import requests
import openai
from googletrans import Translator
import deepl
class HybridTranslationSystem:
def __init__(self):
self.deepl_translator = deepl.Translator(auth_key="YOUR_DEEPL_KEY")
self.google_translator = Translator()
self.openai_client = openai.OpenAI(api_key="YOUR_OPENAI_KEY")
def translate_document(self, text, source_lang, target_lang, document_type):
        """文書タイプに応じた最適な翻訳手法を選択"""
if document_type == "marketing":
return self.marketing_translation(text, source_lang, target_lang)
elif document_type == "technical":
return self.technical_translation(text, source_lang, target_lang)
elif document_type == "legal":
return self.legal_translation(text, source_lang, target_lang)
else:
return self.general_translation(text, source_lang, target_lang)
def marketing_translation(self, text, source_lang, target_lang):
        """マーケティング文書の翻訳（DeepL + ChatGPT）"""
# 1. DeepLで基本翻訳
deepl_result = self.deepl_translator.translate_text(
text, 
source_lang=source_lang, 
target_lang=target_lang
)
# 2. ChatGPTで文化的適応
cultural_adaptation_prompt = f"""
        以下の翻訳文を、{target_lang}の文化に適した自然な表現に調整してください。
        マーケティング効果を最大化するよう、現地の慣用表現や価値観を考慮してください。
        原文: {text}
        翻訳文: {deepl_result.text}
        """
response = self.openai_client.chat.completions.create(
model="gpt-4",
messages=[{"role": "user", "content": cultural_adaptation_prompt}]
)
return response.choices[0].message.content
def technical_translation(self, text, source_lang, target_lang):
        """技術文書の翻訳（ChatGPT中心）"""
technical_prompt = f"""
        以下の技術文書を{source_lang}から{target_lang}に翻訳してください。
        専門用語の正確性を重視し、技術的な文脈を保持してください。
        翻訳対象: {text}
        """
response = self.openai_client.chat.completions.create(
model="gpt-4",
messages=[{"role": "user", "content": technical_prompt}]
)
return response.choices[0].message.content

class MultilingualContentManager:
def __init__(self):
self.translation_system = HybridTranslationSystem()
self.terminology_db = TerminologyDatabase()
self.quality_checker = TranslationQualityChecker()
def process_content_update(self, content_id, updated_text, source_language):
        """コンテンツ更新時の自動多言語化処理"""
# 対象言語リスト取得
target_languages = self.get_target_languages(content_id)
results = {}
for target_lang in target_languages:
try:
# 翻訳実行
translated_text = self.translation_system.translate_document(
updated_text, 
source_language, 
target_lang, 
self.get_document_type(content_id)
)
# 用語集チェック・適用
corrected_text = self.terminology_db.apply_terminology(
translated_text, target_lang
)
# 品質チェック
quality_score = self.quality_checker.evaluate(
updated_text, corrected_text, source_language, target_lang
)
results[target_lang] = {
'translated_text': corrected_text,
'quality_score': quality_score,
'status': 'completed' if quality_score > 0.8 else 'review_required'
}
except Exception as e:
results[target_lang] = {
'error': str(e),
'status': 'failed'
}
return results

class MultilingualSupportSystem:
def __init__(self):
self.translator = HybridTranslationSystem()
self.support_knowledge_base = SupportKnowledgeBase()
def handle_customer_inquiry(self, message, customer_language):
        """多言語カスタマーサポート処理"""
# 1. 顧客メッセージを日本語に翻訳
japanese_message = self.translator.translate_document(
message, customer_language, "ja", "support"
)
# 2. ナレッジベースから回答候補を検索
answer_candidates = self.support_knowledge_base.search(japanese_message)
# 3. 最適な回答を選択・生成
japanese_response = self.generate_response(
japanese_message, answer_candidates
)
# 4. 顧客の言語に翻訳して返答
localized_response = self.translator.translate_document(
japanese_response, "ja", customer_language, "support"
)
return {
'response': localized_response,
'confidence': self.calculate_confidence(answer_candidates),
'escalation_required': self.should_escalate(japanese_message)
}

import nltk
from textstat import flesch_reading_ease
from googletrans import Translator
class TranslationQualityEvaluator:
def __init__(self):
self.reference_translator = Translator()
def evaluate_translation(self, source_text, translated_text, 
source_lang, target_lang):
        """翻訳品質の総合評価"""
scores = {}
# 1. 言語的正確性評価
scores['linguistic_accuracy'] = self.evaluate_linguistic_accuracy(
translated_text, target_lang
)
# 2. 意味的正確性評価
scores['semantic_accuracy'] = self.evaluate_semantic_accuracy(
source_text, translated_text, source_lang, target_lang
)
# 3. 読みやすさ評価
scores['readability'] = self.evaluate_readability(
translated_text, target_lang
)
# 総合スコア計算
overall_score = (
scores['linguistic_accuracy'] * 0.4 +
scores['semantic_accuracy'] * 0.4 +
scores['readability'] * 0.2
)
return {
'overall_score': overall_score,
'detailed_scores': scores,
'quality_level': self.get_quality_level(overall_score)
}
def evaluate_linguistic_accuracy(self, text, language):
        """言語的正確性の評価"""
# 文法チェック、語彙適切性チェック等
# 実装は言語によって異なる
return 0.85  # サンプル値
def evaluate_semantic_accuracy(self, source, target, src_lang, tgt_lang):
        """意味的正確性の評価"""
# 逆翻訳による意味保持度チェック
back_translation = self.reference_translator.translate(
target, src=tgt_lang, dest=src_lang
).text
# 原文と逆翻訳の類似度計算
similarity = self.calculate_similarity(source, back_translation)
return similarity
def get_quality_level(self, score):
        """品質レベルの判定"""
if score >= 0.9:
return "excellent"
elif score >= 0.8:
return "good"
elif score >= 0.7:
return "acceptable"
else:
return "needs_improvement"