The media dependence theory suggests that the more complex the social changes, the deeper the investors rely on the media. Because of the nontransparency of Taiwan's real estate market, the news media serves as a channel to obtain information. This study analyzed the effects of media on the real estate market by using text mining to compile real estate media sentiment index and analyzing its relationship with the real estate market. The media sentiment index had no significant relationship with price concession and transaction volume, but was significantly related to transaction price and time on the market, indicating that the media’s attitude toward the real estate market affects investors. The frequency of real estate news reports significantly correlated with the price, volume, and time on market. The results also indicated that the increase in news reports boosts investors' expectations. Through copula's dynamic correlation analysis, we found that the dynamic correlation between house prices, transaction volume, and media sentiment index began to change in 2012. This study confirmed the role of media sentiment in the market, capturing the nonfundamental side of the real estate market. Our findings can help the government to make more efficient real estate market policies.
web crawler, text mining, real estate market, media sentiment