Tracking

With the tracking service, you can upload articles to our system and get a feed containing matching articles. This can be used to track the spread of press releases or the like.

Upload articles using either FTP or the API. Articles picked up by Opoint will be compared to the uploaded content and given a score. If the score is sufficiently high, it is considered a match and will be delivered.

Score

The score is reported in the tracking object for each article in the delivery feed. The threshold for delivery can be tuned, by supplying a parameter when uploading. Otherwise, a default value of 20 is used.

How the score is computed

Identical content gives the maximum score of a 100. This would mean the upload was republished in its entirety somewhere else.

chevron_right Show example 
Volvo Cars reported its best-ever second half profit and sales volumes in 2020, illustrating the company’s success in mitigating and recovering from the impact of the coronavirus pandemic earlier in the year.

The company reported revenue of 151 billion SEK and an operating profit of 9.5 billion SEK for the second six months of 2020. Profits increased by 8.2 per cent during the period, while revenues were up by 4.9 per cent. The profit margin came in at a strong 6.3 per cent.

Score: 100

If the original text is included in another article, it suggests that the article has quoted the upload, and would thus receive a high score.

chevron_right Show example 
Volvo Cars reported its best-ever second half profit and sales volumes in 2020, illustrating the company’s success in mitigating and recovering from the impact of the coronavirus pandemic earlier in the year.

The company reported revenue of 151 billion SEK and an operating profit of 9.5 billion SEK for the second six months of 2020. Profits increased by 8.2 per cent during the period, while revenues were up by 4.9 per cent. The profit margin came in at a strong 6.3 per cent. The result reflects the company’s best ever second half in terms of sales, driven by strong demand for Volvo Cars’ Recharge line-up of chargeable cars.

Score: 87

It is not necessary that all the text from the upload is found in the other article, and a high score can be given if a large part of the original text is found.

chevron_right Show example 
Volvo Cars reported its best-ever second half profit and sales volumes in 2020, illustrating the company’s success in mitigating and recovering from the impact of the coronavirus pandemic earlier in the year.

The company reported revenue of 151 billion SEK and an operating profit of 9.5 billion SEK for the second six months of 2020. Profits increased by 6.2 per cent during the period, while revenues were up by 4.9 per cent. The profit margin came in at a strong 6.3 per cent.

Score: 79

The score does not depend only on how many common words or sentences are found. If all the sentences or paragraphs of the upload are used in another article, but some other text is inserted between them, the score will be lowered.

chevron_right Show example 
Volvo Cars reported its best-ever second half profit and sales volumes in 2020, illustrating the company’s success in mitigating and recovering from the impact of the coronavirus pandemic earlier in the year.

The rest of the company’s press release stated that: The company reported revenue of 151 billion SEK and an operating profit of 9.5 billion SEK for the second six months of 2020. Profits increased by 8.2 per cent during the period, while revenues were up by 4.9 per cent. The profit margin came in at a strong 6.3 per cent.

Score: 69

The use of <blockquote> tags to mark one or more parts of the upload affects the score computation. Each block gets its own score, which is 100 if the text is found exactly in the article, or lower if some words have been changed.

The final score is then taken as the max between the score of the article itself and the scores of all blockquotes.

chevron_right Show example

Upload

For the full year 2020, Volvo Cars reported revenue of 263 billion SEK and an operating profit of 8.5 billion SEK. It managed to reduce fixed costs in combination with growth, which had a positive influence on cash flow and liquidity.

<blockquote>“We acted decisively to limit the impact of the pandemic,”</blockquote> said Håkan Samuelsson, chief executive. “After a safe restart of our operations, we recovered strongly and reported the best second half in the company’s history. It is also promising to see the fast-growing demand for our Recharge line-up of chargeable cars, which we expect to continue in 2021.”

Article

Volvo Cars’ chief executive stated that they “acted decisively to limit the impact of the pandemic.”

Score for article: 15 Score for quotes: 80 Score: 80

Upload

You can upload articles for tracking either using FTP, or via the API.

  • POST /tracking/

    Upload an article to the database for tracking.

    Request body

    • link partner_article_id required, string

      A unique id identifying the article.

    • link title required, string

      Title of the article.

    • link summary string

      Lede of the article.

    • link text string

      Body text of the article.

    • link threshold int, between 15 and 100, default 20

      Only articles with a tracking score higher than the threshold will count as matches.

    • link duration int, between 1 and 30, default 30

      Number of days to look for matches.

    • link language string

      Language code of the language the article is written in. If not supplied, the language will be guessed.

    • link url string

      URL the article was published on.

    • link time required, string

      Time the article was published on the format YYYY-MM-DDThh:mm:ss.uuuuuuZ. If the time is in the future, it is truncated to the current time.

    Response

    • link id_site int

      Opoint site id of the created article.

    • link id_article int

      Opoint article id of the created article.

    Request body example

    {
  "title": "My news story",
  "summary": "This is the articles lede",
  "text": "Example text",
  "language": "en",
  "url": "https://mynewssite.com/articles/1/example.html",
  "time": "2020-06-14T12:00:04.00Z",
  "partner_article_id": "1"
}

  • POST /tracking/search/

    Retrieve information about uploaded articles using organization specific ids.

    You generally do not need to use this endpoint. The matches are fetched from the StoredSearch feed endpoint.

    Request body

    • link partner_article_ids required, list of strings

      List of organization ids. Duplicate ids will be removed. Ids are expected to match articles already uploaded.

    Request body example

    {
  "partner_article_ids": [
    "1",
    "2",
    "6"
  ]
}

Delivery of matches

Matched articles can be pushed to FTP or pulled in a feed using the Storedsearch endpoint:

  • GET /storedsearch/feed/

    Retrieve articles matching any of the stored searches, starting at the from parameter, searching forward in time.

    If you are continuously pulling data from the feed, a good strategy is to first pick an initial timestamp from where to get data.. Also decide on an interval.

    Then, make a request, passing the from and interval parameters. From the response, get the next_from field. This can be used as the from parameter in the next request to get the next batch of articles. Make sure to add some appropriate sleeping logic in between calls.

    Query parameters

    • link from required, int

      Unix timestamp to search from.

    • link to int

      Unix timestamp to search until.

    • link interval int, between 120 and 1800

      The number of seconds to search for, forward in time. If the ending time is in the future, the search will return content until the current time. Does not apply if to is set. If not provided, it will be set to 900 seconds.

    • link format json or xml

      Return format.

    • link num_art int, default 25000

      Maximum number of articles to return for each of the stored searches.

    • link topics int, 0 or 1

      Add a topics fields to the reponse.

    • link matches int, 0 or 1

      Mark search matches with <match> tags.

    • link equalgroup int, 0 or 1

      Add a equalgroup field to the response, with each articles group id.

    • link readership int
      error_outline Requires license

      Add readership data to the response. Please note that article level readership estimates can be uncertain when the article has just been published, as it is unclear for how long the article will stay visible. Updated readership estimates can be fetched from the search API.

    • link textrazor int
      error_outline Requires license

      Add textrazor data to the response. The parameter values are interpreted the same way as the search @param/textrazor parameter.

    Response

    • link next_from int or missing

      Unix timestamp indicating where the current search stopped, and where the next one can start from. Only applies if interval is used.

    • link covered int or missing

      Difference between the from parameter and next_from. Can be negative if the from parameter was set to the future. Only applies if interval is used.