Trapezium presents news articles sorted by vibe.
We perform sentiment analysis on a range of popular news articles, giving them a colour that represents the positive, negative, or neutral “vibe” of the article content.
NOTE: We got our api working and spitting out correct data less than an hour before due date so couldnt fix some relate bugs with integration with the UI. so the data that went live is just our test data [a single static article] :(
Works BEST in latest Chrome, but it works in all browsers minus IE and old safari [requires requestAnimationFrame]
Readability api, Facebook/Twitter/Linked in search api, Sentiment npm module Express js
Voting is now closed.
yeah unfortunately we ran out of time to iron out integration issues with the api. that single article is our test data. can I ask what mobile browser you were using? it should "work" on mobile - definitely needs work on mobile though.
Learning Objects Inc.
CompliSpace Technology Pty Ltd
Yeah we had to make a hard decision in the minutes before the deadline to either release a version with a slick ui but only one piece of data, or to release a version with live data but ui bugs from integration issues. We'll be releasing a version with live data after judging so be sure to check back after. Thanks for the vote!
Yeah its a mad shame we ran out of time to integrate the api which was completed too late to iron out integration issues. We'll be releasing a finished version after judging so be sure to check back :) Thanks for voting :)
Yeah totally fair call, the api works we just ran out of time to iron out integration issues. You can see it in a branch on the github - not that makes a difference in the context of the competition. Thanks for looking :) We'll be releasing a finished version after judging so be sure to check back then!
Thanks for the kind words! As mentioned in other comments here, we weren't able to ship with a fully functional backend which is why the news sources aren't as robust as we'd have liked. However, after the comp check back to see it as we'd originally intended.
Thanks for the feedback, yeah things get all sorts of real with the actual feeds turned on! Keep an eye out :)
We do sentiment analysis on the content of the article which basically uses a predefined list of weighted words to calculate an overall score by going through the nouns, verbs and adjectives in the document, giving them a score from the list then tallying them. There are already a few npm modules out there that do this using the AFINN wordlist which is widely believed to be one of the best wordlists out there for sentiment analysis. See https://github.com/thisandagain/sentiment for more information and to demo it yourself :) Thanks for the vote!
Thanks for the detailed explanation. :) Nice job.
We will definately be releasing a complete version. Hopefully the day after the judging has finished. Look forward to you trying it again then and giving feedback once again :) Thanks for the vote!
We were so very close to getting the version with live data up - in the last minutes it was a desicion between a buggy UI with live data or a clean UI with dummy data. We chose the latter. Upon reflection we should have had a mroe complete set of dummy data, but oh well. Thanks for the kind words :)
Harsh, dude. The point is to choose what to read based solely on the sentiment. News source bias to either extreme, and sensationalism (a grouping at either end of the spectrum with no neutral ground) are interesting elements, but by design they're things that emerge with use, and not part of the initial experience.
I should clarify my review: I like your idea, and it certainly has value added. My rating doesn't mean that your service is not useful. I reviewed your app based on my own experience reading news: I felt that a service that'd help me pick a news source (rather than specific articles from different sources) based on the overall sentiment would have been more useful to me. Is the service useful to others? It certainly is, as demonstrated by the other reviews. Or maybe I missed the point. The work done in 48 hours is certainly impressive, and I should up my "Innovation" rating. I'm sure that the final version you couldn't deploy is impressive. I apologize if I sounded harsh: I didn't mean to.
It's worth noting that we actually do have a card in our story board that adds a rating to each article that shows on aver age how negative/positive a news source is in general. There was also a card for the ability for usres to add their own news sources, or to scrape the news articles from their twitter/facebook feeds. We decided they would be a value add but that the core product was to display news articles based on sentiment first. The sky is the limit with content aggregators, we wanted to impliment our points of difference first :)
The new articles are picked based on how many times it has been shared, so "in theory" should be more likely to be interesting to anyone [ala other news aggreagtors]. Admitedly that aspect could have been protrayed better but in our user testing they found having 2 metrics displayed confusing.
As Jonny said, we were concious from the start about how we would rate not only the articles, but the source. The source is a bit harder to rate than the article though.
My hope would be that it is up to the reader to notice that SourceA has more negative votes than sourceB so the later is followed. I think shares is the only metric we can rely on though. Please let us know if there is another factor/library that you think we can use. I think we would all be happy implementing it.
Also, regarding the 3D stuff. It degrades in other browsers to 2D (I know because mine doesn't support it). I know everyone that saw the app before launch thought it was cool.
Please send any requests to the GitHub repo (hopefully given) and we'll definitely work to it. I can't imagine it will be too hard for us to add.
Personally I was surprised no one had used SA for news like this before - seemed like then most obvious use for it. We're looking forward to deploying the final version with live data - it was oh so close! Thanks for the vote :)
Yeah it is a shame we'll only be judged based on this current version. Keep an eye out for when the judging comes to a close we'll release the final "if only we had more time" version!
Our api wasnt spitting up the data we needed in time. We're as bummed as you are, but we have it ready now. Just needed those extra few hours - as always! Thanks for voting :)
Just to add, please check back after the competition because we are all planning on finishing this.
Thanks for the kind words! We have a pretty much fully working version ready to push once voting ends. Just needed that extra hour - but thats what its always like right?
Yes, thank you for the feedback. All of us are really keen to get the working version up there eventually. We just lost out in those last few hours.
The Node Firm
Yeah really sad that we couldn't ship with a working API, things got a little messy toward the end in that regard. I think this accurately shows the direction we wanted to take, and ultimately the app isn't that far off being complete. Just couldn't get it there for the deadline.
As Josh said, it got frantic towards the end, but please do check it again after the competition because we should have the working version done.
Thanks! Yeah not having the live data is a let down. It was so close! Another hour and it would have been fine - but thats always the way with hackathons.
SO close! After the judging has come to a close the next push will literally be the app as we'd intended to ship for this comp.