• 대한수사과학회지
• /
• 제1권2호
• /
• pp.5-19
• /
• 2006

데이터 마이닝은 컴퓨터와 정보처리의 발전으로 각기 다른 차원에서 다량으로 수집되는 데이터 속에서 숨은 의미나 패턴을 발견하는 유용한 기법이다. 의사결정나무, 신경망 모형, 규칙 귀납, K-평균 군집화, 시각화 등의 데이터 마이닝 개별 기법들은 산재해 있는 데이터에서 연관성을 분석하고, 이를 분류함으로써 일반화된 개념을 정의하고, 새로운 지식을 추론함으로써 실제 생활에 적용 가능한 예측을 가능하게 한다. 따라서 현재 데이터 마이닝은 기업의 마케팅 분야, 금융기관의 고객 분석, 통신 회사의 고객 이탈 방지 등에서 유용하게 활용되고 있다. 우리가 접해야 하는 정보의 양이 늘어나는 것은 범죄 수사에 있어서도 마찬가지 현상이다. 범죄와 범죄자에 대한 데이터는 축적되어 가지만 정작 개별 사안에 있어서는 중요한 데이터가 접근조차 되지 않고 있으며, 많은 데이터 속에서 이것이 내포하고 있는 숨은 의미를 지나치게 되는 경우도 많다. 본 연구에서는 선행 연구와 사례 적용을 통해 데이터 마이닝의 범죄 수사 적용 가능성과 한계점을 살펴보고자 하였다. 미제 사건으로 남는 경우가 많은 절도나 사기 같은 습관적 상습 범죄의 경우 데이터 마이닝의 분류, 군집화 기능을 활용 한다면 향후 여죄 추적에 효율적으로 활용될 수 있음을 파악할 수 있었고, 특히 다양한 문제에 적용 가능하고, 잡음에 대한 견고성이 있음에도 예측의 정확성을 지니고 있는 신경망 모형의 경우 패턴 인식을 통하여 범죄자 프로파일링이나 화상 자료 대비 시스템 구축에 충분히 활용될 것으로 생각한다. 특히 보험 사기 사례 적용에서 살펴본 바와 같이 마약, 테러와 같은 조직적 범죄수사나 자금세탁과 같은 금융 추적 수사의 경우 해당 자료의 방대함과 모호성으로 인해 수사를 하는 데 많은 어려움이 있지만 이러한 데이터 마이닝 가시화 기법을 적절히 활용한다면 전체적인 윤곽을 파악하는 데 매우 유용하며, 효율적인 수사가 가능함을 확인할 수 있었다. 그러나 데이터 마이닝은 예측 모델이므로 오류를 내재하고 있다는 점에서 수사 기관의 데이터 마이닝 접근은 조심스러워야 하며, 정보 독점화 현상과 개인 사생활 보호라는 측면에서 각 수사기관은 해당 법률에 정한 범위 내에서 해당 사건별로 데이터를 수집하고 이를 통합, 재구성하여 활용하는 측면으로 적용되어야 할 것이다. 또한 각 수사기관별로는 자신의 보유하고 있는 데이터에 대해 다차원 처리가 가능하도록 데이터베이스 시스템을 구축하여 데이터 마이닝이 적용 가능한 환경을 구축하도록 하여야 할 것이다. 아직은 논의의 초기 단계이므로 효과가 크게 부각되지는 않았지만 지금까지 제시한 문제에 대한 연구가 계속 이루어진다면 인권중심, 증거중심의 수사 개념을 바탕으로 적법절차에 의한 수사 활동을 요구받는 시대에 새로운 대안으로 자리 잡을 것이며, 수사의 과학화에 기여할 것으로 전망한다.

