The Customer Support on Twitter dataset is a large, modern corpus of tweets and replies to aid innovation in natural language understanding and conversational models, and for study of modern customer support practices and impact.

https://i.imgur.com/nTv3Iuu.png" alt="Example Analysis - Inbound Volume for the Top 20 Brands">

Context

Natural language remains the densest encoding of human experience we have, and innovation in NLP has accelerated to power understanding of that data, but the datasets driving this innovation don't match the real language in use today. The Customer Support on Twitter dataset offers a large corpus of modern English (mostly) conversations between consumers and customer support agents on Twitter, and has three important advantages over other conversational text datasets:

• Focused - Consumers contact customer support to have a specific problem solved, and the manifold of problems to be discussed is relatively small, especially compared to unconstrained conversational datasets like the reddit Corpus.
• Natural - Consumers in this dataset come from a much broader segment than those in the Ubuntu Dialogue Corpus and have much more natural and recent use of typed text than the Cornell Movie Dialogs Corpus.
• Succinct - Twitter's brevity causes more natural responses from support agents (rather than scripted), and to-the-point descriptions of problems and solutions. Also, its convenient in allowing for a relatively low message limit size for recurrent nets.

Inspiration

The size and breadth of this dataset inspires many interesting questions:

• Can we predict company responses? Given the bounded set of subjects handled by each company, the answer seems like yes!
• Do requests get stale? How quickly do the best companies respond, compared to the worst?
• Can we learn high quality dense embeddings or representations of similarity for topical clustering?
• How does tone affect the customer support conversation? Does saying sorry help?
• Can we help companies identify new problems, or ones most affecting their customers?

Content

The dataset is a CSV, where each row is a tweet. The different columns are described below. Every conversation included has at least one request from a consumer and at least one response from a company. Which user IDs are company user IDs can be calculated using the inbound field.

tweet_id

A unique, anonymized ID for the Tweet. Referenced by response_tweet_id and in_response_to_tweet_id.

author_id

A unique, anonymized user ID. @s in the dataset have been replaced with their associated anonymized user ID.

inbound

Whether the tweet is "inbound" to a company doing customer support on Twitter. This feature is useful when re-organizing data for training conversational models.

created_at

Date and time when the tweet was sent.

text

Tweet content. Sensitive information like phone numbers and email addresses are replaced with mask values like _email_.

response_tweet_id

IDs of tweets that are responses to this tweet, comma-separated.

in_response_to_tweet_id

ID of the tweet this tweet is in response to, if any.

Contributing

Know of other brands the dataset should include? Found something that needs to be fixed? Start a discussion, or email me directly at $FIRSTNAME@$LASTNAME.com!

Acknowledgements

A huge thank you to my friends who helped bootstrap the list of companies that do customer support on Twitter! There are many rocks that would have been left un-turned were it not for your suggestions!

Relevant Resources

• NLTK - casual_tokenize for social media text tokenizing, vader sentiment analysis for social media text
• SciKit Learn - BoW Count Vectorizer, Multinomial Naive Bayes Classifier
• Topic Modeling via Phrase detection with gensim
• facebook research - fastText text classifier

Dataset

This dataset was created by Galal Qassas

Released under MIT

Contents

Dataset

This dataset was created by Muhammad Asif

Released under Apache 2.0

Contents

Dataset

This dataset was created by Rock-Lagoon

Contents

- this data contains 10000 tweets for a telecom company's customer care account on Twitter.

- this data need to use in Sentiment Analysis in Arabic.

Overview

This dataset contains tweets related to major US airlines and is widely used for NLP and sentiment analysis tasks. Each record includes the tweet text, timestamp, airline name, and sentiment label (positive, negative, neutral). This uploaded version is prepared to support advanced text processing, machine learning, and anomaly detection experiments.

