c25 home

Assignment 2 Comments


This is a brief summary of my most common critiques of the second assignment. Use these notes to


Comment

Meaning

not a question

You wrote down a topic, e.g., music, buying things, not a category of common questions. Related comment: "too broad."

too broad

Your category is way too broad to be doable or useful, e.g., "how to buy things." If you're too broad, there are (1) too many places that are relevant (2) way too many kinds and forms of questions people can ask. Leads to: "too many fixed content keys."

too vague

You gave a vague category of question, or a vague rule for modifying a question.

too many fixed content keys

Your rules added content keys that don't make sense for all the questions you claim to handle. Either because those keys are wrong, or because you have too broad a category. Follows from: "too broad."

inconsistent

The rules you gave don't match the example modifications you gave.

incomplete

The rules you gave don't explain all the example modifications you gave.

too narrow

Your category is too specific, e.g., "buy cheap computers." Leads to: "one site with all answers" and "needs more input variety."

one site with all answers

It seems like all you have to do is send your user to the same site everytime. Follows from: "too narrow."

needs more input variety

You had basically the same question in all three examples. Follows from: "too narrow."

ambiguous queries

Your example queries or questions were so short and/or vague that it's not clear how any assistant could help, no matter how intelligent.

parsing is non-trivial

You gave a rule that said "parse input to get ..." Parsing is non-trivial and part of your rules has to be how to understand a question.

what does "is known" mean

You wrote a rule that said "if ... is known...." Known to whom?

what improvement

What if any evidence do you have that your modified query is better than the original question?

how merge with health stats

You picked a statistics type of question. Therefore, you should think about how your statistics relate to the health statistics being done in class.


Comments? Send mail to Chris Riesbeck.