DeskDrop Articles & Interactions
Data from the DeskDrop platform: more than 3,000 articles (EN/PT) accompanied by user interactions (views, likes, comments) over 12 months.
Approximately 73,000 logs and 3,000 articles, CSV
Open Database/Content (free reuse with attribution)
Description
The dataset DeskDrop Articles & Interactions contains the logs of 12 months of activity of the internal DeskDrop article sharing platform. It includes more than 3,000 articles in English and Portuguese, and nearly 73,000 interactions (consultations, likes, comments). Each interaction is contextualized (timestamp, customer used, geolocation), and each article is accompanied by its raw content, title, URL and language.
What is this dataset for?
- Train and test recommendation engines (collaborative, content-based, or hybrid filtering)
- Analyzing reading habits in a professional context
- Experiment with multilingual NLP algorithms on short and dynamic content
Can it be enriched or improved?
Yes. It is possible to aggregate interactions by user or item, create weighted interest scores, or add sentiment analyses. NLP vectorization or semantic embeddings can also transform textual content into usable vectors for machine learning.
🔎 In summary
🧠 Recommended for
- Projects in recommendation systems
- Multilingual NLP
- User behavior analysis
🔧 Compatible tools
- Python (Pandas, Scikit-Learn, Surprise)
- TensorFlow Recommenders
- LDA
- LightFM
💡 Tip
Combine the implicit interest score (views < likes < comments) to create a more nuanced engagement signal in your recommendation engine
Frequently Asked Questions
Are users identifiable in the data?
No, identifiers are anonymized and do not allow users to be directly identified.
Is the dataset multilingual?
Yes, the articles are provided in English and Portuguese, which allows for bilingual NLP experiments.
Can this data be used for commercial purposes?
Yes, the license allows commercial reuse provided that the specified attribution rights are respected.