What's Included

• Tweets.csv – Full collection of airline-related tweets
• Text content suitable for NLP tasks
• Timestamp information (useful for time-based analysis)
• Sentiment labels for classification and evaluation
• Cleaned text field for direct use in ML pipelines

Purpose of This Dataset

This dataset is used in a machine learning workflow focused on: - sentiment analysis
- embedding generation (transformers)
- dimensionality reduction (PCA, UMAP)
- clustering and visualization
- unsupervised anomaly detection using Isolation Forest

It is especially suited for exploring changes in public sentiment, event detection, and contextual analysis in social media data.

Key Use Cases

• Building and testing NLP models
• Semantic similarity and embedding-based analysis
• Sentiment classification
• Detecting anomalous posts or time periods
• Visualizing tweet clusters using UMAP
• Studying customer feedback patterns in the airline industry

Source

Originally derived from the Twitter US Airline Sentiment dataset on Kaggle.
This uploaded version is intended for educational, analytical, and research purposes.

Notes

If you're using this dataset in a notebook, ensure you update your file path accordingly: ```python df = pd.read_csv("/kaggle/input/twitter-airline-sentiment-dataset/Tweets.csv")

Overview

This dataset simulates a production-grade customer support system for a multi-product SaaS company. It contains 2,000 support tickets submitted through 7 different channels — email, chat, phone, WhatsApp, Twitter DM, helpdesk portal, and web form — covering a broad range of issue types, customer segments, and global regions.

Each ticket captures the full lifecycle of a customer issue: from the moment it was created, through triage and agent assignment, to resolution. The dataset is ideal for building and benchmarking NLP classifiers, customer sentiment models, resolution time predictors, churn-risk scorers, and support analytics dashboards.

What Makes This Dataset Unique

Real customer messages — natural language ticket messages across 6 languages (English, Hindi, Arabic, Spanish, French, Japanese, German) End-to-end ticket lifecycle — creation time, resolution time, first response time, reopened/escalated flags Multi-dimensional — product, plan tier, region, channel, priority, and agent all captured Rich target variables — CSAT score (1–5), sentiment (5 levels), resolution status, escalation flag, and resolution time in hours Balanced classes — thoughtfully distributed across categories, regions, and channels

Potential Use Cases

🤖 NLP / Text Classification — Classify support tickets by category or priority from message text 😊 Sentiment Analysis — Predict or validate customer sentiment from message tone ⏱️ Regression — Predict resolution time or first response time 📊 Support Analytics — Visualize agent performance, channel effectiveness, or regional patterns 🔁 Escalation Prediction — Build models to flag tickets likely to be escalated ⭐ CSAT Prediction — Predict customer satisfaction scores from ticket features 🌍 Multilingual NLP — Work with support messages in 7 languages

Dataset at a Glance

PropertyValueRows2,000Columns23Time PeriodJan 2024 – Jun 2025Channels7 (Email, Chat, Phone, WhatsApp, Twitter DM, Helpdesk Portal, Web Form)Products5 (TaskBoard, DataDash, CloudSync Pro, InvoiceFlow, SecureVault)Regions7 (North America, Europe, South Asia, Southeast Asia, Middle East, Africa, Latin America)Languages7 (English, Hindi, Arabic, Spanish, French, Japanese, German)Priority Levels4 (Low, Medium, High, Critical)Ticket Statuses5 (Open, In Progress, Resolved, Closed, Escalated)

File Description

This dataset contains Twitter support conversations collected from various company accounts. It includes customer inquiries and corresponding support responses. The data is useful for training AI chatbots, analyzing customer service trends, and developing sentiment analysis models.

Column Description

This dataset contains customer support interactions on Twitter. It includes the following columns: tweet_id: A unique identifier for each tweet. author_id: The unique ID of the user who posted the tweet. inbound: A boolean value indicating whether the tweet is from a customer (True) or from the support team (False). created_at: The timestamp of when the tweet was posted (in UTC format). text: The content of the tweet. response_tweet_id: The unique ID of the response tweet, if applicable. in_response_to_tweet_id: The ID of the original tweet to which this tweet is responding.

