How to Detect Cloud Sentiment Anomalies with the Pulsebit API (Python)

We recently stumbled upon a striking anomaly: a 24h momentum spike of +0.375 in the sentiment surrounding the cloud sector. This spike isn't just a number; it reflects a significant shift in public sentiment that could impact decision-making. The momentum increase suggests that something noteworthy is brewing in the cloud space, and we need to dive deeper to understand it.

Without a robust pipeline that accounts for multilingual origins or dominant entities, you risk missing critical insights like this one. Imagine your model missing this anomaly by a mere few hours. The leading language in discussions around cloud technology might skew your results if you don't account for it properly, ultimately leading to a misinterpretation of public sentiment. In this case, that could mean missing out on crucial insights from emerging markets or non-English speaking regions.

Arabic coverage led by 4.2 hours. English at T+4.2h. Confidence scores: Arabic 0.82, Mandarin 0.68, English 0.41 Source: Pulsebit /sentiment_by_lang.

To capture such anomalies, we need to implement a solution that leverages our API effectively. Here’s how to catch that spike.

import requests

![Left: Python GET /news_semantic call for 'cloud'. Right: ret](https://pub-c3309ec893c24fb9ae292f229e1688a6.r2.dev/figures/g3_code_output_split_1773076432683.png)
*Left: Python GET /news_semantic call for 'cloud'. Right: returned JSON response structure (clusters: 0). Source: Pulsebit /news_semantic.*


# Define the parameters for the API request
topic = 'cloud'
score = +0.000
confidence = 0.87
momentum = +0.375

# Geographic origin filter - Note: No geo filter data returned
geo_filter = {
    "language": "en",  # Replace with actual language or country if available
    "topic": topic
}

![[DATA UNAVAILABLE: countries — verify /news_recent is return](https://pub-c3309ec893c24fb9ae292f229e1688a6.r2.dev/figures/g3_geo_output_1773076432782.png)
*[DATA UNAVAILABLE: countries — verify /news_recent is returning country/region values for topic: cloud]*


# This is where you would apply the geo filter if data was available
response = requests.get(f"https://api.pulsebit.com/data?topic={topic}&geo={geo_filter}")
data = response.json()

# Check if the expected data is present
if response.status_code == 200 and data:
    print("Geo-filtered data retrieved successfully.")
else:
    print("No geo filter data returned — verify /dataset/daily_dataset and /news_recent.")

Next, we want to analyze the sentiment narrative itself. For this, we’ll run a meta-sentiment moment using our API to score the cluster narrative.

# Input for meta-sentiment analysis
narrative_input = "Cloud narrative sentiment cluster analysis"

# POST request to analyze the sentiment of the narrative
response = requests.post("https://api.pulsebit.com/sentiment", json={"text": narrative_input})
meta_sentiment_data = response.json()

# Check the response
if response.status_code == 200:
    print("Meta-sentiment analysis successful:", meta_sentiment_data)
else:
    print("Failed to analyze meta-sentiment.")

With these two components, we can build some useful tools around this anomaly. Here are three specific builds we can create:

1. Geo-filtered Anomaly Detection: Set a threshold for momentum spikes, e.g., momentum > +0.300, and filter results based on geographic language. This could help you identify trends emerging in specific regions faster.

2. Meta-Sentiment Monitoring: Use the narrative sentiment analysis build to track how sentiment shifts over time. A threshold like confidence > 0.80 can be set to trigger alerts when narratives shift significantly, allowing for quick responses.

3. Composite Signals: Combine the geo-filtered data and the meta-sentiment analysis into a composite signal. For instance, if momentum exceeds +0.375 and the sentiment score is below 0.1, you could trigger a deeper investigation into the underlying reasons.

If you’re ready to start leveraging these insights, head over to our documentation at pulsebit.lojenterprise.com/docs. You can copy-paste this code and have it running in under 10 minutes. Let’s catch those anomalies together!