix ans, dans un effort pour lutter contre l'inflation galopante dans la zone euro.",
    "language": "English-French"
  },
  {
    "id": "news_sim_002",
    "text": "A powerful 7.2 magnitude earthquake struck southern Haiti early Saturday morning, causing widespread destruction and leaving thousands without shelter in the impoverished Caribbean nation.",
    "article_2": "Rescuers pulled survivors from the rubble of a collapsed factory in Bangladesh after a structural failure caused the six-story building to pancake during morning work hours.",
    "language": "English-English"
  }
]

// ... and 8 more items

Get This Design

View on GitHub

Clone or download from the repository

Quick start:

git clone https://github.com/davidjurgens/potato-showcase.git
cd potato-showcase/semeval/2022/task08-news-similarity
potato start config.yaml

Dataset & paper

Chen et al., SemEval 2022, pages 1094-1106

Citation (BibTeX)

bibtex
@inproceedings{chen-etal-2022-semeval,
    title = "{S}em{E}val-2022 Task 8: Multilingual news article similarity",
    author = {Chen, Xi  and Zeynali, Ali  and Camargo, Chico  and Fl{\"o}ck, Fabian  and Gaffney, Devin  and Grabowicz, Przemyslaw  and Hale, Scott A.  and Jurgens, David  and Samory, Mattia},
    booktitle = "Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022)",
    month = jul,
    year = "2022",
    address = "Seattle, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2022.semeval-1.155/",
    doi = "10.18653/v1/2022.semeval-1.155",
    pages = "1094--1106"
}

Details

Annotation Types

likert

Domain

NLPMedia AnalysisSemEval

Use Cases

Document SimilarityNews AnalysisMultilingual NLP

Tags

semevalsemeval-2022shared-tasknews-similaritymultilingualdocument-similarity

Found an issue or want to improve this design?

Open an Issue