How This Data Can Be Used? Training a chatbot: Helps in generating automated support responses. Sentiment analysis: Can analyze whether tweets are complaints, queries, or feedback. Conversation tracking: By linking response tweets with original messages.

originalAuthor : MANORAMA Source : https://www.kaggle.com/datasets/manovirat/aspect/data

Note: This dataset is shared for educational and research purposes only.

This data was collected from several customer care accounts as inquiries of the customers.

"fullText": This variable contains the full-text content of the tweet. "lang": This variable indicates the language in which the tweet is written. "viewsCount": This variable represents the count of views or impressions the tweet has received. "bookmarkCount": This variable represents the count of times the tweet has been bookmarked by users. "favoriteCount": This variable represents the count of times the tweet has been favorited by users. "replyCount": This variable represents the count of replies the tweet has received. "retweetCount": This variable represents the count of times the tweet has been retweeted by users. "quoteCount": This variable represents the count of times the tweet has been quoted by users.

Twitter data: Approx 525,000 Tweets (0.5m) with keyword 'weather' for 3-21 Dec 2022 including RT retweets. You can download this data and more; visit our site for more data twtdata.com Please contact mark@twtdata.com if you need more data.

Twitter Depression Detection Dataset

Overview

This dataset contains Twitter posts with both text and images for depression detection research.

Dataset Statistics

• Total samples: 32363
• Depression samples: 2335
• Non-depression samples: 30028
• Total images: 32363

File Structure

twitter_depression_dataset/
├── images/
│  ├── depresi/     # Images from depression-related posts
│  └── nondepresi/    # Images from non-depression posts
├── metadata/
│  ├── full_dataset.csv # Complete dataset with all metadata
│  └── image_labels.csv # Simple image-label mapping
└── dataset-metadata.json # Kaggle dataset metadata

Data Fields

• full_text: Original tweet text
• label_text: Text label (0=non-depression, 1=depression)
• label_image: Image label (0=non-depression, 1=depression)
• kaggle_output_path: Path to image file in this dataset
• created_at: Tweet creation timestamp
• favorite_count: Number of likes
• Additional Twitter metadata...

Usage

import pandas as pd

# Load the dataset
df = pd.read_csv('metadata/full_dataset.csv')

# Load image-label mapping
labels = pd.read_csv('metadata/image_labels.csv')

Citation

If you use this dataset, please cite appropriately and ensure ethical use for research purposes.

Ethical Considerations

This dataset contains social media data. Please ensure: - Respectful use for research purposes - Proper anonymization in publications - Compliance with platform terms of service - Consideration of mental health sensitivity

Context

This dataset is a part of our research work titled "Opinion Mining of Customer Reviews Using Supervised Learning Algorithms". If you use this dataset then please cite our work. You can find the article in https://ieeexplore.ieee.org/document/9733435

Content

Nowadays, a lot of people express their opinions on various topics using social networking sites. Twitter has become a famous social networking site where people can express their opinions to the point and so it has become a great source for opinion mining. In this research, the goal was to train and build a model that can automatically and accurately categorize the opinion of customer tweet reviews about popular cell phone brands. We have used python TextBlob library for getting the polarity values of all the tweet reviews of the dataset. We have also used Support Vector Machine (SVM), Naïve Bayes, Logistic Regression, Decision Tree and Random Forest algorithms along with Bag of Words and TF-IDF vectorizers separately to train and build the model. We have investigated the opinions using five classes which are Strongly Positive, Positive, Neutral, Negative and Strongly Negative.

When referencing this dataset please cite the below paper

Bibtex @inproceedings{arif2021opinion, title={Opinion Mining of Customer Reviews Using Supervised Learning Algorithms}, author={Arif, Shibbir Ahmed and Hossain, Taslima Binte}, booktitle={2021 5th International Conference on Electrical Information and Communication Technology (EICT)}, pages={1--6}, year={2021}, organization={IEEE} }

Unleash the Power of Apple Sentiment Analysis: A High-Quality, Labeled Tweet Dataset

Struggling to find the perfect dataset for fine-tuning your Transformer model on Apple sentiment analysis? Look no further! Tired of dealing with scattered, low-quality data on Kaggle? We've been there too. That's why we created this comprehensive dataset, meticulously crafted to meet your needs.

