Empirical study of topic modeling in Twitter

By Liangjie Hong, Brian D. Davison. 2010

1,233 citations recorded by the source corpus.

Social networks such as Facebook, LinkedIn, and Twitter have been a crucial source of information for a wide spectrum of users. In Twitter, popular information that is deemed important by the community propagates through the network. Studying the characteristics of content in the messages becomes important for a number of tasks, such as breaking news detection, personalized message recommendation, friends recommendation, sentiment analysis and others. While many researchers wish to use standard text mining tools to understand messages on Twitter, the restricted length of those messages prevents them from being employed to their full potential. We address the problem of using standard topic models in micro-blogging environments by studying how the models can be trained on the dataset. We propose several schemes to train a standard topic model and compare their quality and effectiveness through a set of carefully designed experiments from both qualitative and quantitative perspectives. We show that by training a topic model on aggregated messages we can obtain a higher quality of learned model which results in significantly better performance in two real-world classification problems. We also discuss how the state-of-the-art Author-Topic model fails to model hierarchical relationships between entities in Social Media.

Subfields: Artificial Intelligence, Statistical and Nonlinear Physics

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 Statistical and Nonlinear Physics

Citations in the corpus

Is the Sample Good Enough? Comparing Data from Twitter's Streaming API with Twitter's Firehose (2013, methodology): “It has been used to collect data for topic modeling (Hong and Davison 2010; Pozdnoukhov and Kaiser 2011), network analysis (Sofean and Smith 2012), and statistical analysis of content (Mathioudakis and Koudas 2010), among others.”
Understanding the Limiting Factors of Topic Modeling via Posterior Contraction Analysis (2014, background): “However, the results are mixed when it is applied to unconventional data sets such as tweets (i.e., short messages posted on Twitter.com) (Hong & Davison, 2010; Zhao et al., 2011), query logs (Bing et al., 2011; Carman et al., 2010), digital books (Mimno & McCallum, 2007), and metadata records…”
A biterm topic model for short texts (2013, methodology): “…occurrences of words in short document play less discriminative role compared to lengthy documents where the model has enough word counts to know how words are related [17] ; 2) The limited contexts make it more d-iﬃcult for topic models to identify the senses of ambiguous words in short documents.”
Using Topic Modeling Methods for Short-Text Data: A Comparative Analysis (2020, methodology): “Hong and Brian Davison (2010) compared the performance of the LDA method and author–topic models on the Twitter platform.”
Of Pins and Tweets: Investigating How Users Behave Across Image- and Text-Based Social Networks (2014, methodology): “…and added the following five additional categories, since they have been commonly identified as important in Twitterrelated studies (Duh et al. 2011; Hong and Davison 2010; Lee et al. 2011; Zhao et al. 2011): " Politics " , " Religion " , " Charity " , " Twitter-Specific " , and " Business " .”
Improving Topic Models with Latent Feature Word Representations (2015, background): “One approach is to combine short texts into long pseudo-documents before training LDA (Hong and Davison, 2010; Weng et al., 2010; Mehrotra et al., 2013).”
Learning similarity functions for topic detection in online reputation monitoring (2014, background): “Hong & Davison [18] conducted experiments on the quality of topics derived from different aggregation strategies.”
Short Text Topic Modeling Techniques, Applications, and Performance: A Survey (2019, background): “Other information includes hashtags, timestamps, and named entities have been tread as metadata to merging short texts [17], [19], [38].”
Topic Modeling for Short Texts with Auxiliary Word Embeddings (2016, methodology): “Other metadata that have been used for short text aggregation include hashtags, timestamps, and named entities [13, 19, 42].”
Comparing Twitter and Traditional Media Using Topic Models (2011, methodology): “Our proposed Twitter-LDA differs from the models studied in [7] in that we model one tweet with one topic label, which is similar to [10,11] but for different applications.”
A Topic Modeling Comparison Between LDA, NMF, Top2Vec, and BERTopic to Demystify Twitter Posts (2022, methodology): “In order to eﬀectively extract features from a large corpus of text data, numerous text mining approaches have been introduced (Li et al., 2019), among which topic modeling serves as the most frequently adopted technique (Hong and Davison, 2010).”
When is it biased?: assessing the representativeness of twitter's streaming API (2014, methodology): “It has been used to collect data for topic modeling [10, 16], network analysis [19], and statistical analysis of content [12], among others.”
Improving LDA topic models for microblogs via tweet pooling and automatic labeling (2013, methodology): “An intuitive solution to this problem is tweet pooling [13, 5]: merging re-lated tweets together and presenting them as a single document to the LDA model.”
DeepHate: Hate Speech Detection via Multi-Faceted Text Representations (2020, background): “Often, in mining topics from Twitter posts, the documents are formed by aggregating the posts by hashtags, published time, or authors [23, 48].”
Predictive Analysis on Twitter: Techniques and Applications (2018, methodology): “Furthermore, utilizing the extracted topics as features for a supervised model improved the performance of the classification task in [69].”