從新聞文件中預測與分析企業競爭活動

Predicting Firms’ Competitive Actions from Business News Documents

陳昱安(Yu-An Chen)

指導教授 ： 魏志平

國立臺灣大學/管理學院/資訊管理學研究所/碩士(2015年)

https://doi.org/10.6342/NTU.2015.00416

摘要

在這個時代，經營環境已經變成高度競爭以及邁向全球化，企業經常發動積極的行動去獲得優勢位置或者產業上的獲利。因此，如何判斷市場時機去發動競爭活動就變得很重要。然而，在遭受競爭活動影響時，企業也會採取競爭性回應行動去回應對手的競爭性活動，削弱對手可取得的競爭優勢。但是，被動的回應可以選擇的行動就會受到限制，也不容易取得更好的競爭優勢，如果可以預測對手的下一步競爭活動，就有辦法及早做出預防行動，或提前破壞對手的計畫，進而奪取更加有利的競爭位置。
因此，本研究目的是建構一個預測系統，依據過去企業本身的競爭行動、市場上所有競爭者的競爭行動、以及企業直接競爭者的競爭行動，來預測該企業未來的競爭行動。考慮到系統的自動化，問卷方式的資料來源並不可行，因此我們決定採取擁有大量資料的新聞文件。本研究會說明如何取得文件以及文件的前處理，並且因為資料的不對稱性，我們提出一個解決方法，除此之外，我們也針對系統嘗試不同的研究。結果顯示，透過新聞文件是有辦法作預測模型，並且以次數的資料表達方式所產生的預測結果更好，最後，我們也發現不同的競爭活動需要考慮的面相不同，資料也應該選用不一樣的特徵。

關鍵字

競爭行動預測 ； 競爭行動探勘 ； 競爭行動 ； 新聞文件運用 ； 文字探勘 ； 資料探勘

並列摘要

Due to highly competitive and globalized business environments, firms frequently take aggressive actions to challenge their competitors in an effort to gain advantageous position on their market or improve relative performance. Therefore, it is an important issue how to identify market opportunities and timing to initiate and develop effective competitive actions. However, when a firm is affected by competitors’ competitive actions, it will undertake competitive counteractions in response to competitors’ actions to destroy or weaken rivals’ competitive advantages. But, passive responses are limited and cannot easily gain competitive advantages. If a firm can predict its rivals’ future competitive actions, it can initiate some preventive actions early or even destroy its rivals’ plan in order for the firm to capture better competitive position. Therefore, the objective of this study is to build a prediction system. According to the earlier competitive actions of a focal firm itself, all competitors’ competitive actions in market, and its direct competitors’ competitive actions, our proposed system can predict the focal firm’s future competitive actions. To be automated, it is not feasible for our system to rely on questionnaire items as its predictors. We thus decide to exploit news documents, which are large in size. In addition, our prediction system also faces a class imbalance problem. Thus, we develop an ensemble approach to address this problem. Besides, we conduct several experiments to evaluate the effectiveness of our proposed system. Our experimental results show that our proposed prediction system can predict competitive actions on the basis of data extracted from news documents, and it will get better result when using frequency as the representation scheme. Finally, we find out that different competitive action types need to consider different types of predictors, so the proposed system should apply different feature sets for different types of competitive actions.

並列關鍵字

Competitive Actions Prediction ； Competitive Actions Mining ； Competitive Actions ； News Documents Applications ； Data Mining ； Text Mining

參考文獻

Bowman, D., & Gatignon, H. (1995). Determinants of competitor response time to a new product introduction. Journal of Marketing Research, 42-53.

Chen, M.-J., & MacMillan, I. C. (1992). Nonresponse and delayed response to competitive moves: The roles of competitor dependence and action irreversibility. Academy of Management Journal, 35(3), 539-570.

Chen, M.-J., Smith, K. G., & Grimm, C. M. (1992). Action characteristics as predictors of competitive responses. Management Science, 38(3), 439-455.

Chi, L., Ravichandran, T., & Andrevski, G. (2010). Information technology, network structure, and competitive action. Information Systems Research, 21(3), 543-570.

Debruyne, M., Moenaertb, R., Griffinc, A., Hartd, S., Hultinke, E. J., & Robben, H. (2002). The impact of new product launch strategies on competitive reaction in industrial markets. Journal of Product Innovation Management, 19(2), 159-170.