Academic research > Research areas

Discover your next research project

Inspiration for your next research study can come from anywhere, and exploring what others have done with Twitter data can help.

Research areas

Academic researchers can use Twitter data to study nearly any topic, across nearly any discipline. If working with social media data is your specialty, then you might be interested in browsing some of Twitter’s research goals.

These are some of the biggest questions and challenges that teams at Twitter are working through today. We believe that more scholarship on these topics, whether from Twitter, or from academic researchers around the world, can help make the world a better place.

While this list of topics is not exhaustive, these are areas where Twitter is keen to learn more from the academic community’s expertise. For your consideration, we’ve shared sample research objectives related to the topic, as well as some high-level reference materials to explore.

Jump to: 
 

Promote healthier online conversations

Identify ways to improve experiences on Twitter and promote healthy online behavior.

Example Research Objectives

  • Research and applied methods to reduce and halt online abuse and hate speech

  • Research or data modeling methods that identify toxicity and unhealthy Tweets in non-English languages

  • Understand the experience, harms and effects of targeted or coordinated online abuse or harassment (e.g. such as with doxxing, dogpiling)

  • Trends or variances in online abuse or harassment across the world and/or in times of conflict

  • Increase understanding of polarizing or controversial topics, and receptivity to new or contradictory information

  • Examine government and elected official responses and its impact on civil discourse (e.g., in public health crisis, in times of public unrest)

  • Study immediate and lasting impacts of Twitter feature changes (e.g., who can reply to a Tweet)

Reference materials from Twitter

 

Responsible machine learning at global scale

Build responsible, responsive, and community-driven machine learning methods, systems, and applications for use.

Example Research Objectives - Machine learning systems/methods

  • NLP methods for extracting entities and/or micro-topics from Tweets

  • NLP techniques to identify and validate sentiment in all languages

  • Modeling techniques for identifying toxicity in Tweets, images, and audio in non-English languages

  • Applications of deep learning or convolutional neural networks to natural language processing

  • Graph topology methods (e.g. how a network’s shape determines the spread of information, comparing diffusion networks)

Example Research Objectives- Applied machine learning/data science

  • Research and development of responsible machine learning systems, such as: systems for removing bias, content amplification, procedural justice

  • Techniques for greater fairness and remove bias in natural language, media, images, and audio classifiers

  • Representation learning data efficiency in model training and inference, including: self-supervision, pre-training, reduced manual data labeling

  • Techniques to improve real-time recommendation systems with active, continual, reinforcement learning

  • Techniques to improve recommendation systems with an implicit feedback loop, casual reinforcement learning

  • Modeling methods to group and identify user interests by behavioral and contextual signals to improve real-time recommendation systems

Reference materials

 

Identify and prevent platform manipulation

Improve detection and prevention of malicious automated activity, misinformation, disinformation, and information operations.

Example Research Objectives- Misinformation or disinformation

  • Research and models that identify the reach and potential for harm of misleading information campaigns or groups on Twitter across the globe

  • Research and models that identify the reach and impact of various misinformation operations across the globe

  • Detection methods and identification of coordinated manipulative behavior (e.g., information operations or state manipulation)

  • Examine government and elected official responses and their impacts on public discourse (e.g., how official responses impact perceptions of public health crisis)

  • Techniques for the identification of misleading information at global scale, in non-English languages

  • Reviews of current Twitter policy and enforcement tactics and its impact on platform manipulation practices

Example Research Objectives- Automation detection

  • Identification methods for spam, filtering methods for spam

  • Identification methods for malicious bots/automated account activity

  • Delineation between automated behavior and genuine users (e.g., new accounts, grassroots movements)

  • Identification methods for automated activity that is beneficial vs. harmful, manipulative

  • Impact analysis around Twitter policies regarding platform manipulation and spam

Reference materials

 

Identify market changes from conversations

Identify how conversations on Twitter impact different industries, markets, and future trends.

Example Research Objectives

  • Techniques for extracting entities and/or micro-topics from Tweets for predictive analysis, futurecasting

  • Prediction models assessing features that impact reach and engagement with Tweets

  • Applications of Twitter data in support of business operations (e.g., supply chain insight, predictive analytics, forecast revenue)

  • Applications of Twitter data for decision-making related to brand reputation, new market penetration, behavioral economics research

Reference materials

    Submit your research publication

    Academics have a lot of expertise to share with Twitter, so we are doing our biggest “lit review” of published academic research that uses Twitter data. If you authored a publication that used the Twitter API, consider sharing it below to enrich our study.

    Submit Here

    Find the right data for your next academic study

    If it’s being talked about in the world, it’s probably being talked about on Twitter. The Academic Research product track grants access to historical and real-time public Twitter data, helping advance research objectives for nearly any discipline. 

    View Academic Research product track

     

    Take the next step

    Learn more about Twitter data for academic researchers.

    Resources →
    Share feedback →
    Researchers forum →
    Guide to the Twitter API →

    By using Twitter’s services you agree to our Cookies Use. We use cookies for purposes including analytics, personalisation, and ads.