What makes this dataset unique?

• Extensive Collection: Merges various Kaggle datasets to deliver a substantial volume of tweets (5810 tweets), exceeding the limitations of individual sources.
• Rigorous Cleaning: Eliminates spam, irrelevant content, and misleading links, ensuring data integrity and model accuracy.
• Clear Labeling: Every tweet is meticulously labeled with positive, negative, or neutral sentiment, ready for immediate training.

No more piecing together fragmented datasets or battling with unreliable information. This pre-processed and labeled collection empowers you to:

• Develop superior sentiment analysis models for understanding Apple brand perception.
• Gain valuable insights into customer opinions and preferences.
• Accurately track sentiment trends over time.
• Fine-tune your Transformer model with confidence and efficiency.(BERT, RoBERTa, DistilBERT etc.)

This dataset contains drug-related text entries structured to resemble tweets. It was generated from the drugsComTest_raw.csv dataset, which originally included patient reviews of medications. Source: Extracted from patient-submitted reviews on Drugs.com. Format: CSV file with two columns: drugName – the name of the drug mentioned. tweet – the review text reformatted to simulate a tweet-like message.

Purpose: To support Natural Language Processing (NLP) tasks such as sentiment analysis, drug-effect classification, and social media mining. To act as a proxy dataset for training or testing models on drug-related discussions, where actual Twitter data collection is restricted or unavailable.

Limitations: Not real Twitter data, but synthetic tweets generated from formal drug reviews. May differ in tone and structure compared to actual tweets.

IMPORTANT (02-Apr-2024)

Kaggle has fixed the issue with gzip files and Version 510 should now reflect properly working files

IMPORTANT (28-Mar-2024)

Please use the version 508 of the dataset, as 509 is broken. See link below of the dataset that is properly working https://www.kaggle.com/datasets/bwandowando/ukraine-russian-crisis-twitter-dataset-1-2-m-rows/versions/508

Context

The context and history of the current ongoing conflict can be found https://en.wikipedia.org/wiki/2022_Russian_invasion_of_Ukraine.

Announcement

[Jun 16] (🌇Sunset) Twitter has finally pulled the plug on all of my remaining TWITTER API accounts as part of their efforts for developers to migrate to the new API. The last tweets that I pulled was dated last Jun 14, and no more data from Jun 15 onwards. It was fun til it lasted and I hope that this dataset was able and will continue to help a lot. I'll just leave the dataset here for future download and reference. Thank you all!

[Apr 19] Two additional developer accounts have been permanently suspended, expect a lower throughtput in the next few weeks. I will pull data til they ban my last account.

[Apr 08] I woke up this morning and saw that Twitter has banned/ permanently suspended 4 of my developer accounts, I have around a few more but it is just a matter of time till all my accounts will most likely get banned as well. This was a fun project that I maintained for as long as I can. I will pull data til my last account gets banned.

[Feb 26] I've started to pull in RETWEETS again, so I am expecting a significant amount of throughput in tweets again on top of the dedicated processes that I have that gets NONRETWEETS. If you don't want RETWEETS, just filter them out.

[Feb 24] It's been a year since I started getting tweets of this conflict and had no idea that a year later this is still ongoing. Almost everyone assumed that Ukraine will crumble in a matter of days, but it is not the case. To those who have been using my dataset, i hope that I am helping all of you in one way or another. Ill do my best to maintain updating this dataset as long as I can.

[Feb 02] I seem to be getting less tweets as my crawlers are getting throttled, i used to get 2500 tweets per 15 mins but around 2-3 of my crawlers are getting throttling limit errors. There may be some kind of update that Twitter has done about rate limits or something similar. Will try to find ways to increase the throughput again.

