高等学校化学学报

高等学校化学学报 ›› 2000, Vol. 21 ›› Issue (S1): 42.

• Analytical Sciences • 上一篇    下一篇

A Validated Approach to Discovering Implication Knowledge from Empirical Data with Applications in Chemometrics

LIU Ji-Ming1, Frank S. C. LEE2, WANG Xiao-Ru3, YING Hai4   

  1. 1. Deaparment of Computer Science;
    2. Department of Chemistry, Hong Kong Baptist University;
    3. The Key Laboratory of Analytical Sciences of MOE and Department of Chemistry, Xiamen University, Xiamen 361005, China;
    4. McGill University
  • 出版日期:2000-12-31 发布日期:2000-12-31

A Validated Approach to Discovering Implication Knowledge from Empirical Data with Applications in Chemometrics

LIU Ji-Ming1, Frank S. C. LEE2, WANG Xiao-Ru3, YING Hai4   

  1. 1. Deaparment of Computer Science;
    2. Department of Chemistry, Hong Kong Baptist University;
    3. The Key Laboratory of Analytical Sciences of MOE and Department of Chemistry, Xiamen University, Xiamen 361005, China;
    4. McGill University
  • Online:2000-12-31 Published:2000-12-31

PDF (PC)

被引次数

6

摘要：

Implication knowledge is useful for drawing conclusions or confirming hypotheses based on some observed data. The challenge here is where/how to derive such knowledge. Traditional knowledge acquisition methodologies or tools have limited power in that they rely on human knowledge engineers' judgements and are applicable only under controlled situations. It would be desirable as well as imperative if we can demonstrate any automatic means of mining such implication knowledge directly from empirical data-this is because in practical applications,data samples are usually available and most importantly the automatically induced implications are less subjective and mathematically sound. As a response to this challenge, our work has developed and validated a new method of automatically discovering implication knowledge by statistically inducing the implication relationships among the data attributes in empirically-obtained sample data.

Abstract:

Implication knowledge is useful for drawing conclusions or confirming hypotheses based on some observed data. The challenge here is where/how to derive such knowledge. Traditional knowledge acquisition methodologies or tools have limited power in that they rely on human knowledge engineers' judgements and are applicable only under controlled situations. It would be desirable as well as imperative if we can demonstrate any automatic means of mining such implication knowledge directly from empirical data-this is because in practical applications,data samples are usually available and most importantly the automatically induced implications are less subjective and mathematically sound. As a response to this challenge, our work has developed and validated a new method of automatically discovering implication knowledge by statistically inducing the implication relationships among the data attributes in empirically-obtained sample data.

TrendMD: