Final Project: Decent Dialogues

Inspiration and Intention

As a foreigner, when I talk with people in English, it’s sometimes a little bit embarrassing. Like when people ask me:”How’s going?” I’m always not if I should answer like “good”/”fine”, or answer it with more details or emotions to show my respect – cuz in China, people never greet by asking questions.

So we are wondering if we can build up a specific structure for questions and answers, based on different people’s styles.

What We Do

  • Select the most frequently-used words quoted from various celebrities from;
  • Based on our source texts, re-construct the words and generate different questions;
  • Answer these special-style-questions with a fixed structure.

Techniques We Use

Full github repo here:

First, I use the python library Selenium for web browser automation and scrapping from To prepare for text manipulations, I save each celebrity’s quotes in a txt file.

For the text manipulation part, we basically use Spacy to extract part-of-speech tags and construct the new sentence in a certain grammar structure. We use five lists to store nouns, verbs, adjectives, adverbs, and pronouns. The structure of a question goes like this: “What if”+ adj + noun + verb + adv + adj + pron + verb + adv. And the answer is a simple sentence consists of adj + noun + verb + adj + noun.

To enable user queries for generating arbitrary celebrity quotes, we use Flask to host a web page.

Examples We’ve Got

Final Results

Leave a Reply

Your email address will not be published. Required fields are marked *