網路旅遊資訊探勘

Abstract

旅遊是無煙囪的產業，所帶來的經濟效益逐年穩定成長，了解旅遊趨勢對於旅遊行程的規劃者或旅遊品質的改善者而言是相當重要的課題。根據交通部觀光局旅遊統計資料顯示，國人國內旅遊普遍集中在居住的區域。以此特性為依據，本研究透過對社群網路上的旅遊主題資訊的分析，希望找出特定區域的旅遊者所關注的議題與趨勢。本研究認為，雖然同區域的旅遊者其旅遊傾向應當相似，然考慮決定旅遊傾向的因素之複雜性，必定可以找出各地方些許的差異。 網路呈現的資訊普遍為非結構化或半結構化，本研究透過一系列的前置處理，資料的過濾與結構化、萃取隱含的資訊，並以Apriori 演算法對桃園縣、新竹市、新竹縣、苗栗縣的旅遊主題資訊進行關聯規則分析。結果顯示，桃園縣與新竹市旅遊傾向相似度高，受關注的旅遊類型為「飲食」類的主題；新竹縣的旅遊傾向較接近新竹市與苗栗縣之間，特別的是負面的旅遊主題普遍受到關注；苗栗縣的旅遊傾向以「賞景」型態的主題最受關注，然而「賞景」型態的主題的持續關注度普遍是低的。

Tourism is one kind of industry without pollution, and brings stable economic growth in last ten years. It is quite important to know the trends of tourism for people who make tourism plan or dedicate to improve the quality of tourism. According to the report of Tourism bureau, most of Taiwanese travel at the region of their current address place of residence. This paper aims to know what kinds of topics will be given special attention. Although tourists have similar behavior at the same region, but consider the complexity of tourism plan making, it must have few differences. Most of information is unstructured or semi-structured on web forum. Through a series of data pre-processing, tourism information can be extracted. Finally, here processing web forum's data is processed with Apriori algorithm. According to the analysis, the behaviors of tourism are highly similar at Taoyuan County and Hsinchu City. People tend to pay attention on topic of tourism about "Food"; the behavior of Hsinchu County's tourists fall between Taoyuan County and Hsinchu City. Moreover, negative topics are paid highly attention; the most hit tourism topics of Miaoli County are all about "sightseeing". But these topics are not paid attention continually.