[Jan 02] For all new datasets, it will now be prefixed by a year, so for Jan 01, 2023, it will be 20230101_XXXX.

[Dec 28] For those looking for a cleaned version of my dataset, with the retweets removed from before Aug 08, here is a dataset by @@vbmokin https://www.kaggle.com/datasets/vbmokin/russian-invasion-ukraine-without-retweets

[Nov 19] I noticed that one of my developer accounts, which ISNT TWEETING ANYTHING and just pulling data out of twitter has been permanently banned by Twitter.com, thus the decrease of unique tweets. I will try to come up with a solution to increase my throughput and signup for a new developer account.

[Oct 19] I just noticed that this dataset is finally "GOLD", after roughly seven months since I first uploaded my gzipped csv files.

[Oct 11] Sudden spike in number of tweets revolving around most recent development(s) about the Kerch Bridge explosion and the response from Russia.

[Aug 19- IMPORTANT] I raised the missing dataset issue to Kaggle team and they confirmed it was a bug brought by a ReactJs upgrade, the conversation and details can be seen here https://www.kaggle.com/discussions/product-feedback/345915 . It has been fixed already and I've reuploaded all the gzipped files that were lost PLUS the new files that were generated AFTER the issue was identified.

[Aug 17] Seems the latest version of my dataset lost around 100+ files, good thing this dataset is versioned so one can just go back to the previous version(s) and download them. Version 188 HAS ALL THE LOST FILES, I wont be reuploading all datasets as it will be tedious and I've deleted them already in my local and I only store the latest 2-3 days.

[Aug 10] 3/5 of my Python processes errored out and resulted to around 10-12 hours of NO data gathering for those processes thus the sharp decrease of tweets for Aug 09 dataset. I've applied an exception/ error checking to prevent this from happening.

[Aug 09] Significant drop in tweets extracted, but I am now getting ORIGINAL/ NON-RETWEETS.

[Aug 08] I've noticed that I had a spike of Tweets extracted, but they are literally thousands of retweets of a single original tweet. I also noticed that my crawlers seem to deviate because of this tactic being used by some Twitter users where they flood Twitter w...

Context

There are total of 20 CSV files including tweets related to COVID-19 from 20 March 2020 to 08 April 2020.

Content

For each file, the following columns are included. Columns: coordinates, created_at, hashtags, media, urls, favorite count, id, in_reply_to_screen_name, in_reply_to_status_id, in_reply_to_user_id.

Acknowledgements

We wouldn't be here without the help of others. If you owe any attributions or thanks, include them here along with any citations of past research.

Inspiration

Your data will be in front of the world's largest data science community. What questions do you want to see answered?

Rights

The relevant sections of Twitter's Terms of Service [1] and Developer Agreement [2]. ** According to Twitter's Developer Policy §6 [3]: "If you provide Content to third parties, including downloadable datasets of Content or an API that returns Content, you will only distribute or allow download of Tweet IDs and/or User IDs" and "any Content provided to third parties via non-automated file download remains subject to this Policy". [1] https://twitter.com/tos?lang=en [2] https://dev.twitter.com/overview/terms/agreement [3] https://dev.twitter.com/overview/terms/policy#6.Update_Be_a_Good_Partner_to_Twitter

This dataset contains tweets posted for various services and products along with the emotion contained in the tweet. It contains three columns, the tweet text, the product/service, and the emotion contained in the tweet. It can be used to train various ML models for analyzing the sentiments in the tweets.

If you find the data useful do give an upvote and let me know in the discussions about any improvements.

Cheers!!

🧠 Dataset: 400k Tweets from 7k Users with Age Labels This dataset contains over 400,000 tweets collected from ~7,000 Twitter users, each associated with a known age label. It is designed to support research in age prediction, user profiling, and social media analytics.

