Characterizing Microblogs with Topic Models

By Daniel Ramage, Susan T. Dumais, Daniel J. Liebling. International Conference on Web and Social Media, 2010

810 citations recorded by the source corpus.

As microblogging grows in popularity, services like Twitter are coming to support information gathering needs above and beyond their traditional roles as social networks. But most users’ interaction with Twitter is still primarily focused on their social graphs, forcing the often inappropriate conflation of “people I follow” with “stuff I want to read.” We characterize some information needs that the current Twitter interface fails to support, and argue for better representations of content for solving these challenges. We present a scalable implementation of a partially supervised learning model (Labeled LDA) that maps the content of the Twitter feed into dimensions. These dimensions correspond roughly to substance, style, status, and social characteristics of posts. We characterize users and tweets using this model, and present results on two information consumption oriented tasks.

Subfields: Artificial Intelligence, Information Systems, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/1948cf217cd76c02171b17aff75ff6980c5f09aa

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Statistical and Nonlinear Physics

Citations in the corpus

Tweeting is believing?: understanding microblog credibility perceptions (2012, methodology): “This has prompted researchers to develop novel interfaces and algorithms for filtering tweets, such as tools like Twahpic [28] or Eddi [1], which perform topic-based clustering and filtering, or algorithms for identifying Twitter authors with authority on specific topics [26].”
Towards Decision Support and Goal Achievement: Identifying Action-Outcome Relationships From Social Media (2015, background): “With the wide-spread adoption of social networking services over the last 10-15 years, much research has focused on understanding people’s motivations and participatory behavior on these sites, both from a qualitative as well as quantitative perspective [27, 14, 19, 21, 13, 32, 26, 36, 6, 42].”
Improving retrieval of short texts through document expansion (2012, methodology): “Ramage, Dumais and Leibling address the vocabulary mismatch problem in microblog text using latent Dirichlet allocation (LDA) [23], allowing the LDA model to supplement observed term probabilities in document representation.”
Topic Modeling for Short Texts with Auxiliary Word Embeddings (2016, background): “…efﬁcient models to discover the latent topics from short texts become fundamental to many applications that require semantic understanding of textual content, such as user interest proﬁling [38], topic detection [37], comment summarization [31], content characterizing [32], and classiﬁcation [34].”
Improving LDA topic models for microblogs via tweet pooling and automatic labeling (2013, background): “Prior work on topic modeling for tweets includes the following: [11] provide a scalable implementation of a partially supervised learning model (Labeled LDA).”
Comparing Twitter and Traditional Media Using Topic Models (2011, background): “Recently, [ 9 ] applied labeled-LDA to Twitter, but the model relies on hashtags in Twitter, which may not include all topics.”
What to do about bad language on the internet (2013, background): “Other investigators have found that much of the content in Twitter concerns social events and self presentation (Ramage et al., 2010), which may encourage the use of less formal registers in which socially-marketed language is uninhibited.”
Tweets are forever: a large-scale quantitative analysis of deleted tweets (2013, background): “Such efforts have involved manual coding of small samples [17], utilizing the crowd [2] and automatic topic modeling for large samples [19].”
#TwitterSearch: a comparison of microblog search and web search (2011, methodology): “We use Latent Dirichlet Allocation (LDA) [3], a popular unsupervised latent variable topic model from the machine learning community that has been applied to Web documents for information retrieval [29] as well as to posts from Twitter [24].”
Of Pins and Tweets: Investigating How Users Behave Across Image- and Text-Based Social Networks (2014, methodology): “We chose L-LDA because prior work has shown it to be more effective than LDA at extracting topics from microblogs (Quercia, Askham, and Crowcroft 2012; Ramage, Dumais, and Liebling 2010).”
Exploiting Social Relations and Sentiment for Stock Prediction (2014, methodology): “For model inference, we use collapsed Gibbs sampling (Bishop, 2006) and the symmetric Dirichlet Priors are set to: 𝜂 = 0.01, 𝛼 = 0.01 as suggested in (Ramage et al., 2010).”
Modeling Sentences in the Latent Space (2012, background): “In contrast, existing work on modeling short texts focuses on exploiting additional data, e.g., Ramage et al. (2010) model tweets using their metadata (author, hashtag, etc.).”
Learning similarity functions for topic detection in online reputation monitoring (2014, background): “The general assumption is that each author has a certain distribution of topics, while each tweet is associated only to one topic [38, 31].”
Spatiotemporal social media analytics for abnormal event detection and examination using seasonal-trend decomposition (2012, background): “Similarly, there are many related works for non-spatial temporal document collections, for example IN-SPIRE [36], which is a general purpose document analysis system that depicts document clusters on a visual landscape of topics.”
Analysis of user comments: An approach for software requirements evolution (2013, methodology): “Topic modeling, including unsupervised analysis and automatic inference of semantic topics, has previously been used to extract opinions from text corpora such as Twitter, YouTube, and blogs and the results have been used to suggest other relevant items (movies/videos, people, events, and groups) to”
Beyond Trending Topics: Real-World Event Identification on Twitter (2011, background): “Twitter has attracted specialized attention among social media sites, with recent efforts focusing on detection of general topics [19] and trending topics [16, 6] in Twitter messages.”
Natural Language Processing to the Rescue? Extracting "Situational Awareness" Tweets During Mass Emergency (2011, background): “Therefore, automatic topic modeling of tweet content is an ongoing interest (Ramage et al., 2010).”
Twitinfo: aggregating and visualizing microblogs for event exploration (2011, background): “Even the summary tweets sometimes could not explain the headline news: statistical summarization interfaces could help convey more meaning [25, 4, 21].”