Social Media Data Analytics for the US Construction Industry: Preliminary Study on Twitter

The increasing use of the Internet for many purposes is creating big data, many of which are generated from social media. These big data potentially could assist in obtaining valuable administrative information and even explore new social phenomena. Traditional ways of collecting data, such as questionnaire surveys, are time-consuming and costly. Therefore, the use of social media affords the opportunity to extract information that might be of benefit to the construction industry in a responsive and inexpensive manner. To this end, this paper explores whether information and knowledge that would be valuable in the construction domain can be generated by analyzing social media data. Twitter was selected for an initial trial analysis because of its wide usage in the United States. Because they represent a majority of the construction users in Twitter, the following four user clusters were selected and analyzed: construction workers, construction companies, construction unions, and construction media. For each user identified in the four clusters, the 3,200 most recent Twitter messages were collected, which were analyzed from the following aspects: sentiment analysis, topic modeling, link analysis, geolocation analysis, and timeline analysis. Different data-analysis methods were used for the specific themes, such as Stanford Natural Language Processing (StanfordNLP) for sentiment analysis. The detailed findings, benefits, and barriers to incorporating social media data analytics in the construction industry, as well as future research directions, are discussed in this paper. For example, the sentiment analysis results indicated that construction workers tend to have a higher proportion of negative messages compared to the other clusters, which may prompt more attention to emotional guidance and understanding by construction companies and the public. This paper benefits academia by testing an alternative way of studying the construction population, which could help decision makers gain a better understanding of real-world situations in the construction industry. (C) 2017 American Society of Civil Engineers.