[HN Gopher] Show HN: Metaheads, a search engine for Facebook com...
       ___________________________________________________________________
        
       Show HN: Metaheads, a search engine for Facebook comments in the US
        
       Author : jawerty
       Score  : 30 points
       Date   : 2021-12-22 13:30 UTC (9 hours ago)
        
 (HTM) web link (metaheads.xyz)
 (TXT) w3m dump (metaheads.xyz)
        
       | ibdf wrote:
       | This is pretty neat. I would include an option to search a date
       | range, or at a minimum display the date of the post and let me
       | sort by date - this would make it easier to find posts that are
       | related to events.
        
         | jawerty wrote:
         | Thank you. The date on each is a priority for new features
        
       | altdataseller wrote:
       | This is Facebook posts posted on FB pages from external ppl, not
       | comments on other page posts. Misleading title.
        
       | DarkContinent wrote:
       | I wasn't able to find comments behind privacy settings--can you
       | add a login extension to make that possible?
        
         | jawerty wrote:
         | Interesting, yes I can do that
        
       | honkdaddy wrote:
       | Where are you getting the comments from? It seems to be a very,
       | very tiny subset.
        
         | jawerty wrote:
         | It's coming from public pages in US cities.
        
       | Karawebnetwork wrote:
       | This needs to take spaces and sentences into account. For
       | example, "hacker news" will find results about regular news and
       | no comments have the word "hacker" in them.
        
         | jawerty wrote:
         | Thanks I'll add this to the feature list.
        
           | chris_l wrote:
           | More like the bug list!
        
       | breakfastduck wrote:
       | Yeah this just doesn't get anything for most terms I've searched
       | where you'd expect thousands if not millions of results. What
       | exactly IS this searching?
        
       | jawerty wrote:
       | Hi,
       | 
       | Just to clarify a few things. This search only works for cities
       | in the US.
        
       | encodedstephen wrote:
       | "Computer vision" doesn't seem to bring back any relevant
       | comments.
        
       | Leary wrote:
       | Great project! One thing I'm sure a lot of people would find
       | useful is a "top comments" page that shows what the most popular/
       | trending topics are.
       | 
       | Or is it possible to find what sort of news articles are the most
       | popular? I often hear both side of the political spectrum saying
       | Facebook is biased. Would be cool to see if it was true.
        
         | jawerty wrote:
         | Thanks for the note. Funny enough I'm working on some sentiment
         | analysis in another project on bias. parallel thinking
        
       | Freak_NL wrote:
       | I don't think it really works. A search for 'Amsterdam' returns
       | only one comment.
        
         | istinetz wrote:
         | 0 hits for Rammstein and Romania. Perhaps OP just scraped
         | something like a 1000 comments?
         | 
         | Also, location filters seem to be hardcoded for US and Canada.
        
         | jawerty wrote:
         | It only works for the Us
        
           | hansel_der wrote:
           | i feel like this should be in the title in order for the rest
           | of humanity to quickly skip it. like an date tag "oh this is
           | from 2010, nevermind".
           | 
           | then again, paywalled sites _grml_
        
             | jawerty wrote:
             | Thanks for the feedback just added
        
       | jayolden wrote:
       | This doesn't work for local languages or country-specific
       | comments, which destroys the purpose of having a search engine
       | for FB comments.
        
       | doubleorseven wrote:
       | so if i'm not from or don't care about US pages, can you at least
       | share some info about what you do there? I'm guessing GQL but I'm
       | just guessing.
        
         | jawerty wrote:
         | Absolutely, I algorithmically fetch prominent public news/media
         | pages from a list of us cities of at least a few thousand and
         | get the comments from each page.
        
       | matt123456789 wrote:
       | Wow, super responsive. Great work. I don't want to ask how you
       | scraped these, but can you talk about your software (solr?) and
       | compute resources?
        
         | jawerty wrote:
         | Mongodb aggregation pipeline working wonders.
        
       ___________________________________________________________________
       (page generated 2021-12-22 23:02 UTC)