Predict Next Word from Text Corpus | Dataford Interview Questions

Given a text corpus, clean it and implement logic to predict the most probable next word

Asked in the onsite_round_3 stage. NLP-focused technical round.

Implement predict_next_word(corpus, current_word). Convert corpus text to lowercase alphanumeric tokens, removing punctuation. Count each word that immediately follows current_word. Return the most frequent successor, breaking ties by lexicographically smaller word. Return None if the cleaned query is absent or has no successor.

Signature: def predict_next_word(corpus, current_word):

Example: "Apple builds great products. Apple builds secure products.", "apple" returns "builds". For "builds", both great and secure occur once, so return "great".

Constraints

1 <= corpus.length <= 50000
The corpus contains printable text and whitespace
current_word represents one word, possibly with surrounding punctuation or capitalization
Tokens consist of contiguous ASCII letters and digits
Return None when no successor is observed

Function Signature

def predict_next_word(corpus, current_word):

Your question is Predict Next Word from Text Corpus. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem