duckduckgo / zeroclickinfo-longtail

DuckDuckGo Instant Answers based on full-text data

Home Page:http://duckduckhack.com/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Stack Overflow: IA showing inconsistent result

adityatandon007 opened this issue · comments

This IA is showing the same result for two different queries. Though the queries in itself aren't ambiguous. Both the queries are in fact quite opposite.

Steps to recreate:
Just type in the search bar these two queries

Here are the links to the queries:

query1: https://duckduckgo.com/?q=how+to+install+beautifulsoup+using+pip&t=canonical&ia=qa
query2: https://duckduckgo.com/?q=how+to+uninstall+beautifulsoup+using+pip&t=canonical&ia=qa

Here are the screen shots:

screenshot from 2017-02-07 20-48-42

screenshot from 2017-02-07 20-49-08

People to notify:

@GuiltyDolphin @pjhampton @gautamkrishnar


IA Page: http://duck.co/ia/view/stack_overflow
Maintainer: @DDG

@pjhampton is this how I should always open issue. Just wanted to ask 😄

Close enough @adityatandon007. Make sure you always read the Contributing.md of any repo. It doesn't look like this repo has predefined issue templates.

This happens because it shows "results for how to install beautifulsoup using pip". Searching how to "uninstall" beautifulsoup using pip shows the correct results.

I don't believe there's anything we can do to fix this. We're using SOLR to search for StackOverflow articles that are relevant and given that the query and result are only off by two letters, it appears to be very relevant, though semantically it is not.

/cc @zachthompson

Can we just use Python and beautifulsoup instead of it. Just like i had used in one of my project SoCLI - Stackoverflow command line interface

@gautamkrishnar I want to ask something about SoCLI. Since it is not regarding this issue I want to ask you separately. Ping me when you're on slack.

Can we just use Python and beautifulsoup instead of it.

No, that's not really an option. The database lookup is faster and allows for better caching on our side. The IA triggers far to much for us to use the StackOverflow API.