kakao-analyzer

Name	kakao-analyzer JSON
Version	1.0.2 JSON
	download
home_page	https://github.com/xistoh162108/kakaotalk_analyzer
Summary	Comprehensive analysis tool for KakaoTalk chat exports
upload_time	2025-08-14 08:53:30
maintainer	None
docs_url	None
author	xistoh162108, Claude AI
requires_python	>=3.8
license	MIT
keywords	kakao kakaotalk chat analysis nlp korean messaging
VCS
bugtrack_url
requirements	pandas numpy matplotlib seaborn wordcloud scikit-learn sentence-transformers torch transformers chardet tqdm requests
Travis-CI	No Travis.
coveralls test coverage	No coveralls.

            # 🔥 카카오톡 대화 분석기 (Kakao Talk Analyzer)

[![PyPI version](https://badge.fury.io/py/kakao-analyzer.svg)](https://badge.fury.io/py/kakao-analyzer)
[![Python versions](https://img.shields.io/pypi/pyversions/kakao-analyzer.svg)](https://pypi.org/project/kakao-analyzer/)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Downloads](https://pepy.tech/badge/kakao-analyzer)](https://pepy.tech/project/kakao-analyzer)

> 카카오톡 대화 내보내기 CSV 파일을 종합적으로 분석하는 **차세대 AI 기반** 올인원 도구

```bash
pip install kakao-analyzer
```

**🎯 카카오톡의 특성을 완벽하게 이해하는 스마트 분석기**
- 연속 메시지 자동 그룹핑 (1분 이내 메시지 통합)
- 한국어 채팅 패턴 인식 (ㅋㅋ, ㅠㅠ, 이모티콘 등)
- 시스템 메시지 자동 필터링 (삭제된 메시지, 투표, 정산 등)
- 실제 대화만 분석하여 더 정확한 인사이트 제공

## 🚀 주요 기능

### 🎯 **카카오톡 특화 기능** ⭐ NEW!
- **스마트 메시지 그룹핑**: 1분 이내 연속 메시지를 하나로 통합하여 실제 의도 파악
- **한국어 채팅 패턴 분석**: ㅋㅋ, ㅎㅎ, ㅠㅠ, ㅇㅋ 등 한국인의 채팅 습관 인식
- **시스템 메시지 필터링**: 삭제된 메시지, 투표, 정산, 입장/퇴장 알림 자동 제거
- **멘션 분석**: @ 기능을 이용한 사용자 언급 패턴과 네트워크 분석
- **감정 표현 분석**: 웃음, 슬픔, 동의 표현 등의 감정 패턴 추적

### 📊 기본 통계 분석
- **사용자별 활동 통계**: 메시지 수, 참여율, 일평균 활동량
- **시간적 패턴 분석**: 시간대별, 요일별, 월별 활동 분포  
- **단어 빈도 분석**: 한국어 형태소 분석 및 키워드 추출
- **메시지별 단어 수 통계**: 원본 및 필터링된 단어 수 분석 (평균, 중간값, 최대값)
- **이모티콘/이모지 사용 통계**: 반응 패턴 분석

### 💬 대화 역학 분석
- **대화 턴 분석**: 자동 대화 세션 구분
- **주제 전환 탐지**: 임베딩 기반 주제 변화 감지
- **참여자 패턴**: 대화 시작자, 응답자 분석 (시작 횟수 순으로 정렬)
- **상호작용 분석**: 사용자 간 소통 패턴
- **멘션 네트워크 분석**: @ 멘션을 통한 사용자 간 언급 관계 및 상호 멘션 분석

### 🎯 재미있는 지표
- **참여 불평등**: 지니 계수를 통한 발언권 분석
- **답장 속도**: 반응 시간 통계 및 분포
- **활동 연속성**: 연속 메시지 및 활동일 분석
- **심야 채팅**: 밤샘 대화 패턴 분석

### 🔍 고급 분석
- **임베딩 기반 주제 분석**: 의미론적 유사도를 통한 주제 세그먼트
- **SPLADE 하이브리드 검색**: Dense + Sparse 벡터를 활용한 대화 검색
- **시계열 분석**: 시간에 따른 대화 패턴 변화
- **감정 분석**: 대화 분위기와 감정 패턴 추적
- **스마트 토픽 클러스터링**: 고급 주제 분류 및 테마 분석
- **대화 리듬 분석**: 응답 패턴과 시간적 리듬 분석
- **대규모 데이터셋 최적화**: 지능형 청킹과 메모리 효율성

### 📈 시각화 (개선됨!)
- **활동 히트맵**: 시간대 × 요일 활동 분포
- **일별 타임라인**: 메시지 수 추이 그래프
- **주제 타임라인**: 주제 세그먼트 시각화 (겹침 문제 해결)
- **워드클라우드**: 한국어 폰트 지원으로 깨짐 현상 해결
- **사용자 활동 차트**: 많은 사용자도 깔끔하게 표시 (겹침 문제 해결)

## 🛠️ 설치 및 실행

### 📦 PyPI를 통한 간편 설치 (권장)

```bash
# 기본 설치
pip install kakao-analyzer

# 전체 기능 설치 (AI 모델 포함)
pip install kakao-analyzer[full]

# 개발 도구 포함 설치
pip install kakao-analyzer[dev]
```

설치 후 바로 사용 가능:
```bash
kakao-analyzer --input your_chat.csv
```

### 💻 개발자용 설치

개발이나 커스터마이징을 원한다면:

```bash
# 저장소 클론
git clone https://github.com/xistoh162108/kakaotalk_analyzer.git
cd kakaotalk_analyzer

# 의존성 설치
pip install -r requirements.txt

# 개발 모드로 설치
pip install -e .
```

### 📋 요구사항
- **Python**: 3.8 이상
- **운영체제**: Windows, macOS, Linux 모두 지원
- **메모리**: 최소 4GB RAM (대용량 데이터는 8GB+ 권장)

## 🚀 빠른 시작

### 1단계: 설치
```bash
pip install kakao-analyzer
```

### 2단계: 카카오톡 대화 내보내기
1. 카카오톡 앱에서 분석하고 싶은 대화방 열기
2. `☰` 메뉴 → `채팅방 설정` → `대화 내용 내보내기`
3. `텍스트 파일(.txt)` 선택하여 내보내기
4. 내보낸 파일을 CSV 형식으로 변환 (또는 직접 CSV로 준비)

### 3단계: 분석 실행
```bash
# 기본 분석
kakao-analyzer --input your_chat.csv

# 고급 AI 분석 포함 (권장)
kakao-analyzer --input your_chat.csv --use-splade --use-ollama
```

### ✨ 사용법
```bash
# 기본 분석 (가장 간단)
kakao-analyzer --input chat.csv

# 출력 디렉토리 지정
kakao-analyzer --input chat.csv --outdir my_analysis

# 도움말 보기
kakao-analyzer --help
```

### 🛠️ 고급 옵션
```bash
kakao-analyzer --input INPUT [옵션들]

필수 옵션:
  --input INPUT                입력 CSV 파일 경로

분석 옵션:
  --window-minutes MINUTES     대화 턴 구분 시간 (기본값: 30분)
  --topic-window-size SIZE     주제 분석 윈도우 크기 (기본값: 15)
  --similarity-threshold FLOAT 유사도 임계값 (기본값: 0.3)
  --batch-size SIZE            임베딩 배치 크기 (기본값: 32)
  --max-workers WORKERS        최대 병렬 작업 수 (기본값: 4)

AI 모델 옵션:
  --use-ollama                 Ollama 임베딩 사용 (로컬)
  --model-name MODEL           Ollama 모델명 (기본값: gpt-oss:20b)
  --embed-model MODEL          임베딩 모델명 (기본값: bge-m3)
  --use-splade                 SPLADE 희소 검색 활성화

대규모 데이터 옵션:
  --subset-size SIZE           테스트용 부분집합 크기  
  --subset-strategy STRATEGY   부분집합 전략 (balanced/temporal/diverse/random)

카카오톡 특화 옵션:
  --disable-message-grouping   연속 메시지 그룹핑 비활성화 (비추천)
  --group-window-seconds N     메시지 그룹핑 시간 창 (기본값: 60초)

출력 옵션:
  --outdir DIRECTORY           출력 디렉토리
  --figure-dpi DPI             그래프 해상도 (기본값: 300)
  --log-level LEVEL            로그 레벨 (DEBUG/INFO/WARNING/ERROR)
  --quiet                      콘솔 출력 억제
  --skip-viz                   시각화 생성 건너뛰기
```

### 💡 사용 예제
```bash
# 🌟 기본 분석 (추천)
kakao-analyzer --input chat.csv

# 🤖 AI 기능 전체 활용
kakao-analyzer --input chat.csv --use-splade --use-ollama --model-name "gpt-oss:20b"

# 📊 특정 디렉토리에 결과 저장
kakao-analyzer --input chat.csv --outdir my_results

# ⚡ 대용량 데이터 빠른 테스트
kakao-analyzer --input big_chat.csv --subset-size 1000 --subset-strategy balanced

# 🔧 세밀한 설정 조정
kakao-analyzer --input chat.csv --topic-window-size 20 --similarity-threshold 0.4

# 🎨 고해상도 그래프 생성
kakao-analyzer --input chat.csv --figure-dpi 600

# 🔇 조용한 실행 (CI/CD용)
kakao-analyzer --input chat.csv --quiet --skip-viz --log-level ERROR
```

## 📁 출력 구조

분석 완료 후 다음과 같은 구조로 결과가 생성됩니다:

```
output_directory/
├── summary/                    # 요약 정보
│   ├── analysis_summary.json   # 전체 분석 요약
│   └── processed_messages.csv  # 처리된 메시지 데이터
├── basics/                     # 기본 통계
│   └── basic_statistics.json   # 상세 통계 데이터
├── context/                    # 맥락 분석
│   ├── topic_segments.json     # 주제 세그먼트 데이터
│   ├── topic_segments.csv      # 주제 세그먼트 테이블
│   ├── mood_analysis.json      # 감정 및 분위기 분석
│   ├── advanced_topics.json    # 고급 주제 클러스터링
│   ├── rhythm_analysis.json    # 대화 리듬 분석
│   ├── mention_analysis.json   # 멘션 패턴 및 네트워크 분석
│   ├── hybrid_index.meta.json  # 검색 인덱스 (SPLADE)
│   └── splade_scores.parquet   # 희소 벡터 데이터 (SPLADE)
├── fun/                        # 재미 지표
│   └── fun_metrics.json        # 참여 불평등, 답장 속도 등
├── figures/                    # 시각화
│   ├── heatmap_hour_weekday.png
│   ├── timeseries_daily.png
│   ├── topic_timeline.png
│   ├── wordcloud_global.png
│   ├── user_activity_distribution.png
│   └── reply_latency_distribution.png
├── reports/                    # 리포트
│   ├── analysis_report.md      # 종합 분석 리포트
│   └── qa_test_report.md       # 품질 보증 리포트 (테스트시)
└── logs/                       # 로그
    ├── pipeline_metadata.json  # 실행 메타데이터
    └── analyzer.log            # 분석 로그
```

## 📋 입력 데이터 형식

카카오톡에서 내보낸 CSV 파일이 필요합니다. 지원하는 컬럼 형식:

### 필수 컬럼
- `user` 또는 `User` 또는 `사용자`: 메시지 작성자
- `message` 또는 `Message` 또는 `메시지`: 메시지 내용  
- `datetime` 또는 `Date` 또는 `시간`: 메시지 시간

### 예제 CSV 형식
```csv
user,message,datetime
김민수,안녕하세요!,2024-01-01 10:00:00
이지은,네 안녕하세요! ㅎㅎ,2024-01-01 10:01:00
박서준,오늘 날씨 좋네요,2024-01-01 10:02:00
```

## 🧪 테스트

### 기본 기능 테스트
```bash
# 기본 기능 테스트 (의존성 없이)
python simple_test.py

# 전체 테스트 스위트 (의존성 설치 후)
python test_runner.py
```

### 테스트 데이터 생성
```bash
# 1000개 메시지 테스트 데이터 생성
python test_runner.py --generate-data
```

## 🔧 고급 설정

### 🚀 GPU 가속 지원 (NEW!)
- **자동 GPU 인식**: CUDA가 설치된 시스템에서 자동으로 GPU 사용
- **메모리 최적화**: GPU 메모리의 75%까지 자동 제한하여 안정성 보장
- **모델 자동 이관**: 임베딩 및 SPLADE 모델을 GPU로 자동 로드
- **배치 크기 최적화**: GPU 메모리에 따라 배치 크기 자동 조정

#### GPU 메모리별 최적화
- **12GB+ (RTX 3080Ti/4080 이상)**: 배치 크기 128, Half precision
- **8-12GB (RTX 3070/4070)**: 배치 크기 64, Half precision  
- **4-8GB (GTX 1660/RTX 3060)**: 배치 크기 48, Mixed precision
- **4GB 미만**: CPU 모드로 자동 전환

### 임베딩 모델
- **기본값**: Mock 임베딩 (외부 의존성 없음)
- **고급 옵션**: SentenceTransformer 모델 (`sentence-transformers` 설치 필요)
- **GPU 가속**: CUDA 지원 시 자동으로 GPU에서 실행

### SPLADE 하이브리드 검색
- **활성화**: `--use-splade` 옵션 사용
- **기능**: Dense + Sparse 벡터를 결합한 고급 검색
- **용도**: 특정 주제나 키워드로 대화 검색
- **GPU 가속**: Half precision으로 GPU에서 고속 처리

### Ollama 연동 (한국어 모델 완전 지원)
- **활성화**: `--use-ollama` 옵션 사용
- **요구사항**: Ollama 서버 실행 중이어야 함
- **기본 모델**: gpt-oss:20b (추천)

#### 🇰🇷 지원 한국어 모델들
| 모델명 | 크기 | 특징 | 사용법 |
|--------|------|------|--------|
| `gpt-oss:20b` | 20B | 범용 한국어 모델 (기본값) | `--model-name "gpt-oss:20b"` |
| `gpt-oss:40b` | 40B | 고성능 한국어 모델 | `--model-name "gpt-oss:40b"` |
| `llama3-ko` | 8B | LLaMA3 한국어 파인튜닝 | `--model-name "llama3-ko"` |
| `aya:8b` | 8B | 다국어 지원 (한국어 포함) | `--model-name "aya:8b"` |
| `gemma2-ko` | 9B | Google Gemma2 한국어 버전 | `--model-name "gemma2-ko"` |
| `solar-ko` | 10.7B | Upstage SOLAR 한국어 모델 | `--model-name "solar-ko"` |
| `eeve-korean` | 10.8B | 한국 전용 대화 모델 | `--model-name "eeve-korean"` |
| `beomi-ko` | 13B | 한국어 특화 고성능 모델 | `--model-name "beomi-ko"` |

#### 🚀 모델 설치 및 사용법
```bash
# Ollama 설치 (한 번만)
curl -fsSL https://ollama.ai/install.sh | sh

# 한국어 모델 다운로드 (예: gpt-oss:20b)
ollama pull gpt-oss:20b

# 분석 실행
kakao-analyzer --input chat.csv --use-ollama --model-name "gpt-oss:20b"
```

## 📊 분석 결과 해석

### 참여 불평등 (지니 계수)
- **0.0-0.3**: 매우 균등한 참여
- **0.3-0.5**: 약간 불균등
- **0.5-0.7**: 불균등한 참여
- **0.7-1.0**: 매우 불균등 (소수가 대화 주도)

### 멘션 분석 지표
- **상호 멘션 관계**: 서로 멘션하는 사용자 쌍의 수와 빈도
- **멘션 네트워크 밀도**: 전체 가능한 멘션 관계 대비 실제 멘션 관계 비율
- **멘션 목적 분류**: 질문, 부탁/요청, 인사, 긴급, 토론/의견 등으로 자동 분류

### 답장 속도 카테고리
- **즉석 답장**: 1분 이내
- **빠른 답장**: 1-5분
- **보통 답장**: 5-30분
- **느린 답장**: 30분 이상

### 주제 세그먼트
- 임베딩 유사도 기반으로 자동 감지
- 각 세그먼트별 키워드, 참여자, 지속시간 제공
- 주제 전환 패턴 분석

## 🎨 시각화 가이드

모든 그래프는 한국어 폰트를 지원하며, 다음 형식으로 저장됩니다:
- **형식**: PNG (기본값)
- **해상도**: 300 DPI (기본값)
- **크기**: 자동 조정

### 주요 시각화
1. **활동 히트맵**: 언제 가장 활발한지 한눈에 파악
2. **일별 타임라인**: 시간에 따른 대화량 변화
3. **주제 타임라인**: 대화 주제의 시간적 흐름
4. **워드클라우드**: 자주 사용된 단어들
5. **사용자 활동**: 각 참여자의 기여도
6. **답장 속도**: 소통 스타일 분석

## 💡 활용 사례

### 개인/그룹 분석
- 친구들과의 대화 패턴 분석
- 가족 단톡방 활동 분석
- 동호회/커뮤니티 소통 패턴

### 연구 목적
- 디지털 커뮤니케이션 연구
- 사회 네트워크 분석
- 언어학 연구 데이터

### 비즈니스 분석
- 팀 커뮤니케이션 효율성
- 고객 소통 패턴
- 커뮤니티 관리 인사이트

## 🛡️ 개인정보 보호

- 모든 분석은 로컬에서 수행됩니다
- 데이터는 외부로 전송되지 않습니다
- 결과 파일에는 원본 메시지가 포함될 수 있으니 공유 시 주의하세요
- 개인 식별 정보가 포함된 분석 결과는 신중히 관리하세요

## 🐛 문제 해결

### 일반적인 문제

**1. 모듈을 찾을 수 없음**
```bash
pip install -r requirements.txt
```

**2. 한국어 폰트 문제**
- macOS: Apple Gothic 자동 사용
- Windows: Malgun Gothic 자동 사용
- Linux: Noto Sans CJK KR 설치 권장

**3. 메모리 부족**
- 큰 데이터셋의 경우 `--topic-window-size` 값을 줄여보세요
- 임베딩 비활성화: 설정에서 `use_embeddings = False`

**4. CSV 인코딩 문제**
- 자동 인코딩 감지 기능 내장
- 수동 지정이 필요한 경우 파일을 UTF-8로 변환

### 로그 확인
```bash
# 상세 로그로 실행
kakao-analyzer --input chat.csv --log-level DEBUG

# 로그 파일 확인 (기본 출력 디렉토리)
cat chat_analysis/logs/run.log
```

## 📈 성능 최적화

### 대용량 데이터 처리
- **배치 크기 조정**: 설정에서 `batch_size` 변경
- **임베딩 비활성화**: 기본 분석만 필요한 경우
- **윈도우 크기 감소**: 메모리 사용량 절약

### 속도 향상
- **병렬 처리**: 가능한 곳에서 자동 적용
- **캐싱**: 중간 결과 자동 저장
- **최적화된 알고리즘**: 대용량 데이터 대응

## 🚧 로드맵

현재 이 프로젝트는 지속적으로 개발 중이며, 다음과 같은 기능들이 추가될 예정입니다:

### 🔮 예정된 기능들
- **🌐 다중 메신저 지원**: 라인, 텔레그램, 디스코드 등 다른 메신저 플랫폼 지원
- **📱 웹 인터페이스**: 브라우저에서 사용할 수 있는 직관적인 UI
- **⚡ 실시간 분석**: 대화 진행 중 실시간 패턴 분석
- **🤖 고급 AI 기능**: GPT-4 통합, 더 정교한 감정 분석
- **📊 대시보드**: 실시간 모니터링 및 알림 시스템
- **🔍 검색 엔진**: 대화 내용 고급 검색 및 필터링
- **📈 트렌드 분석**: 시간에 따른 대화 패턴 변화 추적
- **🎨 테마 커스터마이제이션**: 시각화 스타일 및 색상 테마 선택
- **💾 클라우드 백업**: 분석 결과 자동 백업 및 동기화
- **📧 리포트 자동 발송**: 정기적인 분석 리포트 이메일 발송

### 🎯 다음 버전 (v1.1.0) 예정 기능
- 음성 메시지 길이 분석
- 파일 공유 패턴 분석
- 대화방 참여자 변화 추적
- 감정 변화 시계열 분석
- 키워드 알림 시스템

## 👨‍💻 개발자

### 🎓 주 개발자

- **xistoh162108** (KAIST)
  - 📧 이메일: xistoh162108@kaist.ac.kr
  - 🏫 소속: 한국과학기술원 (KAIST)
  - 🔗 GitHub: [@xistoh162108](https://github.com/xistoh162108)

### 🤖 AI 협력 개발

- **Claude AI** (Anthropic)
  - 코드 아키텍처 설계 및 구현
  - 알고리즘 최적화 및 한국어 특화 기능
  - 문서화 및 사용자 경험 개선

이 프로젝트는 **인간-AI 협업**의 결과물로, 학술 연구와 실용적 도구 개발의 만남을 보여줍니다.

## 🤝 기여하기

### 🔗 협업 방식

- **이슈 리포팅**: [GitHub Issues](https://github.com/xistoh162108/kakaotalk_analyzer/issues)
- **기능 제안**: Pull Request 환영
- **학술 협력**: 연구 목적 사용 시 인용 부탁드립니다

### 💡 개선 아이디어

- 새로운 분석 지표 추가
- 다른 메신저 지원 (라인, 텔레그램 등)
- 웹 인터페이스 개발
- 실시간 분석 기능

### 📝 인용 (Citation)

학술 논문이나 연구에서 사용하실 때:

```bibtex
xistoh162108, & Claude AI. (2025). Kakao Talk Analyzer: 
Comprehensive analysis tool for KakaoTalk chat exports (v1.0.1). 
https://github.com/xistoh162108/kakaotalk_analyzer
```

## 📄 라이선스

이 프로젝트는 MIT 라이선스 하에 배포됩니다. 자세한 내용은 [LICENSE](LICENSE) 파일을 참고하세요.

## 🙏 감사의 말

- **한국어 자연어 처리**: KoNLPy, mecab-python3
- **시각화**: matplotlib, seaborn, wordcloud
- **데이터 처리**: pandas, numpy
- **머신러닝**: scikit-learn, sentence-transformers

---

**🤖 Generated with [Claude Code](https://claude.ai/code)**

Co-Authored-By: Claude <noreply@anthropic.com>

            

Raw data

            {
    "_id": null,
    "home_page": "https://github.com/xistoh162108/kakaotalk_analyzer",
    "name": "kakao-analyzer",
    "maintainer": null,
    "docs_url": null,
    "requires_python": ">=3.8",
    "maintainer_email": null,
    "keywords": "kakao kakaotalk chat analysis nlp korean messaging",
    "author": "xistoh162108, Claude AI",
    "author_email": "xistoh162108@kaist.ac.kr",
    "download_url": "https://files.pythonhosted.org/packages/84/6d/863a8d2c93da2cb710da2d1432b4e0bd13937e6c97bccda867f49192f532/kakao_analyzer-1.0.2.tar.gz",
    "platform": null,
    "description": "# \ud83d\udd25 \uce74\uce74\uc624\ud1a1 \ub300\ud654 \ubd84\uc11d\uae30 (Kakao Talk Analyzer)\n\n[![PyPI version](https://badge.fury.io/py/kakao-analyzer.svg)](https://badge.fury.io/py/kakao-analyzer)\n[![Python versions](https://img.shields.io/pypi/pyversions/kakao-analyzer.svg)](https://pypi.org/project/kakao-analyzer/)\n[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)\n[![Downloads](https://pepy.tech/badge/kakao-analyzer)](https://pepy.tech/project/kakao-analyzer)\n\n> \uce74\uce74\uc624\ud1a1 \ub300\ud654 \ub0b4\ubcf4\ub0b4\uae30 CSV \ud30c\uc77c\uc744 \uc885\ud569\uc801\uc73c\ub85c \ubd84\uc11d\ud558\ub294 **\ucc28\uc138\ub300 AI \uae30\ubc18** \uc62c\uc778\uc6d0 \ub3c4\uad6c\n\n```bash\npip install kakao-analyzer\n```\n\n**\ud83c\udfaf \uce74\uce74\uc624\ud1a1\uc758 \ud2b9\uc131\uc744 \uc644\ubcbd\ud558\uac8c \uc774\ud574\ud558\ub294 \uc2a4\ub9c8\ud2b8 \ubd84\uc11d\uae30**\n- \uc5f0\uc18d \uba54\uc2dc\uc9c0 \uc790\ub3d9 \uadf8\ub8f9\ud551 (1\ubd84 \uc774\ub0b4 \uba54\uc2dc\uc9c0 \ud1b5\ud569)\n- \ud55c\uad6d\uc5b4 \ucc44\ud305 \ud328\ud134 \uc778\uc2dd (\u314b\u314b, \u3160\u3160, \uc774\ubaa8\ud2f0\ucf58 \ub4f1)\n- \uc2dc\uc2a4\ud15c \uba54\uc2dc\uc9c0 \uc790\ub3d9 \ud544\ud130\ub9c1 (\uc0ad\uc81c\ub41c \uba54\uc2dc\uc9c0, \ud22c\ud45c, \uc815\uc0b0 \ub4f1)\n- \uc2e4\uc81c \ub300\ud654\ub9cc \ubd84\uc11d\ud558\uc5ec \ub354 \uc815\ud655\ud55c \uc778\uc0ac\uc774\ud2b8 \uc81c\uacf5\n\n## \ud83d\ude80 \uc8fc\uc694 \uae30\ub2a5\n\n### \ud83c\udfaf **\uce74\uce74\uc624\ud1a1 \ud2b9\ud654 \uae30\ub2a5** \u2b50 NEW!\n- **\uc2a4\ub9c8\ud2b8 \uba54\uc2dc\uc9c0 \uadf8\ub8f9\ud551**: 1\ubd84 \uc774\ub0b4 \uc5f0\uc18d \uba54\uc2dc\uc9c0\ub97c \ud558\ub098\ub85c \ud1b5\ud569\ud558\uc5ec \uc2e4\uc81c \uc758\ub3c4 \ud30c\uc545\n- **\ud55c\uad6d\uc5b4 \ucc44\ud305 \ud328\ud134 \ubd84\uc11d**: \u314b\u314b, \u314e\u314e, \u3160\u3160, \u3147\u314b \ub4f1 \ud55c\uad6d\uc778\uc758 \ucc44\ud305 \uc2b5\uad00 \uc778\uc2dd\n- **\uc2dc\uc2a4\ud15c \uba54\uc2dc\uc9c0 \ud544\ud130\ub9c1**: \uc0ad\uc81c\ub41c \uba54\uc2dc\uc9c0, \ud22c\ud45c, \uc815\uc0b0, \uc785\uc7a5/\ud1f4\uc7a5 \uc54c\ub9bc \uc790\ub3d9 \uc81c\uac70\n- **\uba58\uc158 \ubd84\uc11d**: @ \uae30\ub2a5\uc744 \uc774\uc6a9\ud55c \uc0ac\uc6a9\uc790 \uc5b8\uae09 \ud328\ud134\uacfc \ub124\ud2b8\uc6cc\ud06c \ubd84\uc11d\n- **\uac10\uc815 \ud45c\ud604 \ubd84\uc11d**: \uc6c3\uc74c, \uc2ac\ud514, \ub3d9\uc758 \ud45c\ud604 \ub4f1\uc758 \uac10\uc815 \ud328\ud134 \ucd94\uc801\n\n### \ud83d\udcca \uae30\ubcf8 \ud1b5\uacc4 \ubd84\uc11d\n- **\uc0ac\uc6a9\uc790\ubcc4 \ud65c\ub3d9 \ud1b5\uacc4**: \uba54\uc2dc\uc9c0 \uc218, \ucc38\uc5ec\uc728, \uc77c\ud3c9\uade0 \ud65c\ub3d9\ub7c9\n- **\uc2dc\uac04\uc801 \ud328\ud134 \ubd84\uc11d**: \uc2dc\uac04\ub300\ubcc4, \uc694\uc77c\ubcc4, \uc6d4\ubcc4 \ud65c\ub3d9 \ubd84\ud3ec  \n- **\ub2e8\uc5b4 \ube48\ub3c4 \ubd84\uc11d**: \ud55c\uad6d\uc5b4 \ud615\ud0dc\uc18c \ubd84\uc11d \ubc0f \ud0a4\uc6cc\ub4dc \ucd94\ucd9c\n- **\uba54\uc2dc\uc9c0\ubcc4 \ub2e8\uc5b4 \uc218 \ud1b5\uacc4**: \uc6d0\ubcf8 \ubc0f \ud544\ud130\ub9c1\ub41c \ub2e8\uc5b4 \uc218 \ubd84\uc11d (\ud3c9\uade0, \uc911\uac04\uac12, \ucd5c\ub300\uac12)\n- **\uc774\ubaa8\ud2f0\ucf58/\uc774\ubaa8\uc9c0 \uc0ac\uc6a9 \ud1b5\uacc4**: \ubc18\uc751 \ud328\ud134 \ubd84\uc11d\n\n### \ud83d\udcac \ub300\ud654 \uc5ed\ud559 \ubd84\uc11d\n- **\ub300\ud654 \ud134 \ubd84\uc11d**: \uc790\ub3d9 \ub300\ud654 \uc138\uc158 \uad6c\ubd84\n- **\uc8fc\uc81c \uc804\ud658 \ud0d0\uc9c0**: \uc784\ubca0\ub529 \uae30\ubc18 \uc8fc\uc81c \ubcc0\ud654 \uac10\uc9c0\n- **\ucc38\uc5ec\uc790 \ud328\ud134**: \ub300\ud654 \uc2dc\uc791\uc790, \uc751\ub2f5\uc790 \ubd84\uc11d (\uc2dc\uc791 \ud69f\uc218 \uc21c\uc73c\ub85c \uc815\ub82c)\n- **\uc0c1\ud638\uc791\uc6a9 \ubd84\uc11d**: \uc0ac\uc6a9\uc790 \uac04 \uc18c\ud1b5 \ud328\ud134\n- **\uba58\uc158 \ub124\ud2b8\uc6cc\ud06c \ubd84\uc11d**: @ \uba58\uc158\uc744 \ud1b5\ud55c \uc0ac\uc6a9\uc790 \uac04 \uc5b8\uae09 \uad00\uacc4 \ubc0f \uc0c1\ud638 \uba58\uc158 \ubd84\uc11d\n\n### \ud83c\udfaf \uc7ac\ubbf8\uc788\ub294 \uc9c0\ud45c\n- **\ucc38\uc5ec \ubd88\ud3c9\ub4f1**: \uc9c0\ub2c8 \uacc4\uc218\ub97c \ud1b5\ud55c \ubc1c\uc5b8\uad8c \ubd84\uc11d\n- **\ub2f5\uc7a5 \uc18d\ub3c4**: \ubc18\uc751 \uc2dc\uac04 \ud1b5\uacc4 \ubc0f \ubd84\ud3ec\n- **\ud65c\ub3d9 \uc5f0\uc18d\uc131**: \uc5f0\uc18d \uba54\uc2dc\uc9c0 \ubc0f \ud65c\ub3d9\uc77c \ubd84\uc11d\n- **\uc2ec\uc57c \ucc44\ud305**: \ubc24\uc0d8 \ub300\ud654 \ud328\ud134 \ubd84\uc11d\n\n### \ud83d\udd0d \uace0\uae09 \ubd84\uc11d\n- **\uc784\ubca0\ub529 \uae30\ubc18 \uc8fc\uc81c \ubd84\uc11d**: \uc758\ubbf8\ub860\uc801 \uc720\uc0ac\ub3c4\ub97c \ud1b5\ud55c \uc8fc\uc81c \uc138\uadf8\uba3c\ud2b8\n- **SPLADE \ud558\uc774\ube0c\ub9ac\ub4dc \uac80\uc0c9**: Dense + Sparse \ubca1\ud130\ub97c \ud65c\uc6a9\ud55c \ub300\ud654 \uac80\uc0c9\n- **\uc2dc\uacc4\uc5f4 \ubd84\uc11d**: \uc2dc\uac04\uc5d0 \ub530\ub978 \ub300\ud654 \ud328\ud134 \ubcc0\ud654\n- **\uac10\uc815 \ubd84\uc11d**: \ub300\ud654 \ubd84\uc704\uae30\uc640 \uac10\uc815 \ud328\ud134 \ucd94\uc801\n- **\uc2a4\ub9c8\ud2b8 \ud1a0\ud53d \ud074\ub7ec\uc2a4\ud130\ub9c1**: \uace0\uae09 \uc8fc\uc81c \ubd84\ub958 \ubc0f \ud14c\ub9c8 \ubd84\uc11d\n- **\ub300\ud654 \ub9ac\ub4ec \ubd84\uc11d**: \uc751\ub2f5 \ud328\ud134\uacfc \uc2dc\uac04\uc801 \ub9ac\ub4ec \ubd84\uc11d\n- **\ub300\uaddc\ubaa8 \ub370\uc774\ud130\uc14b \ucd5c\uc801\ud654**: \uc9c0\ub2a5\ud615 \uccad\ud0b9\uacfc \uba54\ubaa8\ub9ac \ud6a8\uc728\uc131\n\n### \ud83d\udcc8 \uc2dc\uac01\ud654 (\uac1c\uc120\ub428!)\n- **\ud65c\ub3d9 \ud788\ud2b8\ub9f5**: \uc2dc\uac04\ub300 \u00d7 \uc694\uc77c \ud65c\ub3d9 \ubd84\ud3ec\n- **\uc77c\ubcc4 \ud0c0\uc784\ub77c\uc778**: \uba54\uc2dc\uc9c0 \uc218 \ucd94\uc774 \uadf8\ub798\ud504\n- **\uc8fc\uc81c \ud0c0\uc784\ub77c\uc778**: \uc8fc\uc81c \uc138\uadf8\uba3c\ud2b8 \uc2dc\uac01\ud654 (\uacb9\uce68 \ubb38\uc81c \ud574\uacb0)\n- **\uc6cc\ub4dc\ud074\ub77c\uc6b0\ub4dc**: \ud55c\uad6d\uc5b4 \ud3f0\ud2b8 \uc9c0\uc6d0\uc73c\ub85c \uae68\uc9d0 \ud604\uc0c1 \ud574\uacb0\n- **\uc0ac\uc6a9\uc790 \ud65c\ub3d9 \ucc28\ud2b8**: \ub9ce\uc740 \uc0ac\uc6a9\uc790\ub3c4 \uae54\ub054\ud558\uac8c \ud45c\uc2dc (\uacb9\uce68 \ubb38\uc81c \ud574\uacb0)\n\n## \ud83d\udee0\ufe0f \uc124\uce58 \ubc0f \uc2e4\ud589\n\n### \ud83d\udce6 PyPI\ub97c \ud1b5\ud55c \uac04\ud3b8 \uc124\uce58 (\uad8c\uc7a5)\n\n```bash\n# \uae30\ubcf8 \uc124\uce58\npip install kakao-analyzer\n\n# \uc804\uccb4 \uae30\ub2a5 \uc124\uce58 (AI \ubaa8\ub378 \ud3ec\ud568)\npip install kakao-analyzer[full]\n\n# \uac1c\ubc1c \ub3c4\uad6c \ud3ec\ud568 \uc124\uce58\npip install kakao-analyzer[dev]\n```\n\n\uc124\uce58 \ud6c4 \ubc14\ub85c \uc0ac\uc6a9 \uac00\ub2a5:\n```bash\nkakao-analyzer --input your_chat.csv\n```\n\n### \ud83d\udcbb \uac1c\ubc1c\uc790\uc6a9 \uc124\uce58\n\n\uac1c\ubc1c\uc774\ub098 \ucee4\uc2a4\ud130\ub9c8\uc774\uc9d5\uc744 \uc6d0\ud55c\ub2e4\uba74:\n\n```bash\n# \uc800\uc7a5\uc18c \ud074\ub860\ngit clone https://github.com/xistoh162108/kakaotalk_analyzer.git\ncd kakaotalk_analyzer\n\n# \uc758\uc874\uc131 \uc124\uce58\npip install -r requirements.txt\n\n# \uac1c\ubc1c \ubaa8\ub4dc\ub85c \uc124\uce58\npip install -e .\n```\n\n### \ud83d\udccb \uc694\uad6c\uc0ac\ud56d\n- **Python**: 3.8 \uc774\uc0c1\n- **\uc6b4\uc601\uccb4\uc81c**: Windows, macOS, Linux \ubaa8\ub450 \uc9c0\uc6d0\n- **\uba54\ubaa8\ub9ac**: \ucd5c\uc18c 4GB RAM (\ub300\uc6a9\ub7c9 \ub370\uc774\ud130\ub294 8GB+ \uad8c\uc7a5)\n\n## \ud83d\ude80 \ube60\ub978 \uc2dc\uc791\n\n### 1\ub2e8\uacc4: \uc124\uce58\n```bash\npip install kakao-analyzer\n```\n\n### 2\ub2e8\uacc4: \uce74\uce74\uc624\ud1a1 \ub300\ud654 \ub0b4\ubcf4\ub0b4\uae30\n1. \uce74\uce74\uc624\ud1a1 \uc571\uc5d0\uc11c \ubd84\uc11d\ud558\uace0 \uc2f6\uc740 \ub300\ud654\ubc29 \uc5f4\uae30\n2. `\u2630` \uba54\ub274 \u2192 `\ucc44\ud305\ubc29 \uc124\uc815` \u2192 `\ub300\ud654 \ub0b4\uc6a9 \ub0b4\ubcf4\ub0b4\uae30`\n3. `\ud14d\uc2a4\ud2b8 \ud30c\uc77c(.txt)` \uc120\ud0dd\ud558\uc5ec \ub0b4\ubcf4\ub0b4\uae30\n4. \ub0b4\ubcf4\ub0b8 \ud30c\uc77c\uc744 CSV \ud615\uc2dd\uc73c\ub85c \ubcc0\ud658 (\ub610\ub294 \uc9c1\uc811 CSV\ub85c \uc900\ube44)\n\n### 3\ub2e8\uacc4: \ubd84\uc11d \uc2e4\ud589\n```bash\n# \uae30\ubcf8 \ubd84\uc11d\nkakao-analyzer --input your_chat.csv\n\n# \uace0\uae09 AI \ubd84\uc11d \ud3ec\ud568 (\uad8c\uc7a5)\nkakao-analyzer --input your_chat.csv --use-splade --use-ollama\n```\n\n### \u2728 \uc0ac\uc6a9\ubc95\n```bash\n# \uae30\ubcf8 \ubd84\uc11d (\uac00\uc7a5 \uac04\ub2e8)\nkakao-analyzer --input chat.csv\n\n# \ucd9c\ub825 \ub514\ub809\ud1a0\ub9ac \uc9c0\uc815\nkakao-analyzer --input chat.csv --outdir my_analysis\n\n# \ub3c4\uc6c0\ub9d0 \ubcf4\uae30\nkakao-analyzer --help\n```\n\n### \ud83d\udee0\ufe0f \uace0\uae09 \uc635\uc158\n```bash\nkakao-analyzer --input INPUT [\uc635\uc158\ub4e4]\n\n\ud544\uc218 \uc635\uc158:\n  --input INPUT                \uc785\ub825 CSV \ud30c\uc77c \uacbd\ub85c\n\n\ubd84\uc11d \uc635\uc158:\n  --window-minutes MINUTES     \ub300\ud654 \ud134 \uad6c\ubd84 \uc2dc\uac04 (\uae30\ubcf8\uac12: 30\ubd84)\n  --topic-window-size SIZE     \uc8fc\uc81c \ubd84\uc11d \uc708\ub3c4\uc6b0 \ud06c\uae30 (\uae30\ubcf8\uac12: 15)\n  --similarity-threshold FLOAT \uc720\uc0ac\ub3c4 \uc784\uacc4\uac12 (\uae30\ubcf8\uac12: 0.3)\n  --batch-size SIZE            \uc784\ubca0\ub529 \ubc30\uce58 \ud06c\uae30 (\uae30\ubcf8\uac12: 32)\n  --max-workers WORKERS        \ucd5c\ub300 \ubcd1\ub82c \uc791\uc5c5 \uc218 (\uae30\ubcf8\uac12: 4)\n\nAI \ubaa8\ub378 \uc635\uc158:\n  --use-ollama                 Ollama \uc784\ubca0\ub529 \uc0ac\uc6a9 (\ub85c\uceec)\n  --model-name MODEL           Ollama \ubaa8\ub378\uba85 (\uae30\ubcf8\uac12: gpt-oss:20b)\n  --embed-model MODEL          \uc784\ubca0\ub529 \ubaa8\ub378\uba85 (\uae30\ubcf8\uac12: bge-m3)\n  --use-splade                 SPLADE \ud76c\uc18c \uac80\uc0c9 \ud65c\uc131\ud654\n\n\ub300\uaddc\ubaa8 \ub370\uc774\ud130 \uc635\uc158:\n  --subset-size SIZE           \ud14c\uc2a4\ud2b8\uc6a9 \ubd80\ubd84\uc9d1\ud569 \ud06c\uae30  \n  --subset-strategy STRATEGY   \ubd80\ubd84\uc9d1\ud569 \uc804\ub7b5 (balanced/temporal/diverse/random)\n\n\uce74\uce74\uc624\ud1a1 \ud2b9\ud654 \uc635\uc158:\n  --disable-message-grouping   \uc5f0\uc18d \uba54\uc2dc\uc9c0 \uadf8\ub8f9\ud551 \ube44\ud65c\uc131\ud654 (\ube44\ucd94\ucc9c)\n  --group-window-seconds N     \uba54\uc2dc\uc9c0 \uadf8\ub8f9\ud551 \uc2dc\uac04 \ucc3d (\uae30\ubcf8\uac12: 60\ucd08)\n\n\ucd9c\ub825 \uc635\uc158:\n  --outdir DIRECTORY           \ucd9c\ub825 \ub514\ub809\ud1a0\ub9ac\n  --figure-dpi DPI             \uadf8\ub798\ud504 \ud574\uc0c1\ub3c4 (\uae30\ubcf8\uac12: 300)\n  --log-level LEVEL            \ub85c\uadf8 \ub808\ubca8 (DEBUG/INFO/WARNING/ERROR)\n  --quiet                      \ucf58\uc194 \ucd9c\ub825 \uc5b5\uc81c\n  --skip-viz                   \uc2dc\uac01\ud654 \uc0dd\uc131 \uac74\ub108\ub6f0\uae30\n```\n\n### \ud83d\udca1 \uc0ac\uc6a9 \uc608\uc81c\n```bash\n# \ud83c\udf1f \uae30\ubcf8 \ubd84\uc11d (\ucd94\ucc9c)\nkakao-analyzer --input chat.csv\n\n# \ud83e\udd16 AI \uae30\ub2a5 \uc804\uccb4 \ud65c\uc6a9\nkakao-analyzer --input chat.csv --use-splade --use-ollama --model-name \"gpt-oss:20b\"\n\n# \ud83d\udcca \ud2b9\uc815 \ub514\ub809\ud1a0\ub9ac\uc5d0 \uacb0\uacfc \uc800\uc7a5\nkakao-analyzer --input chat.csv --outdir my_results\n\n# \u26a1 \ub300\uc6a9\ub7c9 \ub370\uc774\ud130 \ube60\ub978 \ud14c\uc2a4\ud2b8\nkakao-analyzer --input big_chat.csv --subset-size 1000 --subset-strategy balanced\n\n# \ud83d\udd27 \uc138\ubc00\ud55c \uc124\uc815 \uc870\uc815\nkakao-analyzer --input chat.csv --topic-window-size 20 --similarity-threshold 0.4\n\n# \ud83c\udfa8 \uace0\ud574\uc0c1\ub3c4 \uadf8\ub798\ud504 \uc0dd\uc131\nkakao-analyzer --input chat.csv --figure-dpi 600\n\n# \ud83d\udd07 \uc870\uc6a9\ud55c \uc2e4\ud589 (CI/CD\uc6a9)\nkakao-analyzer --input chat.csv --quiet --skip-viz --log-level ERROR\n```\n\n## \ud83d\udcc1 \ucd9c\ub825 \uad6c\uc870\n\n\ubd84\uc11d \uc644\ub8cc \ud6c4 \ub2e4\uc74c\uacfc \uac19\uc740 \uad6c\uc870\ub85c \uacb0\uacfc\uac00 \uc0dd\uc131\ub429\ub2c8\ub2e4:\n\n```\noutput_directory/\n\u251c\u2500\u2500 summary/                    # \uc694\uc57d \uc815\ubcf4\n\u2502   \u251c\u2500\u2500 analysis_summary.json   # \uc804\uccb4 \ubd84\uc11d \uc694\uc57d\n\u2502   \u2514\u2500\u2500 processed_messages.csv  # \ucc98\ub9ac\ub41c \uba54\uc2dc\uc9c0 \ub370\uc774\ud130\n\u251c\u2500\u2500 basics/                     # \uae30\ubcf8 \ud1b5\uacc4\n\u2502   \u2514\u2500\u2500 basic_statistics.json   # \uc0c1\uc138 \ud1b5\uacc4 \ub370\uc774\ud130\n\u251c\u2500\u2500 context/                    # \ub9e5\ub77d \ubd84\uc11d\n\u2502   \u251c\u2500\u2500 topic_segments.json     # \uc8fc\uc81c \uc138\uadf8\uba3c\ud2b8 \ub370\uc774\ud130\n\u2502   \u251c\u2500\u2500 topic_segments.csv      # \uc8fc\uc81c \uc138\uadf8\uba3c\ud2b8 \ud14c\uc774\ube14\n\u2502   \u251c\u2500\u2500 mood_analysis.json      # \uac10\uc815 \ubc0f \ubd84\uc704\uae30 \ubd84\uc11d\n\u2502   \u251c\u2500\u2500 advanced_topics.json    # \uace0\uae09 \uc8fc\uc81c \ud074\ub7ec\uc2a4\ud130\ub9c1\n\u2502   \u251c\u2500\u2500 rhythm_analysis.json    # \ub300\ud654 \ub9ac\ub4ec \ubd84\uc11d\n\u2502   \u251c\u2500\u2500 mention_analysis.json   # \uba58\uc158 \ud328\ud134 \ubc0f \ub124\ud2b8\uc6cc\ud06c \ubd84\uc11d\n\u2502   \u251c\u2500\u2500 hybrid_index.meta.json  # \uac80\uc0c9 \uc778\ub371\uc2a4 (SPLADE)\n\u2502   \u2514\u2500\u2500 splade_scores.parquet   # \ud76c\uc18c \ubca1\ud130 \ub370\uc774\ud130 (SPLADE)\n\u251c\u2500\u2500 fun/                        # \uc7ac\ubbf8 \uc9c0\ud45c\n\u2502   \u2514\u2500\u2500 fun_metrics.json        # \ucc38\uc5ec \ubd88\ud3c9\ub4f1, \ub2f5\uc7a5 \uc18d\ub3c4 \ub4f1\n\u251c\u2500\u2500 figures/                    # \uc2dc\uac01\ud654\n\u2502   \u251c\u2500\u2500 heatmap_hour_weekday.png\n\u2502   \u251c\u2500\u2500 timeseries_daily.png\n\u2502   \u251c\u2500\u2500 topic_timeline.png\n\u2502   \u251c\u2500\u2500 wordcloud_global.png\n\u2502   \u251c\u2500\u2500 user_activity_distribution.png\n\u2502   \u2514\u2500\u2500 reply_latency_distribution.png\n\u251c\u2500\u2500 reports/                    # \ub9ac\ud3ec\ud2b8\n\u2502   \u251c\u2500\u2500 analysis_report.md      # \uc885\ud569 \ubd84\uc11d \ub9ac\ud3ec\ud2b8\n\u2502   \u2514\u2500\u2500 qa_test_report.md       # \ud488\uc9c8 \ubcf4\uc99d \ub9ac\ud3ec\ud2b8 (\ud14c\uc2a4\ud2b8\uc2dc)\n\u2514\u2500\u2500 logs/                       # \ub85c\uadf8\n    \u251c\u2500\u2500 pipeline_metadata.json  # \uc2e4\ud589 \uba54\ud0c0\ub370\uc774\ud130\n    \u2514\u2500\u2500 analyzer.log            # \ubd84\uc11d \ub85c\uadf8\n```\n\n## \ud83d\udccb \uc785\ub825 \ub370\uc774\ud130 \ud615\uc2dd\n\n\uce74\uce74\uc624\ud1a1\uc5d0\uc11c \ub0b4\ubcf4\ub0b8 CSV \ud30c\uc77c\uc774 \ud544\uc694\ud569\ub2c8\ub2e4. \uc9c0\uc6d0\ud558\ub294 \uceec\ub7fc \ud615\uc2dd:\n\n### \ud544\uc218 \uceec\ub7fc\n- `user` \ub610\ub294 `User` \ub610\ub294 `\uc0ac\uc6a9\uc790`: \uba54\uc2dc\uc9c0 \uc791\uc131\uc790\n- `message` \ub610\ub294 `Message` \ub610\ub294 `\uba54\uc2dc\uc9c0`: \uba54\uc2dc\uc9c0 \ub0b4\uc6a9  \n- `datetime` \ub610\ub294 `Date` \ub610\ub294 `\uc2dc\uac04`: \uba54\uc2dc\uc9c0 \uc2dc\uac04\n\n### \uc608\uc81c CSV \ud615\uc2dd\n```csv\nuser,message,datetime\n\uae40\ubbfc\uc218,\uc548\ub155\ud558\uc138\uc694!,2024-01-01 10:00:00\n\uc774\uc9c0\uc740,\ub124 \uc548\ub155\ud558\uc138\uc694! \u314e\u314e,2024-01-01 10:01:00\n\ubc15\uc11c\uc900,\uc624\ub298 \ub0a0\uc528 \uc88b\ub124\uc694,2024-01-01 10:02:00\n```\n\n## \ud83e\uddea \ud14c\uc2a4\ud2b8\n\n### \uae30\ubcf8 \uae30\ub2a5 \ud14c\uc2a4\ud2b8\n```bash\n# \uae30\ubcf8 \uae30\ub2a5 \ud14c\uc2a4\ud2b8 (\uc758\uc874\uc131 \uc5c6\uc774)\npython simple_test.py\n\n# \uc804\uccb4 \ud14c\uc2a4\ud2b8 \uc2a4\uc704\ud2b8 (\uc758\uc874\uc131 \uc124\uce58 \ud6c4)\npython test_runner.py\n```\n\n### \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130 \uc0dd\uc131\n```bash\n# 1000\uac1c \uba54\uc2dc\uc9c0 \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130 \uc0dd\uc131\npython test_runner.py --generate-data\n```\n\n## \ud83d\udd27 \uace0\uae09 \uc124\uc815\n\n### \ud83d\ude80 GPU \uac00\uc18d \uc9c0\uc6d0 (NEW!)\n- **\uc790\ub3d9 GPU \uc778\uc2dd**: CUDA\uac00 \uc124\uce58\ub41c \uc2dc\uc2a4\ud15c\uc5d0\uc11c \uc790\ub3d9\uc73c\ub85c GPU \uc0ac\uc6a9\n- **\uba54\ubaa8\ub9ac \ucd5c\uc801\ud654**: GPU \uba54\ubaa8\ub9ac\uc758 75%\uae4c\uc9c0 \uc790\ub3d9 \uc81c\ud55c\ud558\uc5ec \uc548\uc815\uc131 \ubcf4\uc7a5\n- **\ubaa8\ub378 \uc790\ub3d9 \uc774\uad00**: \uc784\ubca0\ub529 \ubc0f SPLADE \ubaa8\ub378\uc744 GPU\ub85c \uc790\ub3d9 \ub85c\ub4dc\n- **\ubc30\uce58 \ud06c\uae30 \ucd5c\uc801\ud654**: GPU \uba54\ubaa8\ub9ac\uc5d0 \ub530\ub77c \ubc30\uce58 \ud06c\uae30 \uc790\ub3d9 \uc870\uc815\n\n#### GPU \uba54\ubaa8\ub9ac\ubcc4 \ucd5c\uc801\ud654\n- **12GB+ (RTX 3080Ti/4080 \uc774\uc0c1)**: \ubc30\uce58 \ud06c\uae30 128, Half precision\n- **8-12GB (RTX 3070/4070)**: \ubc30\uce58 \ud06c\uae30 64, Half precision  \n- **4-8GB (GTX 1660/RTX 3060)**: \ubc30\uce58 \ud06c\uae30 48, Mixed precision\n- **4GB \ubbf8\ub9cc**: CPU \ubaa8\ub4dc\ub85c \uc790\ub3d9 \uc804\ud658\n\n### \uc784\ubca0\ub529 \ubaa8\ub378\n- **\uae30\ubcf8\uac12**: Mock \uc784\ubca0\ub529 (\uc678\ubd80 \uc758\uc874\uc131 \uc5c6\uc74c)\n- **\uace0\uae09 \uc635\uc158**: SentenceTransformer \ubaa8\ub378 (`sentence-transformers` \uc124\uce58 \ud544\uc694)\n- **GPU \uac00\uc18d**: CUDA \uc9c0\uc6d0 \uc2dc \uc790\ub3d9\uc73c\ub85c GPU\uc5d0\uc11c \uc2e4\ud589\n\n### SPLADE \ud558\uc774\ube0c\ub9ac\ub4dc \uac80\uc0c9\n- **\ud65c\uc131\ud654**: `--use-splade` \uc635\uc158 \uc0ac\uc6a9\n- **\uae30\ub2a5**: Dense + Sparse \ubca1\ud130\ub97c \uacb0\ud569\ud55c \uace0\uae09 \uac80\uc0c9\n- **\uc6a9\ub3c4**: \ud2b9\uc815 \uc8fc\uc81c\ub098 \ud0a4\uc6cc\ub4dc\ub85c \ub300\ud654 \uac80\uc0c9\n- **GPU \uac00\uc18d**: Half precision\uc73c\ub85c GPU\uc5d0\uc11c \uace0\uc18d \ucc98\ub9ac\n\n### Ollama \uc5f0\ub3d9 (\ud55c\uad6d\uc5b4 \ubaa8\ub378 \uc644\uc804 \uc9c0\uc6d0)\n- **\ud65c\uc131\ud654**: `--use-ollama` \uc635\uc158 \uc0ac\uc6a9\n- **\uc694\uad6c\uc0ac\ud56d**: Ollama \uc11c\ubc84 \uc2e4\ud589 \uc911\uc774\uc5b4\uc57c \ud568\n- **\uae30\ubcf8 \ubaa8\ub378**: gpt-oss:20b (\ucd94\ucc9c)\n\n#### \ud83c\uddf0\ud83c\uddf7 \uc9c0\uc6d0 \ud55c\uad6d\uc5b4 \ubaa8\ub378\ub4e4\n| \ubaa8\ub378\uba85 | \ud06c\uae30 | \ud2b9\uc9d5 | \uc0ac\uc6a9\ubc95 |\n|--------|------|------|--------|\n| `gpt-oss:20b` | 20B | \ubc94\uc6a9 \ud55c\uad6d\uc5b4 \ubaa8\ub378 (\uae30\ubcf8\uac12) | `--model-name \"gpt-oss:20b\"` |\n| `gpt-oss:40b` | 40B | \uace0\uc131\ub2a5 \ud55c\uad6d\uc5b4 \ubaa8\ub378 | `--model-name \"gpt-oss:40b\"` |\n| `llama3-ko` | 8B | LLaMA3 \ud55c\uad6d\uc5b4 \ud30c\uc778\ud29c\ub2dd | `--model-name \"llama3-ko\"` |\n| `aya:8b` | 8B | \ub2e4\uad6d\uc5b4 \uc9c0\uc6d0 (\ud55c\uad6d\uc5b4 \ud3ec\ud568) | `--model-name \"aya:8b\"` |\n| `gemma2-ko` | 9B | Google Gemma2 \ud55c\uad6d\uc5b4 \ubc84\uc804 | `--model-name \"gemma2-ko\"` |\n| `solar-ko` | 10.7B | Upstage SOLAR \ud55c\uad6d\uc5b4 \ubaa8\ub378 | `--model-name \"solar-ko\"` |\n| `eeve-korean` | 10.8B | \ud55c\uad6d \uc804\uc6a9 \ub300\ud654 \ubaa8\ub378 | `--model-name \"eeve-korean\"` |\n| `beomi-ko` | 13B | \ud55c\uad6d\uc5b4 \ud2b9\ud654 \uace0\uc131\ub2a5 \ubaa8\ub378 | `--model-name \"beomi-ko\"` |\n\n#### \ud83d\ude80 \ubaa8\ub378 \uc124\uce58 \ubc0f \uc0ac\uc6a9\ubc95\n```bash\n# Ollama \uc124\uce58 (\ud55c \ubc88\ub9cc)\ncurl -fsSL https://ollama.ai/install.sh | sh\n\n# \ud55c\uad6d\uc5b4 \ubaa8\ub378 \ub2e4\uc6b4\ub85c\ub4dc (\uc608: gpt-oss:20b)\nollama pull gpt-oss:20b\n\n# \ubd84\uc11d \uc2e4\ud589\nkakao-analyzer --input chat.csv --use-ollama --model-name \"gpt-oss:20b\"\n```\n\n## \ud83d\udcca \ubd84\uc11d \uacb0\uacfc \ud574\uc11d\n\n### \ucc38\uc5ec \ubd88\ud3c9\ub4f1 (\uc9c0\ub2c8 \uacc4\uc218)\n- **0.0-0.3**: \ub9e4\uc6b0 \uade0\ub4f1\ud55c \ucc38\uc5ec\n- **0.3-0.5**: \uc57d\uac04 \ubd88\uade0\ub4f1\n- **0.5-0.7**: \ubd88\uade0\ub4f1\ud55c \ucc38\uc5ec\n- **0.7-1.0**: \ub9e4\uc6b0 \ubd88\uade0\ub4f1 (\uc18c\uc218\uac00 \ub300\ud654 \uc8fc\ub3c4)\n\n### \uba58\uc158 \ubd84\uc11d \uc9c0\ud45c\n- **\uc0c1\ud638 \uba58\uc158 \uad00\uacc4**: \uc11c\ub85c \uba58\uc158\ud558\ub294 \uc0ac\uc6a9\uc790 \uc30d\uc758 \uc218\uc640 \ube48\ub3c4\n- **\uba58\uc158 \ub124\ud2b8\uc6cc\ud06c \ubc00\ub3c4**: \uc804\uccb4 \uac00\ub2a5\ud55c \uba58\uc158 \uad00\uacc4 \ub300\ube44 \uc2e4\uc81c \uba58\uc158 \uad00\uacc4 \ube44\uc728\n- **\uba58\uc158 \ubaa9\uc801 \ubd84\ub958**: \uc9c8\ubb38, \ubd80\ud0c1/\uc694\uccad, \uc778\uc0ac, \uae34\uae09, \ud1a0\ub860/\uc758\uacac \ub4f1\uc73c\ub85c \uc790\ub3d9 \ubd84\ub958\n\n### \ub2f5\uc7a5 \uc18d\ub3c4 \uce74\ud14c\uace0\ub9ac\n- **\uc989\uc11d \ub2f5\uc7a5**: 1\ubd84 \uc774\ub0b4\n- **\ube60\ub978 \ub2f5\uc7a5**: 1-5\ubd84\n- **\ubcf4\ud1b5 \ub2f5\uc7a5**: 5-30\ubd84\n- **\ub290\ub9b0 \ub2f5\uc7a5**: 30\ubd84 \uc774\uc0c1\n\n### \uc8fc\uc81c \uc138\uadf8\uba3c\ud2b8\n- \uc784\ubca0\ub529 \uc720\uc0ac\ub3c4 \uae30\ubc18\uc73c\ub85c \uc790\ub3d9 \uac10\uc9c0\n- \uac01 \uc138\uadf8\uba3c\ud2b8\ubcc4 \ud0a4\uc6cc\ub4dc, \ucc38\uc5ec\uc790, \uc9c0\uc18d\uc2dc\uac04 \uc81c\uacf5\n- \uc8fc\uc81c \uc804\ud658 \ud328\ud134 \ubd84\uc11d\n\n## \ud83c\udfa8 \uc2dc\uac01\ud654 \uac00\uc774\ub4dc\n\n\ubaa8\ub4e0 \uadf8\ub798\ud504\ub294 \ud55c\uad6d\uc5b4 \ud3f0\ud2b8\ub97c \uc9c0\uc6d0\ud558\uba70, \ub2e4\uc74c \ud615\uc2dd\uc73c\ub85c \uc800\uc7a5\ub429\ub2c8\ub2e4:\n- **\ud615\uc2dd**: PNG (\uae30\ubcf8\uac12)\n- **\ud574\uc0c1\ub3c4**: 300 DPI (\uae30\ubcf8\uac12)\n- **\ud06c\uae30**: \uc790\ub3d9 \uc870\uc815\n\n### \uc8fc\uc694 \uc2dc\uac01\ud654\n1. **\ud65c\ub3d9 \ud788\ud2b8\ub9f5**: \uc5b8\uc81c \uac00\uc7a5 \ud65c\ubc1c\ud55c\uc9c0 \ud55c\ub208\uc5d0 \ud30c\uc545\n2. **\uc77c\ubcc4 \ud0c0\uc784\ub77c\uc778**: \uc2dc\uac04\uc5d0 \ub530\ub978 \ub300\ud654\ub7c9 \ubcc0\ud654\n3. **\uc8fc\uc81c \ud0c0\uc784\ub77c\uc778**: \ub300\ud654 \uc8fc\uc81c\uc758 \uc2dc\uac04\uc801 \ud750\ub984\n4. **\uc6cc\ub4dc\ud074\ub77c\uc6b0\ub4dc**: \uc790\uc8fc \uc0ac\uc6a9\ub41c \ub2e8\uc5b4\ub4e4\n5. **\uc0ac\uc6a9\uc790 \ud65c\ub3d9**: \uac01 \ucc38\uc5ec\uc790\uc758 \uae30\uc5ec\ub3c4\n6. **\ub2f5\uc7a5 \uc18d\ub3c4**: \uc18c\ud1b5 \uc2a4\ud0c0\uc77c \ubd84\uc11d\n\n## \ud83d\udca1 \ud65c\uc6a9 \uc0ac\ub840\n\n### \uac1c\uc778/\uadf8\ub8f9 \ubd84\uc11d\n- \uce5c\uad6c\ub4e4\uacfc\uc758 \ub300\ud654 \ud328\ud134 \ubd84\uc11d\n- \uac00\uc871 \ub2e8\ud1a1\ubc29 \ud65c\ub3d9 \ubd84\uc11d\n- \ub3d9\ud638\ud68c/\ucee4\ubba4\ub2c8\ud2f0 \uc18c\ud1b5 \ud328\ud134\n\n### \uc5f0\uad6c \ubaa9\uc801\n- \ub514\uc9c0\ud138 \ucee4\ubba4\ub2c8\ucf00\uc774\uc158 \uc5f0\uad6c\n- \uc0ac\ud68c \ub124\ud2b8\uc6cc\ud06c \ubd84\uc11d\n- \uc5b8\uc5b4\ud559 \uc5f0\uad6c \ub370\uc774\ud130\n\n### \ube44\uc988\ub2c8\uc2a4 \ubd84\uc11d\n- \ud300 \ucee4\ubba4\ub2c8\ucf00\uc774\uc158 \ud6a8\uc728\uc131\n- \uace0\uac1d \uc18c\ud1b5 \ud328\ud134\n- \ucee4\ubba4\ub2c8\ud2f0 \uad00\ub9ac \uc778\uc0ac\uc774\ud2b8\n\n## \ud83d\udee1\ufe0f \uac1c\uc778\uc815\ubcf4 \ubcf4\ud638\n\n- \ubaa8\ub4e0 \ubd84\uc11d\uc740 \ub85c\uceec\uc5d0\uc11c \uc218\ud589\ub429\ub2c8\ub2e4\n- \ub370\uc774\ud130\ub294 \uc678\ubd80\ub85c \uc804\uc1a1\ub418\uc9c0 \uc54a\uc2b5\ub2c8\ub2e4\n- \uacb0\uacfc \ud30c\uc77c\uc5d0\ub294 \uc6d0\ubcf8 \uba54\uc2dc\uc9c0\uac00 \ud3ec\ud568\ub420 \uc218 \uc788\uc73c\ub2c8 \uacf5\uc720 \uc2dc \uc8fc\uc758\ud558\uc138\uc694\n- \uac1c\uc778 \uc2dd\ubcc4 \uc815\ubcf4\uac00 \ud3ec\ud568\ub41c \ubd84\uc11d \uacb0\uacfc\ub294 \uc2e0\uc911\ud788 \uad00\ub9ac\ud558\uc138\uc694\n\n## \ud83d\udc1b \ubb38\uc81c \ud574\uacb0\n\n### \uc77c\ubc18\uc801\uc778 \ubb38\uc81c\n\n**1. \ubaa8\ub4c8\uc744 \ucc3e\uc744 \uc218 \uc5c6\uc74c**\n```bash\npip install -r requirements.txt\n```\n\n**2. \ud55c\uad6d\uc5b4 \ud3f0\ud2b8 \ubb38\uc81c**\n- macOS: Apple Gothic \uc790\ub3d9 \uc0ac\uc6a9\n- Windows: Malgun Gothic \uc790\ub3d9 \uc0ac\uc6a9\n- Linux: Noto Sans CJK KR \uc124\uce58 \uad8c\uc7a5\n\n**3. \uba54\ubaa8\ub9ac \ubd80\uc871**\n- \ud070 \ub370\uc774\ud130\uc14b\uc758 \uacbd\uc6b0 `--topic-window-size` \uac12\uc744 \uc904\uc5ec\ubcf4\uc138\uc694\n- \uc784\ubca0\ub529 \ube44\ud65c\uc131\ud654: \uc124\uc815\uc5d0\uc11c `use_embeddings = False`\n\n**4. CSV \uc778\ucf54\ub529 \ubb38\uc81c**\n- \uc790\ub3d9 \uc778\ucf54\ub529 \uac10\uc9c0 \uae30\ub2a5 \ub0b4\uc7a5\n- \uc218\ub3d9 \uc9c0\uc815\uc774 \ud544\uc694\ud55c \uacbd\uc6b0 \ud30c\uc77c\uc744 UTF-8\ub85c \ubcc0\ud658\n\n### \ub85c\uadf8 \ud655\uc778\n```bash\n# \uc0c1\uc138 \ub85c\uadf8\ub85c \uc2e4\ud589\nkakao-analyzer --input chat.csv --log-level DEBUG\n\n# \ub85c\uadf8 \ud30c\uc77c \ud655\uc778 (\uae30\ubcf8 \ucd9c\ub825 \ub514\ub809\ud1a0\ub9ac)\ncat chat_analysis/logs/run.log\n```\n\n## \ud83d\udcc8 \uc131\ub2a5 \ucd5c\uc801\ud654\n\n### \ub300\uc6a9\ub7c9 \ub370\uc774\ud130 \ucc98\ub9ac\n- **\ubc30\uce58 \ud06c\uae30 \uc870\uc815**: \uc124\uc815\uc5d0\uc11c `batch_size` \ubcc0\uacbd\n- **\uc784\ubca0\ub529 \ube44\ud65c\uc131\ud654**: \uae30\ubcf8 \ubd84\uc11d\ub9cc \ud544\uc694\ud55c \uacbd\uc6b0\n- **\uc708\ub3c4\uc6b0 \ud06c\uae30 \uac10\uc18c**: \uba54\ubaa8\ub9ac \uc0ac\uc6a9\ub7c9 \uc808\uc57d\n\n### \uc18d\ub3c4 \ud5a5\uc0c1\n- **\ubcd1\ub82c \ucc98\ub9ac**: \uac00\ub2a5\ud55c \uacf3\uc5d0\uc11c \uc790\ub3d9 \uc801\uc6a9\n- **\uce90\uc2f1**: \uc911\uac04 \uacb0\uacfc \uc790\ub3d9 \uc800\uc7a5\n- **\ucd5c\uc801\ud654\ub41c \uc54c\uace0\ub9ac\uc998**: \ub300\uc6a9\ub7c9 \ub370\uc774\ud130 \ub300\uc751\n\n## \ud83d\udea7 \ub85c\ub4dc\ub9f5\n\n\ud604\uc7ac \uc774 \ud504\ub85c\uc81d\ud2b8\ub294 \uc9c0\uc18d\uc801\uc73c\ub85c \uac1c\ubc1c \uc911\uc774\uba70, \ub2e4\uc74c\uacfc \uac19\uc740 \uae30\ub2a5\ub4e4\uc774 \ucd94\uac00\ub420 \uc608\uc815\uc785\ub2c8\ub2e4:\n\n### \ud83d\udd2e \uc608\uc815\ub41c \uae30\ub2a5\ub4e4\n- **\ud83c\udf10 \ub2e4\uc911 \uba54\uc2e0\uc800 \uc9c0\uc6d0**: \ub77c\uc778, \ud154\ub808\uadf8\ub7a8, \ub514\uc2a4\ucf54\ub4dc \ub4f1 \ub2e4\ub978 \uba54\uc2e0\uc800 \ud50c\ub7ab\ud3fc \uc9c0\uc6d0\n- **\ud83d\udcf1 \uc6f9 \uc778\ud130\ud398\uc774\uc2a4**: \ube0c\ub77c\uc6b0\uc800\uc5d0\uc11c \uc0ac\uc6a9\ud560 \uc218 \uc788\ub294 \uc9c1\uad00\uc801\uc778 UI\n- **\u26a1 \uc2e4\uc2dc\uac04 \ubd84\uc11d**: \ub300\ud654 \uc9c4\ud589 \uc911 \uc2e4\uc2dc\uac04 \ud328\ud134 \ubd84\uc11d\n- **\ud83e\udd16 \uace0\uae09 AI \uae30\ub2a5**: GPT-4 \ud1b5\ud569, \ub354 \uc815\uad50\ud55c \uac10\uc815 \ubd84\uc11d\n- **\ud83d\udcca \ub300\uc2dc\ubcf4\ub4dc**: \uc2e4\uc2dc\uac04 \ubaa8\ub2c8\ud130\ub9c1 \ubc0f \uc54c\ub9bc \uc2dc\uc2a4\ud15c\n- **\ud83d\udd0d \uac80\uc0c9 \uc5d4\uc9c4**: \ub300\ud654 \ub0b4\uc6a9 \uace0\uae09 \uac80\uc0c9 \ubc0f \ud544\ud130\ub9c1\n- **\ud83d\udcc8 \ud2b8\ub80c\ub4dc \ubd84\uc11d**: \uc2dc\uac04\uc5d0 \ub530\ub978 \ub300\ud654 \ud328\ud134 \ubcc0\ud654 \ucd94\uc801\n- **\ud83c\udfa8 \ud14c\ub9c8 \ucee4\uc2a4\ud130\ub9c8\uc774\uc81c\uc774\uc158**: \uc2dc\uac01\ud654 \uc2a4\ud0c0\uc77c \ubc0f \uc0c9\uc0c1 \ud14c\ub9c8 \uc120\ud0dd\n- **\ud83d\udcbe \ud074\ub77c\uc6b0\ub4dc \ubc31\uc5c5**: \ubd84\uc11d \uacb0\uacfc \uc790\ub3d9 \ubc31\uc5c5 \ubc0f \ub3d9\uae30\ud654\n- **\ud83d\udce7 \ub9ac\ud3ec\ud2b8 \uc790\ub3d9 \ubc1c\uc1a1**: \uc815\uae30\uc801\uc778 \ubd84\uc11d \ub9ac\ud3ec\ud2b8 \uc774\uba54\uc77c \ubc1c\uc1a1\n\n### \ud83c\udfaf \ub2e4\uc74c \ubc84\uc804 (v1.1.0) \uc608\uc815 \uae30\ub2a5\n- \uc74c\uc131 \uba54\uc2dc\uc9c0 \uae38\uc774 \ubd84\uc11d\n- \ud30c\uc77c \uacf5\uc720 \ud328\ud134 \ubd84\uc11d\n- \ub300\ud654\ubc29 \ucc38\uc5ec\uc790 \ubcc0\ud654 \ucd94\uc801\n- \uac10\uc815 \ubcc0\ud654 \uc2dc\uacc4\uc5f4 \ubd84\uc11d\n- \ud0a4\uc6cc\ub4dc \uc54c\ub9bc \uc2dc\uc2a4\ud15c\n\n## \ud83d\udc68\u200d\ud83d\udcbb \uac1c\ubc1c\uc790\n\n### \ud83c\udf93 \uc8fc \uac1c\ubc1c\uc790\n\n- **xistoh162108** (KAIST)\n  - \ud83d\udce7 \uc774\uba54\uc77c: xistoh162108@kaist.ac.kr\n  - \ud83c\udfeb \uc18c\uc18d: \ud55c\uad6d\uacfc\ud559\uae30\uc220\uc6d0 (KAIST)\n  - \ud83d\udd17 GitHub: [@xistoh162108](https://github.com/xistoh162108)\n\n### \ud83e\udd16 AI \ud611\ub825 \uac1c\ubc1c\n\n- **Claude AI** (Anthropic)\n  - \ucf54\ub4dc \uc544\ud0a4\ud14d\ucc98 \uc124\uacc4 \ubc0f \uad6c\ud604\n  - \uc54c\uace0\ub9ac\uc998 \ucd5c\uc801\ud654 \ubc0f \ud55c\uad6d\uc5b4 \ud2b9\ud654 \uae30\ub2a5\n  - \ubb38\uc11c\ud654 \ubc0f \uc0ac\uc6a9\uc790 \uacbd\ud5d8 \uac1c\uc120\n\n\uc774 \ud504\ub85c\uc81d\ud2b8\ub294 **\uc778\uac04-AI \ud611\uc5c5**\uc758 \uacb0\uacfc\ubb3c\ub85c, \ud559\uc220 \uc5f0\uad6c\uc640 \uc2e4\uc6a9\uc801 \ub3c4\uad6c \uac1c\ubc1c\uc758 \ub9cc\ub0a8\uc744 \ubcf4\uc5ec\uc90d\ub2c8\ub2e4.\n\n## \ud83e\udd1d \uae30\uc5ec\ud558\uae30\n\n### \ud83d\udd17 \ud611\uc5c5 \ubc29\uc2dd\n\n- **\uc774\uc288 \ub9ac\ud3ec\ud305**: [GitHub Issues](https://github.com/xistoh162108/kakaotalk_analyzer/issues)\n- **\uae30\ub2a5 \uc81c\uc548**: Pull Request \ud658\uc601\n- **\ud559\uc220 \ud611\ub825**: \uc5f0\uad6c \ubaa9\uc801 \uc0ac\uc6a9 \uc2dc \uc778\uc6a9 \ubd80\ud0c1\ub4dc\ub9bd\ub2c8\ub2e4\n\n### \ud83d\udca1 \uac1c\uc120 \uc544\uc774\ub514\uc5b4\n\n- \uc0c8\ub85c\uc6b4 \ubd84\uc11d \uc9c0\ud45c \ucd94\uac00\n- \ub2e4\ub978 \uba54\uc2e0\uc800 \uc9c0\uc6d0 (\ub77c\uc778, \ud154\ub808\uadf8\ub7a8 \ub4f1)\n- \uc6f9 \uc778\ud130\ud398\uc774\uc2a4 \uac1c\ubc1c\n- \uc2e4\uc2dc\uac04 \ubd84\uc11d \uae30\ub2a5\n\n### \ud83d\udcdd \uc778\uc6a9 (Citation)\n\n\ud559\uc220 \ub17c\ubb38\uc774\ub098 \uc5f0\uad6c\uc5d0\uc11c \uc0ac\uc6a9\ud558\uc2e4 \ub54c:\n\n```bibtex\nxistoh162108, & Claude AI. (2025). Kakao Talk Analyzer: \nComprehensive analysis tool for KakaoTalk chat exports (v1.0.1). \nhttps://github.com/xistoh162108/kakaotalk_analyzer\n```\n\n## \ud83d\udcc4 \ub77c\uc774\uc120\uc2a4\n\n\uc774 \ud504\ub85c\uc81d\ud2b8\ub294 MIT \ub77c\uc774\uc120\uc2a4 \ud558\uc5d0 \ubc30\ud3ec\ub429\ub2c8\ub2e4. \uc790\uc138\ud55c \ub0b4\uc6a9\uc740 [LICENSE](LICENSE) \ud30c\uc77c\uc744 \ucc38\uace0\ud558\uc138\uc694.\n\n## \ud83d\ude4f \uac10\uc0ac\uc758 \ub9d0\n\n- **\ud55c\uad6d\uc5b4 \uc790\uc5f0\uc5b4 \ucc98\ub9ac**: KoNLPy, mecab-python3\n- **\uc2dc\uac01\ud654**: matplotlib, seaborn, wordcloud\n- **\ub370\uc774\ud130 \ucc98\ub9ac**: pandas, numpy\n- **\uba38\uc2e0\ub7ec\ub2dd**: scikit-learn, sentence-transformers\n\n---\n\n**\ud83e\udd16 Generated with [Claude Code](https://claude.ai/code)**\n\nCo-Authored-By: Claude <noreply@anthropic.com>\n",
    "bugtrack_url": null,
    "license": "MIT",
    "summary": "Comprehensive analysis tool for KakaoTalk chat exports",
    "version": "1.0.2",
    "project_urls": {
        "Bug Reports": "https://github.com/xistoh162108/kakaotalk_analyzer/issues",
        "Documentation": "https://github.com/xistoh162108/kakaotalk_analyzer/blob/main/README.md",
        "Homepage": "https://github.com/xistoh162108/kakaotalk_analyzer",
        "Source": "https://github.com/xistoh162108/kakaotalk_analyzer"
    },
    "split_keywords": [
        "kakao",
        "kakaotalk",
        "chat",
        "analysis",
        "nlp",
        "korean",
        "messaging"
    ],
    "urls": [
        {
            "comment_text": null,
            "digests": {
                "blake2b_256": "e3b63e0dabe4720f1dfd8b4c4aa95e9fbc19e6b8157a4b23809c5f1ca192de8f",
                "md5": "ee055c18974ab2b384871b2363a12ff9",
                "sha256": "2a3c154a5b04d8936c6d8d140b689e5c97d1a291387df9c6527c4d8832efc7b7"
            },
            "downloads": -1,
            "filename": "kakao_analyzer-1.0.2-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "ee055c18974ab2b384871b2363a12ff9",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": ">=3.8",
            "size": 107244,
            "upload_time": "2025-08-14T08:53:29",
            "upload_time_iso_8601": "2025-08-14T08:53:29.031586Z",
            "url": "https://files.pythonhosted.org/packages/e3/b6/3e0dabe4720f1dfd8b4c4aa95e9fbc19e6b8157a4b23809c5f1ca192de8f/kakao_analyzer-1.0.2-py3-none-any.whl",
            "yanked": false,
            "yanked_reason": null
        },
        {
            "comment_text": null,
            "digests": {
                "blake2b_256": "846d863a8d2c93da2cb710da2d1432b4e0bd13937e6c97bccda867f49192f532",
                "md5": "6eef039d2a61b6b75e3fb7916f44add0",
                "sha256": "d77e46fb96b2bda7e8082dda971862909ea54021c015ae742a14249612f1f562"
            },
            "downloads": -1,
            "filename": "kakao_analyzer-1.0.2.tar.gz",
            "has_sig": false,
            "md5_digest": "6eef039d2a61b6b75e3fb7916f44add0",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": ">=3.8",
            "size": 103692,
            "upload_time": "2025-08-14T08:53:30",
            "upload_time_iso_8601": "2025-08-14T08:53:30.664790Z",
            "url": "https://files.pythonhosted.org/packages/84/6d/863a8d2c93da2cb710da2d1432b4e0bd13937e6c97bccda867f49192f532/kakao_analyzer-1.0.2.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2025-08-14 08:53:30",
    "github": true,
    "gitlab": false,
    "bitbucket": false,
    "codeberg": false,
    "github_user": "xistoh162108",
    "github_project": "kakaotalk_analyzer",
    "travis_ci": false,
    "coveralls": false,
    "github_actions": false,
    "requirements": [
        {
            "name": "pandas",
            "specs": [
                [
                    ">=",
                    "1.5.0"
                ]
            ]
        },
        {
            "name": "numpy",
            "specs": [
                [
                    ">=",
                    "1.21.0"
                ]
            ]
        },
        {
            "name": "matplotlib",
            "specs": [
                [
                    ">=",
                    "3.5.0"
                ]
            ]
        },
        {
            "name": "seaborn",
            "specs": [
                [
                    ">=",
                    "0.11.0"
                ]
            ]
        },
        {
            "name": "wordcloud",
            "specs": [
                [
                    ">=",
                    "1.8.0"
                ]
            ]
        },
        {
            "name": "scikit-learn",
            "specs": [
                [
                    ">=",
                    "1.1.0"
                ]
            ]
        },
        {
            "name": "sentence-transformers",
            "specs": [
                [
                    ">=",
                    "2.2.0"
                ]
            ]
        },
        {
            "name": "torch",
            "specs": [
                [
                    ">=",
                    "1.13.0"
                ]
            ]
        },
        {
            "name": "transformers",
            "specs": [
                [
                    ">=",
                    "4.21.0"
                ]
            ]
        },
        {
            "name": "chardet",
            "specs": [
                [
                    ">=",
                    "4.0.0"
                ]
            ]
        },
        {
            "name": "tqdm",
            "specs": [
                [
                    ">=",
                    "4.60.0"
                ]
            ]
        },
        {
            "name": "requests",
            "specs": [
                [
                    ">=",
                    "2.25.0"
                ]
            ]
        }
    ],
    "lcname": "kakao-analyzer"
}