In view of the inadequacies of the monitoring and analysis methods of the "supply" and "demand" sides of drugs in the "New Generation Anti-drug Strategy 2.0", the application of an innovative idea of emerging technology to Taiwan's anti-drug strategy can not only complement the inadequacies of the previous statistical and analytical methods, but also trigger the introduction of international innovative ideas, interdisciplinary cooperation and diversified applications in the future anti-drug strategy. This study is an attempt to use new technologies to provide a basis for developing a new approach by conducting two subprojects. In the first subproject, a web-based crawler was used to collect information on the transactions and sales of ketamine "snorting equipment" on an online auction platform and to explore the drug market. The results of Subproject 1 are as follows: 1. It is feasible to monitor the drug market using online sales information, and the monitoring results can alert investigative units and help them take action in advance; 2. Online auction merchants value consumer feedback, and it is appropriate to require merchants to reinforce warnings when interacting with consumers to remind them that ketamine use is against the law and dangerous to their health; 3. The cheap and plain color of the highly sought-after ketamine snorting equipment implies the financial disadvantage of most young consumers who buy it; 4. The types and prices of ketamine snorting equipment are varied, and the price of popular styles is about 900-1500 NT; 5. The lower the price, the more options and the more reviews, the higher the sales of ketamine snorting equipment; 6. According to Narcotics Hazard Prevention Act, it is difficult to punish online auction merchants for selling ketamine snorting equipment, and it is appropriate to introduce corporate responsibility to encourage online shopping platforms to self-regulate.

　　The second subproject focuses on the "demand side" of the drug market. First, this study uses national population-based data to track recidivism rates, survival rates, and survival time in DPCCAT (Deferred Prosecution with Conditions to Complete Addiction Treatment (DPCCAT). In addition, this study uses multilevel Bayesian analysis to examine the association between drug use and violent crime. Finally, the study analysed comorbidity statistics for mental illness, drug use, and violent crime using official statistics from the Agency of Corrections. Analysis of recidivism and survival showed that 65.9% of drug users who participated in DPCCAT in 2015-2016 recidivated within 5 years. However, after further differentiating whether or not participants completed DPCCAT, we found that the recidivism rate within 5 years of completing DPCCAT was 39.7%. In addition, we found that the 5-year survival rate, recidivism rate, and average survival time of those who completed DPCCAT differed by type of drug use as follows: schedule I drug use alone (70%, 30%, 3.99 years), Schedule II drug use alone (71%, 29%, 3.89 years), combination of Schedule I and Schedule II drug use (45%, 55%, 2.9 years), and heterogeneous group (51%, 55%, 2.9 years). Kaplan-Meier and Cox analyses revealed a significantly higher risk of relapse in the group with schedule I and II drug use group and in the heterogeneous group compared with drug users alone.

　　Subproject 2's multilevel Bayesian analysis found that the probability of engaging in violent crime after drug use was relatively low compared to persistent drug users, with only a minority of 5-6% of drug users subsequently involved in violent crime. However, when a person's drug use was accompanied by a violent crime, there was a 31.4% probability of continuing down the path to violent crime. In addition, using correctional statistics, the study found that 2.99 to 4.48% of Schedule I and II drug users in correctional facilities with a psychosis diagnosis had been convicted of violent crimes. This study suggests that more emphasis should be placed on evidence-based recidivism assessment for deferred prosecution and addiction treatment. Integration of prosecution and medical assessment databases should be developed to promote scientific and evidence-based diversified treatment and triage for early and effective triage assessment of drug users. Diversified treatment should offer different treatment interventions to drug users at different stages. Treatment and psychiatric needs should be addressed for disadvantaged groups with drug dependence and mental disorders.

Keywords: web crawlers, innovative research, drug use equipment, Shopee, big data, data exploration, drug use, mental disorder, violent crime, drug-related violent crime

For the full paper, please visit 應用爬蟲程式與數據探勘毒品施用行為創新先導研究 (only in traditional Chinese version).