​
Introduction

Ever feel like you’re seeing the same news story over and over? You’re not alone. In today’s fast-paced digital world, major events often get covered by multiple outlets, resulting in a flood of similar articles. This can make it challenging to find unique, valuable content. That’s where article deduplication comes in. It’s a feature we’ve built into News API v3 to help you cut through the noise and get to the heart of the news.

​
What is article deduplication?

Article deduplication is a process that identifies and filters out nearly identical news articles from a large collection of content. It’s smarter than simple text matching - it uses advanced language processing to recognize articles that say the same thing, even if they use different words or structures. Here’s what article deduplication does:
  • Provides you with a diverse set of unique articles.
  • Reduces information overload by eliminating redundant content.
  • Improves the efficiency of content analysis and research.
  • Enhances your experience by presenting a more manageable amount of relevant information.

​
How does it work?

Our deduplication system uses a multi-step process to accurately identify duplicate articles while maintaining high precision. Let’s break it down:

​
Semantic similarity comparison

First, we compare the meaning of articles:
  1. We convert article texts into vector representations (embeddings) using our Natural Language Processing (NLP) pipeline.
  2. These embeddings capture the meaning and relationships of the content.
  3. We use cosine similarity to compare these embeddings, with a threshold of 0.95 to identify potential duplicates.
This method catches similar articles even if they use different wording or sentence structures. It’s great at identifying rewrites or articles from different sources covering the same event.

​
Levenshtein distance analysis

After the initial screening, we refine our process using the Levenshtein distance, which measures the minimum number of single-character edits needed to change one text into another. We use specific thresholds for accuracy:
  • 0.97 for article titles
  • 0.92 for article content
This step helps us distinguish between articles that discuss similar topics differently and those that are true duplicates. It reduces the chance of false positives in our deduplication process.

​
Identifying the original article

Our system doesn’t just spot duplicates - it also identifies which article is likely the original or most authoritative version. We use a scoring algorithm that considers factors like:
  • Domain credibility
  • Author’s reputation
  • Publication timestamp
The article with the highest score becomes the “parent” article. This status can change if we find a new duplicate with a higher score, reflecting the dynamic nature of news content. By identifying the original article, you get access to the most credible and comprehensive version of a story, while still filtering out redundant content.

​
Continuous updates and historical lookup

To keep our deduplication system effective, we continuously update our database and compare each new article against articles from the past seven days. This approach catches duplicates that may appear days after the original publication, accounting for delayed reporting or republishing content. By implementing this comprehensive process, News API v3 provides you with a rich, diverse, and non-redundant set of news articles, enhancing the overall quality and usability of the data.

​
How to use deduplication

Now that you understand how article deduplication works, let’s look at how to use this feature in your News API v3 queries.

​
Enable deduplication

To exclude duplicate articles from your search results, set the exclude_duplicates parameter to true in your API request. Here’s what you need to know:

​
Understand the API response

When you enable deduplication, each article object in the response includes additional fields related to duplication:
  • duplicate_count: The number of duplicates associated with the article.
  • duplicate_articles_group_id: A unique identifier for the group of duplicates associated with the article.
These fields give you valuable information about the extent of duplication for each article. You can use this data for further analysis or filtering in your application.

​
Code example

Here’s an example of how to make a request using Python that includes the deduplication feature:
request.py
import requests
import json

url = "https://v3-api.newscatcherapi.com/api/search"
payload = json.dumps({
  "q": "market value",
  "lang": "en",
  "theme": "Tech",
  "exclude_duplicates": True
})
headers = {
  'x-api-token': 'YOUR-API-KEY',
  'Content-Type': 'application/json',
  'Accept': 'application/json'
}
response = requests.request("POST", url, headers=headers, data=payload)
print(response.text)
In the response, you’ll find deduplication information for each article. Here’s an example of what you might see:
response.json
{
  "title": "Global Financial Services Industry - Insights Around Market Size, Key Trends And Forecast, 2024: Grand View Research, Inc.",
  ...
  "duplicate_count": 5,
  "duplicate_articles_group_id": "542def7ce3844c269d5f1a929309e6da"
}
This shows that the article has five duplicates, which have been excluded from the results due to the exclude_duplicates parameter.

​
Use cases

Let’s explore some key scenarios where deduplication can make a significant impact:
  • Content curation and aggregation: Ensure users of your news aggregator see a diverse range of articles without redundancy, improving user experience by reducing information overload and increasing the variety of perspectives presented.
  • Media monitoring and analysis: Focus on unique brand mentions or industry trends without getting overwhelmed by repetitive content, leading to more accurate sentiment analysis and trend identification.
  • Research and trend analysis: Streamline data collection for researchers and analysts by eliminating duplicate articles, making it easier to identify unique data points and trends for more accurate and efficient analysis.
  • Personalized news feeds: Enhance user engagement and satisfaction by ensuring readers see a wider variety of content rather than multiple versions of the same story in their personalized news feeds.

​
Deduplication vs clustering

While both deduplication and clustering organize and manage large sets of articles, they serve different purposes and are used in different contexts.

​
Key differences

  • Deduplication identifies and removes nearly identical articles, preserving only the most relevant or original version.
  • Clustering groups similar articles together without removing any content, allowing you to see multiple perspectives on a topic.

​
When to use deduplication vs clustering

  • Use deduplication to eliminate redundancy and present only unique content to your users.
  • Use clustering to group related articles together to show different angles or developments of a story over time.
For more information on our clustering feature, check out Clustering news articles.