Demo Performing Text Processing and Analysis

Explore four fundamental NLP tasks—summarization, sentiment analysis, translation, and data formatting—using OpenAI’s GPT-3.5-Turbo model in a Jupyter notebook. We’ll define a helper function for chat completions and reuse it throughout.

import os
import openai

openai.api_key = os.getenv("OPENAI_API_KEY")
# Alternatively, set your key directly:
def get_word_completion(prompt):
    messages = [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user",   "content": prompt}
    ]
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=messages
    )
    return response.choices[0].message.content

1. Text Summarization

We’ll summarize Steve Jobs’ 2005 Stanford commencement address in two ways:

1. A ~500-word narrative
2. Bullet-point highlights

# Load the speech into `context`
context = """
[Steve Jobs’ 2005 Stanford Commencement Address full text goes here…]
"""

1.1 Full-Length (~500 words)

prompt = f"""
Create a summary capturing the main points and key details in 500 words based on the content delimited by triple backticks.
```{context}```
"""
response = get_word_completion(prompt)
print(response)
```text

### 1.2 Bullet-Point Summary

```python
prompt = f"""
Create a summary capturing the main points and key details as bullet points based on the content delimited by triple backticks.
```{context}```
"""
response = get_word_completion(prompt)
print(response)

Sample Bullet Output:

• Jobs shared three stories: connecting the dots, love and loss, and reflections on mortality.
• He credited calligraphy classes—taken after dropping out—for the Macintosh’s design.
• Being fired from Apple sparked his creativity at NeXT and Pixar.
• A cancer diagnosis taught him the value of time.
• He concluded with “Stay Hungry. Stay Foolish,” quoting The Whole Earth Catalog.

2. Sentiment Analysis

Label each movie review as positive, negative, or neutral.

context = """
1. If you sometimes like to go to the movies to have fun, Wasabi is a good place to start.
2. An idealistic love story that brings out the latent 15-year-old romantic in everyone.
3. The story loses its bite in a last-minute happy ending that's even less plausible than the rest of the picture.
"""

prompt = f"""
Analyze the sentiment of the reviews delimited in triple backticks.
First show the actual review and then add the sentiment – positive, negative, or neutral.
```{context}```
"""
response = get_word_completion(prompt)
print(response)
```text

**Expected Output**:

1. If you sometimes like to go to the movies to have fun, Wasabi is a good place to start.  
   Sentiment: Positive  
2. An idealistic love story that brings out the latent 15-year-old romantic in everyone.  
   Sentiment: Positive  
3. The story loses its bite in a last-minute happy ending that's even less plausible than the rest of the picture.  
   Sentiment: Negative  

---

## 3. Translation

Translate a French poem by Victor Hugo into English while preserving its poetic style.

```python
context = """
Demain, dès l’aube, à l’heure où blanchit la campagne,
Je partirai. Vois-tu, je sais que tu m’attends.
"""

prompt = f"""
Write an English poem based on the French poem delimited in triple backticks.
```{context}```
"""
response = get_word_completion(prompt)
print(response)

Sample Translation:

Tomorrow, at dawn's early light,
I shall depart, for I know you wait.
Through misty fields and silent night,
I press ahead despite my fate.
...

4. Data Formatting (JSON, XML, JSONL)

Convert a list of U.S. states and capitals into three structured formats.

context = """
1. Alabama - Montgomery
2. California - Sacramento
3. Florida - Tallahassee
4. Georgia - Atlanta
5. Illinois - Springfield
6. Massachusetts - Boston
7. New York - Albany
8. Texas - Austin
9. Pennsylvania - Harrisburg
10. Washington - Olympia
"""

prompt = f"""
From the content delimited in triple backticks, format it into JSON, XML, and JSONL.
```{context}```
"""
response = get_word_completion(prompt)
print(response)

Summary of Techniques

These examples demonstrate how GPT-3.5-Turbo can accelerate common text-processing workflows. Next up: word embeddings and similarity search.

Links and References

• OpenAI Chat Completion API
• Kubernetes Documentation
• Terraform Registry