Abstract

In the era of rapid growth and high competition, a company must possess an information/knowledge advantage in order to hold the upper hand in the industry. Therefore, the company has to continuously monitor its competitors in order to get enough information and convert the information into competitive knowledge. Although information technology has been used in many areas and has many successful examples, it is rarely the case that information technology was used for the task of competitor intelligence. Accordingly, this study devises a method of time series pattern extraction and processing for the task of obtaining place-prospect competitor intelligence in order to advance an enterprise with a competitive knowledge advantage. The purposes of the method are two-fold: (1) For a product manufactured by the company, the gathered data are mined into a knowledge advantage—an appropriate amount of the stock to be allocated at a timely fashion at a retailer in face of competition. (2) This knowledge advantage alerts the company's decision makers to what is unknown and forces them to make good decisions on the stock allocation problem, freeing them from the dilemma of over-stock or under-stock with respect to competitors’ stock. Our approach differs from traditional inventory management in the grounds they are based: traditional inventory management is based on the perspective of cash flow while our approach is based on the perspective of competition encountered. The results show our method is quite promising to this end, obtaining the intelligence to gain competitive advantage.

Author Keywords: Competitive intelligence; Time series pattern extraction

Article Outline

1. Introduction

2. Competitive information and the architecture of the CIMiner

3. Time series pattern extraction in the CIMiner

3.1. The general framework of time series pattern extraction

3.2. A Novel way of deploying the framework

3.3. The Algorithms

3.3.1. Algorithm 1

3.3.2. Algorithm 2

3.3.3. Algorithm 3

4. Time series pattern processing in the CIMiner

4.1. Correction models

4.2. An example

5. Evaluation

5.1. UnitLength

5.2. Threshold T

5.3. The experiment results

6. Conclusion

Appendix A

References