台灣房地產市場輿論與從眾行為之房價泡沫分析

近年來台灣充斥著許多房價泡沫相關的新聞，許多專家皆對房市的走勢有著不同的切入觀點。在社群論壇上，也有許多民眾會討論自己所在區域的房價概況。近期在財務的研究議題上，也開始出現財務數據與非結構化的文字資料結合的文獻，文字可透過文字探勘的技術量化成數據來分析。因此有許多文獻透過網路爬蟲抓取財務相關新聞、論壇等文字資料，建構情緒指標來探討文字與財務數據之間的關聯性。而根據過去的文獻，民眾情緒、從眾行為 (herding behavior) 以及房價泡沫的程度皆會影響到房價的走勢。本文透過網路爬蟲抓取 Mobile01 論壇以及 PTT 論壇中房屋討論版的文章與留言。藉由上述論壇中的文字資料建構情緒指標，探討在不同的社群論壇與台灣直轄市之下，情緒指標、從眾行為與房價泡沫對於房價影響的差異。

關鍵字

情緒指標；從眾行為；房價泡沫；房價指數；文字探勘

並列摘要

In recent years, Taiwan has been flooded with news related to the housing price bubble, and many experts have different perspectives on the trend of the housing market. On community forums, there are also many people who will discuss the general situation of housing prices in their area. Recently, on financial research topics, documents combining financial data and unstructured text data have also begun to appear. Text can be quantified into data for analysis through text mining technology. Therefore, many documents use web crawlers to grab financial-related news, forums and other textual data to construct sentiment indicators to explore the correlation between text and financial data. According to past literature, public sentiment, herding behavior, and the degree of housing price bubbles will all affect the trend of housing prices. This article uses web crawlers to grab articles and comments on the housing discussion board in the Mobile01 forum and the PTT forum. Construct sentiment indicators based on the text data in the above forums, and explore the differences in the impact of sentiment indicators, herd behavior, and housing price bubbles on housing prices in different community forums and municipalities of Taiwan.