What is an N-gram Model?

Simplified N-gram Explanation

An N-gram models use (n-1) words of context in the sentence to predict the next word.

E.g., let’s say we are using a trigram model (uses 2 words for context)

If you type - “I am”

The message app might suggest - “going”, “happy”, “fine”, etc.

---

Real Example with Bigrams:

Given a sentence: "I love to eat ice cream and I love to eat pizza."

The bigrams are:

• I love

• love to

• to eat

• eat ice

• ice cream

• cream and

• and I

• I love

• love to

• to eat

• eat pizza

From this, the model learns:

• After “I”, “love” always comes next → 100% probability

• After “eat”, both “ice” and “pizza” appear → 50% probability each

Using these probabilities, the app predicts the next word.

---

Application:

• Spell or grammar correction: Needs only a small context → bigram model is often enough.

• Plagiarism detection: Needs larger context → higher-order n-grams (like 5-grams or 6-grams) are better.