🍺 On a hot summer’s day, many people would like to have a cold beer, but drink driving often leads to regrets. Based on statistics from the National Police Agency, in the last five years, there were more than 50,000 cases of “prosecution due to drink driving” in a year on average.
🚔 Drink driving is one of the social issues that many people are concerned about. Recently, the government has amended regulations and increased penalties because of this. However, whenever the outcome of a court decision for a major drink driving case is released, doubts continue to be raised among public opinion: why “only a light penalty”?
💁 This thorny issue has seen new breakthroughs recently. During the Presidential Hackathon this year, the Judicial Yuan team comprising legal and data experts proposed two ideas of “real-time reference” and the “sentencing reference aid.”
🔧 In the last two years, the Judician Yuan has successively developed several tools, the “dictionary of difficult terms” and “dictionary of sentencing terms” to allow the public to look up professional legal terms and search for court verdicts through the “court verdict query system.” However, it is not possible to cross-reference using these tools, which instead led to difficulties for users.
🔗 The Judicial Yuan team was inspired by the design of the Wikipedia interface and connected these tools to develop the “real-time reference” browser plug-in. When looking up court verdicts in the future, all you need to do is hover the mouse cursor over the term to see the corresponding explanation, allowing people who may be interested to quickly understand the contents of the court verdict.
🖥 The “sentencing information aid,” on the other hand, applies machine learning. Taking the “Offense of Unsafe to Drive” caused by drink driving as an example, they would first locate about 50,000 related court verdicts from the existing court verdict database, sort the factors involved in these offenses, and set up these factors on the sentencing information system.
📲 Anyone who is curious about the verdict only has to input different factors such as “breath alcohol content” or “vehicle type” to not only find the scope of proposed penalties, but also see the verdicts of similar cases. This way, everyone will be able to see that the same offense may result in very different sentencing outcomes, even if they only differ in one of the factors.
⚖ The Judicial Yuan team was eventually selected as one of the five outstanding teams of the Presidential Hackathon. Through them, we have seen not only the immense potential of “machine learning,” but also the innovative power brought about by “joint learning.” We look forward to future achievements of the team in resolving more doubts raised in society and putting into practice the cause of “judicial reforms that give people a sense of change” so emphasized by President Tsai.
🙏 Lastly, I am especially grateful to our colleagues from the Board of Science and Technology, National Development Council, Ministry of Foreign Affairs, and Institute for Information Industry for their assistance. It is thanks to their hard work over the last few months that the Presidential Hackathon was completed successfully.