search

claimskg / claimskg-extractor

Claim Review extractor for ClaimsKG
19 stars 13 forks source link

Update outdated sources (websites 2021) #5

Open Slashdacoda opened 3 years ago

Slashdacoda commented 3 years ago

WIP related to: https://github.com/claimskg/claimskg-extractor/issues/15

Gautamshahi commented 3 years ago

How do you use the labels from the fact-checking websites? Leadstories don;t give a direct label which can be mapped into four category(proposed by you)

Slashdacoda commented 3 years ago

Hi @Gautamshahi, I have forked the code. For more detailed questions, better contact @twktheainur.

According to the description of https://github.com/claimskg/claimskg-extractor#normalizing-truth-values-ratings-across-fact-checking-websites. the labels are divided into 4 areas. TRUE, FALSE, MIXTURE and OTHER

After my customization of snopes i took the rating from the site: https://www.snopes.com/fact-check-ratings/ https://github.com/Slashdacoda/claimskg-extractor/blob/6c7ccef926a561f06cd37d13878685a38b17cef0/claim_extractor/extractors/snopes.py#L135 Currently the restriction is not made. See here the result in column "rating_alternateName": https://github.com/Slashdacoda/claimskg-extractor/blob/snopes_2021_refactor/output_got.csv

To what extent the result is used. I can not currently estimate because I do not yet have the necessary knowledge about how the processing then takes place from the generator. Here also the following issue should be kept in mind: https://github.com/Slashdacoda/claimskg_generator/issues/2

Slashdacoda commented 3 years ago
Slashdacoda commented 3 years ago

@twktheainur, for the sake of completeness, can you help me? related to Vishvasnews: image

  1. I cant login (I dont have the credentials), is there some TAGME_API_KEY problems? or ist the service not working?
  2. After creating an account to setup a new API (how do I get a new API Key?), I figure out the GCUBE service private... image
  3. At the moment i have deactivated the tagme functionality. Should I enable the tagme service for all the other "outdated" sources I mentioned in the problem description?
twktheainur commented 3 years ago

Hi, sorrry, only seeing this now as it goes on my Gmail, that classifies those notifications away.

I hadn't undated the extractor to be in line with the others, in that entity extraction is done later and not inside individual extractors. The students who implemented it added that annotation step, but the right course of action is indeed to remove this functionality.

Best, Andon

On Mon, 31 May 2021, 12:29 Sascha Schüller, @.***> wrote:

@twktheainur https://github.com/twktheainur, for the sake of completeness, can you help me? related to Vishvasnews: [image: image] https://user-images.githubusercontent.com/22930947/120177855-210e1f80-c209-11eb-84f7-71118f51f163.png

  1. I cant login (I dont have the credentials), is there some TAGME_API_KEY problems? or ist the service not working?

  2. After creating an account to setup a new API (how do I get a new API Key?), I figure out the GCUBE service private... [image: image] https://user-images.githubusercontent.com/22930947/120178423-c88b5200-c209-11eb-99ae-61cc2bce1ba2.png

  3. At the moment i have deactivated the tagme functionality. Should I enable the tagme service for all the other "outdated" sources I mentioned in the problem description?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/claimskg/claimskg-extractor/issues/5#issuecomment-851393223, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAUBEG7RVMDAVU6EVGBKFDTQNQHPANCNFSM43IAHK5Q .

Slashdacoda commented 3 years ago

Related to: EU vs Disinfo -> 403 'Forbidden' https://github.com/claimskg/claimskg-extractor/issues/5#issuecomment-848725483

Access problem with euvsdisinfo still exists: image

Posible solution: