HW$\_1$ Lexicon$-$based Text Mining

With Whom Do People Spend Happy Moments?

Data

HappyDB is a corpus of $100,000+$ crowd$-$sourced descriptions of happy moments. HappyDB is

available on GitHub $[1].$ The cleaned data set is located at $[2].$

Task $(10$ points in total; four subtasks, $25\%$ each$)$

$(1)$ HappyDB provided a $$people dictionary$[3],$ which is a lexicon of common social

relationships. Use this lexicon to find the top three social relationships mentioned in

happy moments, e$.$g$.$ spouse, parents, children, friends, or someone else.

$(2)$ Use your world knowledge to assess the strengths and weaknesses of the $$people

dictionary$$ in terms of answering the question $$with whom do people spend happy

moments?$$ You can define $$people$,$strength$$ and $$weakness$$ in your own way here.

If you think this dictionary is already perfect, you can articulate your argument and skip

task #$3$ instead.

$(3)$ Modify the $$people dictionary$$ to fix the weaknesses that you have identified. Use the

revised lexicon to redo task #$1.$

$(4)$ Extract five context words before and after the most mentioned people, sort by frequency,

and explore patterns in the $100$ most frequent context words, for example, any words

indicating activities that people do in happy moments?

Submission

Write a Python script in Jupyter Notebook and submit your code file named

$$HW$1\_$firstname$\_$lastname.ipynb$.$ The file should include your code and your explanations in

comments. Make sure to include your name in both filename and comment on top of your code.

Grading criteria:

$-$ Accuracy and reproducibility in analysis

$-$ Clarity in explaining the analytical process and result