Sentiment Analysis Tracks Public Reaction to Geopolitical Events

Risk Management, Sentiment Analysis, Social Media Analysis

Sentiment Analysis to track public reaction to geopolitical events

Governments and Other Organizations Need to Assess Public Opinion Regarding World Events

In order to craft effective and sound policies, Government needs a good understanding of public opinion. To achieve this, there must be an effective and timely way of monitoring public attitudes towards issues of public concern.

Monitoring attitudes is not a simple matter. The on-going Gaza war is a good illustration of the complexity of understanding the public’s attitudes both to the event itself and to associated events such as various countries’ response to it. The types of assessments needed include:

  • Overall levels of pro-Hamas and pro-Israel support both worldwide and in specific regions and countries
  • Public sentiment towards the Hamas October 7 attack both worldwide and in specific regions and countries
  • Attitudes toward the Israeli response and the way it is conducting the war
  • Attitudes towards U.S. support of Israel
  • Attitudes towards the U.S. policies towards the Middle East as a whole
  • etc.

The Russo-Ukrainian war is another example where similar assessments of public attitudes are useful. In particular, where Americans stand right now on the question of  continued aid to Ukraine is something that should be monitored.

Sentiment Analysis Is an AI technology than Can Help

Today by far the major source for understanding public attitudes towards geopolitical events is social media. The immense volume of information contained there is clearly beyond solely human means to evaluate.

What is needed is a technology that provides a fine-grained analysis at scale and in real time of the public’s reaction to geopolitical events and to Government responses. There is such a technology: Sentiment Analysis.

How Sentiment Analysis Works

Sentiment Analysis identifies people’s attitudes in unstructured text data such as that found in social media. Basic versions of Sentiment Analysis simplistically identify sentiment-bearing words such as “great” or “awful” as positive or negative without understanding any of the surrounding context. It cannot associate a sentiment with the entity toward which the specific sentiment is expressed, such as a country name or a specific politician.

A more advanced version is called Entity-Based Sentiment Analysis, and it provides a detailed analysis of sentiment by identifying not only a sentiment but also the target of the sentiment and any associated entities.  For instance, Entity-Based Sentiment Analysis captures the following sentiments:

  • We will not forget the barbaric evilness of Hamas and the other extremists in Gaza (source)

Sentiment: negative

Predicate: barbaric evilness

Object: Hamas

Object: extremists

  • I want Israel to understand that they can scream ” hamas monsters ”  from here till eternity and that will still never, could never, justify Israel’s CRUEL, INSANE, FASCIST SLAUGHTER OF INNOCENTS (source)

Sentiment: negative

Predicate: cruel

Predicate: insane

Predicate: fascist

Object: slaughter

Associated entity: Israel

  • It would send a strong signal to the Kremlin, and the Russian population, if players opposing Russia’s brutal and illegal war against Ukraine, choose not to accept the awards. (source)

Sentiment: negative

Predicate: brutal

Object: war

Associated entity: Russia

This capability allows us to capture multiple sentiments accurately when more than one sentiment is mentioned in the same post.

Advanced Entity-Based Sentiment Analysis also makes possible the aggregation of sentiment found in a multitude of social media posts. This is because it normalizes all the sentiments and entities it identifies. Human language can express the same sentiment in many different ways:

    • China has always been an authoritarian regime
    • China’s authoritarian regime
    • an authoritarian Chinese regime

These three statements mean the same thing. Advanced Entity-Based Sentiment Analysis converts each of these sentiments to the same structured representation of the sentiment, such as:

Sentiment: negative

Predicate: authoritarian

Object: regime

Associated entity: China

This normalization allows sentiments to be aggregated, giving organizations a statistically based view of public attitudes.

The time stamps of the social media posts can be used to provide a view of how public attitudes change over time. This supports a dashboard-style display with specific quantitative evidence of how public attitudes evolve.


Based on AI techniques, Sentiment Analysis is an effective tool for monitoring public attitudes. It gives a government or other organizations real-time access to what the public thinks about geopolitical events.