📦 Dataset Contents username: Twitter handle or pseudonym (anonymized if necessary)

age: User’s age (numeric or bucketed in your downstream tasks)

tweet_id: Unique ID for each tweet (can be hydrated using Twitter API)

📊 Use Cases Age prediction from social media text

Sociolinguistic analysis by age group

Demographic classification for digital behavior modeling

Pretraining or fine-tuning language models with age-aware context

📚 Origin This dataset was compiled for the research project: “Age Matters: A Deep Learning Framework for Demographic Prediction on X with a Publicly Available Dataset” (2025) by Awais and Bashir The data was collected, cleaned, and annotated with user ages for academic purposes.

If you use this dataset, please cite or acknowledge the research work.

🔒 Notes This dataset includes only tweet IDs (per Twitter policy). Use tools like twarc2 to hydrate tweets.

Usernames are included to group tweets per user but should not be used for identity linking unless properly anonymized.

🔁 Suggested Next Steps Hydrate the tweet text using the Twitter API

Preprocess tweets (remove noise, lemmatize, etc.)

Use models like BERTweet or TF-IDF + classifier for age prediction

Subtitle:

Sentiment analysis of tweets about 6 major US airlines, scraped in February 2015

About Dataset:

This dataset contains around 14,640 tweets about six major US airlines — American, Delta, Southwest, United, US Airways, and Virgin America. The tweets were scraped from Twitter in February 2015 and manually labeled by contributors as positive, negative, or neutral. For tweets marked negative, contributors also tagged a reason (e.g. "late flight", "rude service", "lost luggage", "customer service issue").

Columns:

tweet_id — unique identifier for the tweet airline_sentiment — labeled sentiment (positive / negative / neutral) airline_sentiment_confidence — confidence score for the sentiment label negativereason — reason for negative sentiment (if applicable) negativereason_confidence — confidence score for the negative reason airline — the airline mentioned airline_sentiment_gold / negativereason_gold — gold-standard labels used for quality control name — Twitter username of the author retweet_count — number of retweets text — the tweet content tweet_coord — geo-coordinates (if available) tweet_created — timestamp of the tweet tweet_location — user-reported location user_timezone — user's timezone

Use cases:

Sentiment classification (positive / negative / neutral) Multi-class text classification for negative reasons NLP preprocessing practice (cleaning tweets, TF-IDF, embeddings) Airline customer service analysis

Source: Originally released by Figure Eight (formerly Crowdflower).

This comprehensive dataset is a meticulously curated collection of mental health statuses tagged from various statements. The dataset amalgamates raw data from multiple sources, cleaned and compiled to create a robust resource for developing chatbots and performing sentiment analysis.

Data Source:

The dataset integrates information from the following Kaggle datasets:

• 3k Conversations Dataset for Chatbot
• Depression Reddit Cleaned
• Human Stress Prediction
• Predicting Anxiety in Mental Health Data
• Mental Health Dataset Bipolar
• Reddit Mental Health Data
• Students Anxiety and Depression Dataset
• Suicidal Mental Health Dataset
• Suicidal Tweet Detection Dataset

Data Overview:

The dataset consists of statements tagged with one of the following seven mental health statuses: - Normal - Depression - Suicidal - Anxiety - Stress - Bi-Polar - Personality Disorder

Data Collection:

The data is sourced from diverse platforms including social media posts, Reddit posts, Twitter posts, and more. Each entry is tagged with a specific mental health status, making it an invaluable asset for:

• Developing intelligent mental health chatbots.
• Performing in-depth sentiment analysis.
• Research and studies related to mental health trends.

Features:

• unique_id: A unique identifier for each entry.
• Statement: The textual data or post.
• Mental Health Status: The tagged mental health status of the statement.

Usage:

This dataset is ideal for training machine learning models aimed at understanding and predicting mental health conditions based on textual data. It can be used in various applications such as:

• Chatbot development for mental health support.
• Sentiment analysis to gauge mental health trends.
• Academic research on mental health patterns.

Acknowledgments:

This dataset was created by aggregating and cleaning data from various publicly available datasets on Kaggle. Special thanks to the original dataset creators for their contributions.