priyaradhakrishnan0/elden

elden's Issues

gold recall is low in TAC4NED mapping

Hi,

I download the TAC4NED candidate mapping and calculate the gold recall of each query in Test1.csv. However, the gold recall is only ~51.4% in this map.

The way I calculate gold recall is like this:

• treat url in "ENTITY" line as gold answer
• check whether it is in the following "CANDIDATE" lines

For example, I found all "Hennepin County" don't have gold answer in its candidate list

ENTITY	text:Hennepin County	qid:EL012291	docId:APW_ENG_20070810.0196.LDC2009T13	url:http://en.wikipedia.org/wiki/Hennepin_County,_Minnesota
CANDIDATE	url:http://en.wikipedia.org/wiki/Randy_Johnson_(council_member)
CANDIDATE	url:http://en.wikipedia.org/wiki/Hennepin_Avenue_Bridge
CANDIDATE	url:http://en.wikipedia.org/wiki/Hennepin_Township,_Putnam_County,_Illinois

ImportError: No module named HtmlTextExtraction

Hi, thanks for uploding code.

I came across this error, when running code for scraping.

% python2.7 WebScraping.py Traceback (most recent call last): File "WebScraping.py", line 13, in <module> from HtmlTextExtraction import getLinks_api_search,getLinks_cmu_search, getLinks_custom_search, extractDataFromLink, extractSentencesFromLink ImportError: No module named HtmlTextExtraction
Although mongodbClass.py was founded in github, but HtmlTextExtraction was not found.

Or, If you could write details for making corpus, I would appreciate it.

Ask for help

Hello, thank you very much for sharing your code, I have a question to ask you, how did you get the Wikipedia hyperlink graph data used in this article?I would appreciate it if you could help me solve my problems.