곡곡λΆλ¬Έμ ν¨μ¨μ±μ μ¦μ§ νΉμ μ
무 νμ μ μ€νν μ μλ μμ΄λμ΄ κ°λ°
μ λͺ©νλ‘ νμ¬ μ°Έκ°μλ€μ μ£Όμ΄μ§ λ°μ΄ν°μ λꡬλ₯Ό μ¬μ©νμ¬ 20μκ° λμ μμ΄λμ΄λ₯Ό ꡬννκ² λ©λλ€.
The goal is to develop ideas that enhance efficiency or enable innovative practices in the public sector. Participants will use the provided data and tools to implement their ideas over a period of 20 hours.
μ
μ±λ―Όμ μνΈμ
νλ‘μ νΈλ 곡무μλ€μ΄ μ¨λΌμΈ μμμ κ²ͺλ μ
μ± λ―Όμ
μΌλ‘ μΈν κ°μ λ
Έλ
λ¬Έμ λ₯Ό ν΄κ²°νκΈ° μν΄ AI
κΈ°λ° μλΉμ€λ₯Ό κ°λ°νμμ΅λλ€. μ΄ μλΉμ€λ Langchain νλ μμν¬
λ₯Ό νμ©ν LLM
κΈ°μ μ νμ©νμ¬, μ¨λΌμΈμμ λ°μνλ μ
μ± λ―Όμμ μλμΌλ‘ κ°μ§νκ³ νν°λ§νμ¬ κ³΅λ¬΄μμ μ
무 λΆλ΄μ μ€μ΄κ³ μ ν©λλ€. μ΄ μλΉμ€λ κΈ°μ‘΄μ μλμ μΈ λ°©μ λμ , κΈ°μ‘΄ μμ€ν
μ μ½κ² ν΅ν©ν μ μλ κ°νΈνκ³ ν¨μ¨μ μΈ λκ΅¬λ‘ μ€κ³λμμ΅λλ€.
μ£Όμ κΈ°λ₯:
μλ κ°μ§ λ° νν°λ§
:AI
κ° μ¨λΌμΈ μ μ± λ―Όμμ μλμΌλ‘ κ°μ§νμ¬ νν°λ§ν©λλ€.κ°νΈν ν΅ν©
: κΈ°μ‘΄ μμ€ν μ μ½κ² ν΅ν©ν μ μλλ‘ μ€κ³λμμ΅λλ€.λΉμμ΄ μν
: λ―Όμ λ΄μ© μ€ λΉμμ΄λ₯Ό μννμ¬ λμ± μΎμ ν μ 무 νκ²½μ μ‘°μ±ν©λλ€.
Malicious Complaint Guardian
project has developed anAI
-based service to address theemotional labor
issues that public officials face frommalicious complaints
online. This service utilizes theLangchain framework
andLLM
technology to automatically detect and filter malicious online complaints, aiming to reduce the workload of public officials. It is designed as a convenient and efficient tool that can be easily integrated into existing systems, replacing traditional manual processes.
Key Features:
Automatic Detection and Filtering
:AI
automatically detects and filters online malicious complaints.Easy Integration
: Designed for easy integration into existing systems.Profanity Moderation
: Moderates profanity in complaints to create a more pleasant working environment.
λ°λͺ¨ μμ ( Demostration Video ) |
---|
λ°νμλ£ ( PDF ) |
- νκ΅μ΄ λΉμμ΄ λ°μ΄ν°κ° λ§€μ° λΆμ‘±ν΄μ νμ΅μ μΆ©λΆν λͺ»νμλ€.
- KISOμμ μΆμν
KSS
APIλ₯Ό μ°λν΄μ μ νμ±μ λμ΄κ³ μΆμμ§λ§ λ²μΈμλ κΈ°μ λ§ νμ©ν΄μ£Όμ΄μ μ°λνμ§ λͺ»νλ€. - Microsoftμμ μ 곡νλ OPENAI ν ν°μ΄ λ§μ΄ λΆμ‘±ν΄μ Personal OEPNAI ν ν° μ¬μ©μ΄ λΆκ°νΌνμλ€.
- μκ°μ΄ λΆμ‘±ν΄μ μ΅μ ν μ§νμ λ§μ΄ λͺ»νλ€.
- The lack of sufficient Korean profanity data hindered thorough training.
- We wanted to enhance accuracy by integrating the
KSS
API released by KISO, but it was only available to registered corporations, so we couldn't integrate it.- The OPENAI tokens provided by Microsoft were insufficient, making the use of Personal OPENAI tokens inevitable.
- Due to time constraints, we were unable to carry out extensive optimization.
Frontend | Frontend | AI |
---|---|---|
μ΄μ λ―Ό | κ³ μ¬λ―Ό | κΉνκ· |