[16:06:26] <muhdusman>	 bmansurov: I am having doubt with the flow.
[16:07:13] <muhdusman>	 So we first get the ids from the first API call or we directly pass the seed?
[16:07:42] <bmansurov>	 muhdusman: The way I see it is that there will be only one API call.
[16:08:01] <bmansurov>	 muhdusman: we'll just get rid of the WDQS API call and adjust the MediaWiki API call.
[16:09:01] <muhdusman>	 bmansurov: So we just pass the seed as morelike:<seed> and then sort the results based on index?
[16:09:19] <bmansurov>	 muhdusman: yes, filter the results.
[16:09:27] <bmansurov>	 muhdusman: we'd make the changes here: https://github.com/wikimedia/mediawiki-services-recommendation-api/blob/master/lib/article.creation.translation.js#L78
[16:09:56] <bmansurov>	 muhdusman: if you can adjust the query parameters to look like the API URL I shared, you'll see that there won't be a need to call WDQS
[16:10:34] <bmansurov>	 muhdusman: and then you'd change the filter function which is being called on the results of the above link: https://github.com/wikimedia/mediawiki-services-recommendation-api/blob/master/lib/article.creation.translation.js#L221
[16:10:55] <bmansurov>	 muhdusman: you'll have to make a similar change in https://github.com/wikimedia/mediawiki-services-recommendation-api/blob/master/lib/article.creation.translation.js#L119
[16:11:42] <bmansurov>	 In the first case the goal is to get similar articles that don't exist in the source language and that's not a disambiguation page.
[16:12:14] <muhdusman>	 bmansurov: So for getArticleBySeed I use the same API as the link you shared and make changes. 
[16:12:22] <bmansurov>	 muhdusman: yes
[16:13:02] <muhdusman>	 For: getArticlesByPageviews I change generator from search to mostviewed. 
[16:13:55] <muhdusman>	 bmansurov: Then the flow is completely like morelike API.
[16:14:09] <bmansurov>	 yes for getArticlesByPageviews
[16:14:27] <bmansurov>	 sure, morelike is a replacement of this api
[16:14:32] <bmansurov>	 but we need to keep both for now
[16:15:10] <bmansurov>	 and morelike uses a machine learning trained model output
[16:15:38] <muhdusman>	 Oh okay. I wrote a python script to check with the URL you shared and the output we get from current API. They seem to be different. 
[16:16:03] <bmansurov>	 muhdusman: yes, they are expected to be different
[16:16:15] <bmansurov>	 muhdusman: hopefully the morelike API results are better
[16:16:48] <muhdusman>	 Oh, that got me fully confused. I thought I was doing something wrong. 
[16:17:15] <bmansurov>	 muhdusman: you're on the right track
[16:17:28] <muhdusman>	 bmansurov: I'll make the changes accordingly then. Thanks a lot!
[16:17:37] <bmansurov>	 muhdusman: np, good luck