id: 06067595
dt: a
an: 06067595
au: Zhang, Lei; Zhang, Qi-ming; Wang, Yi-guo; Yu, Dong-lin
ti: Selecting an appropriate interestingness measure to evaluate the
    correlation between syndrome elements and symptoms.
so: Cao, Longbing (ed.) et al., New frontiers in applied data mining. PAKDD
    2011 international workshops, Shenzhen, China, May 24‒27, 2011.
    Revised selected papers. Berlin: Springer (ISBN 978-3-642-28319-2/pbk).
    Lecture Notes in Computer Science 7104. Lecture Notes in Artificial
    Intelligence, 372-383 (2012).
py: 2012
pu: Berlin: Springer
la: EN
cc: 
ut: interestingness measure; syndrome element; symptom; traditional Chinese
    medicine
ci: 
li: doi:10.1007/978-3-642-28320-8_32
ab: Summary: In order to select the best interestingness measure appropriate
    for evaluating the correlation between syndrome elements and symptoms,
    60 objective interestingness measures were selected from different
    subjects. Firstly, a hypothesis for a good measure was proposed. Based
    on the hypothesis, an experiment was designed to evaluate the measures.
    The experiment was based on the clinical record database of past
    dynasties including 51,186 clinical cases. The selected dataset in this
    study had 44,600 records. Han and Re were selected as the experimental
    syndrome elements. Three indicators calculated according to the
    distances between two syndrome elements were obtained in the experiment
    and were combined into one indicator. The Z score, $φ$-coefficient and
    Kappa were selected from 60 measures after the experiment. The Z score
    and $φ$-coefficient were selected according to subjective
    interestingness. Finally, the $φ$-coefficient was selected as the best
    measure for its low computational complexity. The method introduced in
    this paper may be used in other similar territories. Further research
    of traditional Chinese medicine can be made based on the conclusion
    made in this paper.
rv: