運用詞頻分析技術於XBRL財報與附註關聯之研究-以投資性不動產為例

Applying Zipf’s Law for XBRL Financial Report and Footnote Related Research-The Example of Investment Property

林聖訓(Sheng-Syun Lin)

指導教授 ： 黃士銘

國立中正大學/管理學院/資訊管理學系/碩士(2013年)

摘要

可延伸商業報導語言（XBRL）可說是一個全球化企業報告的共通語言，為了與國際先進國家接軌，我國政府積極推動XBRL與IFRS之導入適用於財務申報。然而，當我國上市櫃公司全面採用IFRS編制報表時需注意與過去一般公認原則差異變動之處，尤為公允價值影響之會計準則部分，投資者在閱讀與分析財報同時，更需仰賴附註揭露提供額外解釋能力以了解企業選擇政策之過程與理由。因此對於表述企業營運之脈絡與價值之文字敘述部分，本研究欲透過系統開發法之雛型開發法，結合XBRL、Zipf法則建立一個詞頻分析機制來輔助查核人員或投資者，運用此機制達到檢核特定公司群或異常詞頻解釋之目的，聚焦詞頻異常之公司以期增加附註判斷之輔助參考。系統建置後，本研究以投資性不動產之重大會計政策附註揭露為例進行實驗測試。測試各模組之試驗與分析數值結果過程中發現揭露之內容會與會計事務所不同而品質有所影響外，透過此機制也能找出與其它揭露差異程度大之公司。

關鍵字

投資性不動產 ； 詞頻分析 ； 齊普夫 ； 可延伸商業報導語言

並列摘要

Extensible Business Reporting Language (XBRL) could be deemed to a common language of the global corporate financial reports. To catch up with the pace of advanced country, our country is actively promoting XBRL and International Financial Reporting Standards (IFRS) for applying to financial reporting. As our country fully prepared to adopt the IFRS as our financial reporting framework, we should note the difference with the Generally Accepted Accounting Principles (GAAP) and the IFRS, particularly the impact of fair value. While reading and analyzing the financial report, we also rely on the disclosure of footnote that provides additional insight into the companies of what the accounting policies and reasons they have chosen.
For these depiction that describe the business operations and the value of the investment, this study’s aim is to establish a mechanism, which combine XBRL and the word-frequency analysis of Zipf’s law to assist users achieving the purpose of finding a specific company checklist group ,and the explanation of unusual word-frequency. It can help us to focus on anomalies word-frequency in order to increase the company's auxiliary reference by judgments on footnote. After the system is built, this study takes the footnotes disclosure of the investment property in significant accounting policies as a testing case. From the result and the data analysis of each module found that the contents and qualities of the footnotes varies due to accountancy firms’ effects. Besides, through this mechanism also be able to identify the degree of difference footnote disclosures with other companies.

並列關鍵字

Investment Property ； XBRL ； Zipf

參考文獻

林千翔、張嘉惠、陳貞伶。(民99)。結合長詞優先與序列標記之中文斷詞研究。中文計算語言學期刊, 15(3-4), 161-179。

Goh, Chooi-Ling, Asahara, Masayuki, & Matsumoto, Yuji. (2005). Chinese word segmentation by classification of characters. Computational Linguistics and Chinese Language Processing, 10(3), 381-396.

葉又豪（民100）。運用文字探勘分析非量化資訊協助投資人預測公司財務表現。國立政治大學會計研究所碩士論文。

江韋達（民100）。運用資料及文字探勘探討不同市場營運概況文字敘述及財務表現之一致性。國立政治大學會計研究所碩士論文。

葉乃菁、王玳琪、張嘉珍、吳騏、賴志遠（民98）。建構創新政策研究工具—文字探勘之應用簡介。科技政策研究與資訊中心。取自http://www.narlabs. org.tw/tw/pressroom/topic/topic.php?group_id=21&topic_id=79。