Big Data Text Analysis Home -- Download -- Instructions -- FAQ

****From April 2023 there is no longer free access to Twitter, unless you have the Academic API****

A free Windows program for keyword, issue, time series, sentiment, gender and content analyses of (mainly) social media texts. *download*

Academic researchers and students can use Academic Research Mozdeh

For updates, follow @Mozdeh4

• This probably stopped working in April 2023 but I have not tested it: Before using Mozdeh please get a sign up to Twitter and register as a developer to get a Twitter developer "bearer token". This gives 500k free tweets per month collected by Mozdeh.
• Free online course 1: Introduction to YouTube and Twitter analysis with Mozdeh.
• Free online course 2: Word association thematic analysis with Mozdeh. See also the Word Association Thematic Analysis information page.
• Presentation files summarising four Covid-19 studies.

Gather or import texts

• [This no longer works with free Twitter] Quick Twitter demo video of downloading and installing Mozdeh, collecting tweets, searching, filtering and keyword analysing tweets.
• [This no longer works with free Twitter] Gather recent tweets with keyword searches. The tweets will be a maximum of 7 days old. Keep Mozdeh switched on to gather a large collection of tweets for weeks, months or years. This allows up to 18,000 tweets per 15 minute window - or up to 1.7 million per day. Its results are not necessarily comprehensive. Older tweets matching keyword searches must be bought from Twitter or a reseller or accessed with Academic Research Mozdeh.
• [This no longer works with free Twitter] Gather tweets from one or more users: up to the most recent 2,000-3,200 tweets per user.
• Get Tweets from the full Twitter archive using the Academic Research Mozdeh (requires permission from Twitter).
• Gather comments on YouTube videos - about a million per day. Video about downloading videos.
• Gather comments from subreddits on Reddit.
• Import academic citations, TripAdvisor reviews, Reddit comments, or other formatted or unformatted texts: see the Import Data button in the Startup Wizard. Facebook importing is no longer possible.

Analyse

These options work for all languages with spaces between words. Special instructions for Japanese and Chinese.

• Search or filter the texts gathered with keyword queries, sentiment strength, country [Twitter only], retweets/likes, and/or author gender (male, female, nonbinary).
• Mine word associations to get insights into important issues within your texts by topic, gender, country, time, popularity, and sentiment.
• **Word Association Thematic Analysis**: to discover themes in your text. Video and Book about the method.
• Researching politics on social media with Mozdeh - video by Dr Rosalynd Southern.
• Search by sentiment, measure the average sentiment for a query, or find terms associating with positive or negative sentiment.
• Mine gender differences in the texts.
• Find the average retweet/citation/like count for texts matching a query/gender/sentiment (includes confidence intervals).
• Draw time series graphs of the volume of posts in your topic or the percentage of texts matching any query in your topic. See also examples of time series analyses.
• Create networks of users from their texts.
• Create time series graphs for an individual user in your collection (thanks to Kim Holmberg for these instructions).
• Draw a time series graph of the average positive and negative sentiment strength of texts overall or for texts matching a query in your topic (see the bottom of the main time series graphs page).
• Find words with a sudden frequency increase (i.e., keyword spike detection) in the data.
• Generate topic models to find the main topics of discussion in the texts.

Save selected texts

• Save a random sample of texts for a content analysis. If you already have a sample and need to randomise it you could use Excel.

Examples of papers and books written with Mozdeh

Thelwall, M. (2021). Word association thematic analysis: A social media text exploration strategy. San Rafael, CA: Morgan & Claypool. [A method to analyse text for themes with Mozdeh.]

Thelwall, M. (2018). Social web text analytics with Mozdeh. University of Wolverhampton. [free short overview book]

Mozdeh was used to collect and analyse data for the following.

Thelwall, M., Thelwall, S. & Fairclough, R. (2021). Male, female, and nonbinary differences in UK Twitter self-descriptions: A fine-grained systematic exploration. Journal of Data and Information Science, 6(2), 1-27.

Thelwall, M., Makita, M., Mas-Bleda, A., & Stuart, E. (2021). “My ADHD hellbrain”: A Twitter data science perspective on a behavioural disorder. Journal of Data and Information Science, 6(1).

Thelwall, S. & Thelwall, M. (2020). Anthropomorphizing atopy: Tweeting about eczema. Journal of the Dermatology Nurses' Association, 12(2), 74-77.

Potts, G., & Radford, D. R. (2019). #Teeth&Tweets: the reach and reaction of an online social media oral health promotion campaign. British Dental Journal, 227(3), 217-222.

Thelwall, M. (2018). Can museums find male or female audiences online with YouTube? Aslib Journal of Information Management, 70(5), 481-497.

Thelwall, M. & Mas-Bleda, A. (2018). YouTube science channel video presenters and comments: Female friendly or vestiges of sexism? Aslib Journal of Information Management, 70(1), 28-46.

Thelwall, M. (2018). Can social news websites pay for content and curation? The SteemIt cryptocurrency model. Journal of Information Science, 44(6), 736–751.

Ahmed, W. (2018). Using Social Media Data for Research: An Overview of Tools. Journal of Communication Technology, 1(1), 77-94.

Thelwall, M. (2018). Social media analytics for YouTube comments: Potential and limitations. International Journal of Social Research Methodology, 21(3), 303-316. doi:10.1080/13645579.2017.1381821 [ local copy]

Holmberg, K., & Hellsten, I. (2014). Analyzing the climate change debate on Twitter: content and differences between genders. In Proceedings of the 2014 ACM conference on Web science (pp. 287-288). ACM.

Terlumun, I. T., Appollm, Y. I., Ibrahim, F. J., Mamman, F. S., Yusuf, A. K., & Ibrahim, A. H. (2018). Determining the Effectiveness of YouTube Videos in Teaching and Learning with Mozdeh Algorithm. International Journal of Education and Evaluation.

Thelwall, M., & Cugelman, B. (2017). Monitoring Twitter strategies to discover resonating topics: The case of the UNDP. El profesional de la información.

Thelwall, M., Buckley, K., & Paltoglou, G. (2011). Sentiment in Twitter events. Journal of the American Society for Information Science and Technology, 62(2), 406-418.

Thelwall, M. (2008). No place for news in social network web sites? Online Information Review, 32(6), 726-744.

Extra resources

• FAQ
• User guide
• Removing spam texts.
• Theoretical overview and introduction to time series and sentiment analysis for Twitter chapter.
• [This no longer works with free Twitter] Crash course.
• [This no longer works with free Twitter] June 2017 one day quick course - slides and workshop.
• The YouTube Comment Term Frequency Comparison (CTFC) method.
• Not about Mozdeh, but this paper gives useful ideas about interpreting the results: Sen, I., Flöck, F., Weller, K., Weiß, B., & Wagner, C. (2021). A total error framework for digital traces of human behavior on online platforms. Public Opinion Quarterly, 85(S1), 399-422.

Mozdeh Screenshots

Sentiment and retweet/reviews/citation graphs and the top of the word frequency table.

Mozdeh can produce a time series graph of the volume of texts for your topic. The graph below is a section of a screenshot showing the hourly volume of tweeting for a corpus of tweets related to the 2011 UK Riots in 2011.

Mozdeh can produce a time series graph of the percentage of texts matching a query for your topic. The graph below is a section of a screenshot showing the hourly percentage of tweets mentioning police in a UK Riots 2011 corpus.

Mozdeh can produce a time series graph of the average positive and negative sentiment of texts for your whole topic or just the texts matching a query for your topic. The graph below is a section of a screenshot showing the hourly average sentiment of tweets mentioning police in a UK Riots 2011 corpus.

Mozdeh can search for texts matching a query for your topic. The picture below is a section of a screenshot showing some of the results of a search for Wolverhampton in a UK Riots 2011 Twitter corpus.

If you can't use Windows, then you might try COSMOS for Mac/Linux or investigate this list of social media software.