Cluster Analysis of Twitter Data: A Review of Algorithms

Abstract

Twitter, a microblogging online social network (OSN), has quickly gained prominence as it provides people with the opportunity to communicate and share posts and topics. Tremendous value lies in automated analysing and reasoning about such data in order to derive meaningful insights, which carries potential opportunities for businesses, users, and consumers. However, the sheer volume, noise, and dynamism of Twitter, imposes challenges that hinder the efficacy of observing clusters with high intra-cluster (i.e. minimum variance) and low inter-cluster similarities. This review focuses on research that has used various clustering algorithms to analyse Twitter data streams and identify hidden patterns in tweets where text is highly unstructured. This paper performs a comparative analysis on approaches of unsupervised learning in order to determine whether empirical findings support the enhancement of decision support and pattern recognition applications. A review of the literature identified 13 studies that implemented different clustering methods. A comparison including clustering methods, algorithms, number of clusters, dataset(s) size, distance measure, clustering features, evaluation methods, and results was conducted. The conclusion reports that the use of unsupervised learning in mining social media data has several weaknesses. Success criteria and future directions for research and practice to the research community are discussed.

An Empirical Performance Evaluation of Semantic-Based Similarity Measures in Microblogging Social Media

“…This problem is more prevalent in Online Social Network (OSN) texts due to the informal nature and the high degree of lexical variations used. Areas of work within related fields, such as classification and clustering of tweets face similar issues when identifying similarities in natural language text presented in Twitter [9]. To illustrate some challenges present in Twitter, consider the following tweet [10]: "#qcpoli enjoyed a hearty laugh today with #plq debate audience for @jflisee #notrehome tune was that the intended reaction?"…”

“…In terms of social media data, particularly Twitter, the task becomes much harder due to many inaccuracies that may be present in the short pieces of text. These inaccuracies include: 1) Poor grammatical and syntactical structure due to the character limit which encourage the frequent use of abbreviations and irregular expressions [9]. 2) Misspellings, out-of-vocabulary words, and acronyms.…”

Consensus Clustering of Tweet Networks via Semantic and Sentiment Similarity Estimation

“…With the development of social media, new forms of communication and information acquisition have been established [1]. Twitter 1 is one of the most popular social media networks [2], [3] and has become an important source of information [4]. Through short text messages called tweets, people can deliver and receive information such as amusing anecdotes and news about natural disasters or political events [4].…”

“…Twitter 1 is one of the most popular social media networks [2], [3] and has become an important source of information [4]. Through short text messages called tweets, people can deliver and receive information such as amusing anecdotes and news about natural disasters or political events [4]. However, the number of accessible tweets is too large for users to easily find their desired information.…”

“…To overcome this difficulty, it is useful to create an overview of many tweets by grouping tweets with similar topics. Because of the nature of social media, tweets with unforeseen topics may appear [4], [5]. Therefore, clustering, i.e., an unsupervised method of grouping data, is especially suitable.…”