• 여성건강간호학회지
• /
• 제2권2호
• /
• pp.298-331
• /
• 1996

Recent improvements in living standard and development in medical care led to an increased interest in life expectancy and personal health, and also led to a more demand for higher quality of life. Thus, the problem of women's health draw a fresh interest nowadays. Since late-middle aged women experience various physical and socio-psychological changes and tend to have chronic illnesses, these women have to take initiatives for their health control by realizing their own responsibility. The basic elements for a healthy life of these women are understanding of their physical and psychological changes and acceptance of these changes. Health promoting behaviors of an individual or a group are actions toward increasing the level of well-being and self-actualization, and are affected by various variables. In Pender's health promoting model, variables are categorized into cognitive factors(individual perceptions), modifying factors, and variables affecting the likelihood for actions, and the model assumes the health promoting behaviors are affected by cognitive factors which are again affected by demographic factors. Since Pender's model was proposed based on a tool broad conceptual frame, many studies done afterwards have included only a limited number of variables of Pender's model. Furthermore, Pender's model did not precisely explain the possibilities of direct and indirect paths effects. The objectives of this study are to evaluate Pender's model and thus propose a model that explains health promoting behaviors among late-middle aged women in order to facilitate nursing intervention for this group of population. The hypothetical model was developed based on the Pender's health promoting model and the findings from past studies on women's health. Data were collected by self-reported questionnaires from 417 women living in Seoul, between July and November 1994. Questionnaires were developed based on instruments of Walker and others' health promotion lifestyle profile, Wallston and others' multidimensional health locus of control, Maoz's menopausal symptom check list and Speake and others' health self-rating scale. IN addition, items measuring self-efficacy were made by the present author based on past studies. In a pretest, the questionnaire items were reliable with Cronbach's alpha ranging from .786 to .934. The models for health promoting behaviors were tested by using structural equation modelling technique with LISREL 7.20. The results were summarized as follows : 1. The overall fit of the hypothetical model to the data was good (chi-square=4.42, df=5, p=.490, GFI=.995, AGFI=.962, RMSR=.024). 2. Paths of the model were modified by considering both its theoretical implication and statistical significance of the parameter estimates. Compared to the hypothetical model, the revised model has become parsimonious and had a better fit to the data (chi-square =4.55, df=6, p=.602, GFI=.995, AGFI=.967, RMSR=.024). 3. The results of statistical testing were as follows : 1) Family function internal health locus of control, self-efficacy, and education level exerted significant effects on health promoting behaviors(${\gamma}_{43}$=.272, T=3.714; ${\beta}_[41}$=.211, T=2.797; ${\beta}_{42}$=.199, T=2.717; ${\gamma}_{41}$=.136, T=1.986). The effect of economic status, physical menopausal symptoms, and perceived health status on health promoting behavior were insignificant(${\gamma}_{42}$=.095, T=1.456; ${\gamma}_{44}$=.101, T=1.143; ${\gamma}_{43}$=.082, T=.967). 2) Family function had a significance direct effect on internal health locus of control (${\gamma}_{13}$=.307, T=3.784). The direct effect of education level on internal health locus of control was insignificant(${\gamma}_{11}$=-.006, T=-.081). 3) The directs effects of family functions & internal health locus of control on self-efficacy were significant(${\gamma}_{23}$=.208, T=2.607; ${\beta}_{21}$=.191, T=2.2693). But education level and economic status did not exert a significant effect on self-efficacy(${\gamma}_{21}$=.137, T=1.814; ${\beta}_{22}$=.137, T=1.814; ${\gamma}_{22}$=.112, T=1.499). 4) Education level had a direct and positive effect on perceived health status, but physical menopausal symptoms had a negative effect on perceived health status and these effects were all significant(${\gamma}_{31}$=.171, T=2.496; ${\gamma}_{34}$=.524, T=-7.120). Internal health locus and self-efficacy had an insignificant direct effect on perceived health status(${\beta}_{31}$=.028, T=.363; ${\beta}_{32}$=.041, T=.557). 5) All predictive variables of health promoting behaviors explained 51.8% of the total variance in the model. The above findings show that health promoting behaviors are explained by personal, environmental and perceptual factors : family function, internal health locus of control, self-efficacy, and education level had stronger effects on health promoting behaviors than predictors in the model. A significant effect of family function on health promoting behaviors reflects an important role of the Korean late-middle aged women in family relationships. Therefore, health professionals first need to have a proper evaluation of family function in order to reflect the family function style into nursing interventions and development of strategies. These interventions and strategies will enhance internal health locus of control and self-efficacy for promoting health behaviors. Possible strategies include management of health promoting programs, use of a health information booklets, and individual health counseling, which will enhance internal health locus of control and self-efficacy of the late-middle aged women by making them aware of health responsibilities and value for oneself. In this study, an insignificant effect of physical menopausal symptoms and perceived health status on health promoting behaviors implies that they are not motive factors for health promoting behaviors. Further analytic researches are required to clarify the influence of physical menopausal symptoms and perceived health status on health promoting behaviors with-middle aged women.