Azure AI Textanalytics Py

1---
2name: azure-ai-textanalytics-py
3description: |
4  Azure AI Text Analytics SDK for sentiment analysis, entity recognition, key phrases, language detection, PII, and healthcare NLP. Use for natural language processing on text.
5  Triggers: "text analytics", "sentiment analysis", "entity recognition", "key phrase", "PII detection", "TextAnalyticsClient".
6package: azure-ai-textanalytics
7---
8 
9# Azure AI Text Analytics SDK for Python
10 
11Client library for Azure AI Language service NLP capabilities including sentiment, entities, key phrases, and more.
12 
13## Installation
14 
15```bash
16pip install azure-ai-textanalytics
17```
18 
19## Environment Variables
20 
21```bash
22AZURE_LANGUAGE_ENDPOINT=https://<resource>.cognitiveservices.azure.com
23AZURE_LANGUAGE_KEY=<your-api-key>  # If using API key
24```
25 
26## Authentication
27 
28### API Key
29 
30```python
31import os
32from azure.core.credentials import AzureKeyCredential
33from azure.ai.textanalytics import TextAnalyticsClient
34 
35endpoint = os.environ["AZURE_LANGUAGE_ENDPOINT"]
36key = os.environ["AZURE_LANGUAGE_KEY"]
37 
38client = TextAnalyticsClient(endpoint, AzureKeyCredential(key))
39```
40 
41### Entra ID (Recommended)
42 
43```python
44from azure.ai.textanalytics import TextAnalyticsClient
45from azure.identity import DefaultAzureCredential
46 
47client = TextAnalyticsClient(
48    endpoint=os.environ["AZURE_LANGUAGE_ENDPOINT"],
49    credential=DefaultAzureCredential()
50)
51```
52 
53## Sentiment Analysis
54 
55```python
56documents = [
57    "I had a wonderful trip to Seattle last week!",
58    "The food was terrible and the service was slow."
59]
60 
61result = client.analyze_sentiment(documents, show_opinion_mining=True)
62 
63for doc in result:
64    if not doc.is_error:
65        print(f"Sentiment: {doc.sentiment}")
66        print(f"Scores: pos={doc.confidence_scores.positive:.2f}, "
67              f"neg={doc.confidence_scores.negative:.2f}, "
68              f"neu={doc.confidence_scores.neutral:.2f}")
69        
70        # Opinion mining (aspect-based sentiment)
71        for sentence in doc.sentences:
72            for opinion in sentence.mined_opinions:
73                target = opinion.target
74                print(f"  Target: '{target.text}' - {target.sentiment}")
75                for assessment in opinion.assessments:
76                    print(f"    Assessment: '{assessment.text}' - {assessment.sentiment}")
77```
78 
79## Entity Recognition
80 
81```python
82documents = ["Microsoft was founded by Bill Gates and Paul Allen in Albuquerque."]
83 
84result = client.recognize_entities(documents)
85 
86for doc in result:
87    if not doc.is_error:
88        for entity in doc.entities:
89            print(f"Entity: {entity.text}")
90            print(f"  Category: {entity.category}")
91            print(f"  Subcategory: {entity.subcategory}")
92            print(f"  Confidence: {entity.confidence_score:.2f}")
93```
94 
95## PII Detection
96 
97```python
98documents = ["My SSN is 123-45-6789 and my email is john@example.com"]
99 
100result = client.recognize_pii_entities(documents)
101 
102for doc in result:
103    if not doc.is_error:
104        print(f"Redacted: {doc.redacted_text}")
105        for entity in doc.entities:
106            print(f"PII: {entity.text} ({entity.category})")
107```
108 
109## Key Phrase Extraction
110 
111```python
112documents = ["Azure AI provides powerful machine learning capabilities for developers."]
113 
114result = client.extract_key_phrases(documents)
115 
116for doc in result:
117    if not doc.is_error:
118        print(f"Key phrases: {doc.key_phrases}")
119```
120 
121## Language Detection
122 
123```python
124documents = ["Ce document est en francais.", "This is written in English."]
125 
126result = client.detect_language(documents)
127 
128for doc in result:
129    if not doc.is_error:
130        print(f"Language: {doc.primary_language.name} ({doc.primary_language.iso6391_name})")
131        print(f"Confidence: {doc.primary_language.confidence_score:.2f}")
132```
133 
134## Healthcare Text Analytics
135 
136```python
137documents = ["Patient has diabetes and was prescribed metformin 500mg twice daily."]
138 
139poller = client.begin_analyze_healthcare_entities(documents)
140result = poller.result()
141 
142for doc in result:
143    if not doc.is_error:
144        for entity in doc.entities:
145            print(f"Entity: {entity.text}")
146            print(f"  Category: {entity.category}")
147            print(f"  Normalized: {entity.normalized_text}")
148            
149            # Entity links (UMLS, etc.)
150            for link in entity.data_sources:
151                print(f"  Link: {link.name} - {link.entity_id}")
152```
153 
154## Multiple Analysis (Batch)
155 
156```python
157from azure.ai.textanalytics import (
158    RecognizeEntitiesAction,
159    ExtractKeyPhrasesAction,
160    AnalyzeSentimentAction
161)
162 
163documents = ["Microsoft announced new Azure AI features at Build conference."]
164 
165poller = client.begin_analyze_actions(
166    documents,
167    actions=[
168        RecognizeEntitiesAction(),
169        ExtractKeyPhrasesAction(),
170        AnalyzeSentimentAction()
171    ]
172)
173 
174results = poller.result()
175for doc_results in results:
176    for result in doc_results:
177        if result.kind == "EntityRecognition":
178            print(f"Entities: {[e.text for e in result.entities]}")
179        elif result.kind == "KeyPhraseExtraction":
180            print(f"Key phrases: {result.key_phrases}")
181        elif result.kind == "SentimentAnalysis":
182            print(f"Sentiment: {result.sentiment}")
183```
184 
185## Async Client
186 
187```python
188from azure.ai.textanalytics.aio import TextAnalyticsClient
189from azure.identity.aio import DefaultAzureCredential
190 
191async def analyze():
192    async with TextAnalyticsClient(
193        endpoint=endpoint,
194        credential=DefaultAzureCredential()
195    ) as client:
196        result = await client.analyze_sentiment(documents)
197        # Process results...
198```
199 
200## Client Types
201 
202| Client | Purpose |
203|--------|---------|
204| `TextAnalyticsClient` | All text analytics operations |
205| `TextAnalyticsClient` (aio) | Async version |
206 
207## Available Operations
208 
209| Method | Description |
210|--------|-------------|
211| `analyze_sentiment` | Sentiment analysis with opinion mining |
212| `recognize_entities` | Named entity recognition |
213| `recognize_pii_entities` | PII detection and redaction |
214| `recognize_linked_entities` | Entity linking to Wikipedia |
215| `extract_key_phrases` | Key phrase extraction |
216| `detect_language` | Language detection |
217| `begin_analyze_healthcare_entities` | Healthcare NLP (long-running) |
218| `begin_analyze_actions` | Multiple analyses in batch |
219 
220## Best Practices
221 
2221. **Use batch operations** for multiple documents (up to 10 per request)
2232. **Enable opinion mining** for detailed aspect-based sentiment
2243. **Use async client** for high-throughput scenarios
2254. **Handle document errors** — results list may contain errors for some docs
2265. **Specify language** when known to improve accuracy
2276. **Use context manager** or close client explicitly
228