Hi there! My friend and I are working on a website called The Newsphere and we’re looking for others who share our interest in understanding current events to help us beta test our platform.
The Newsphere acts as a virtual map of world news, enabling users to see the news as the result of a complex web of relationships between people, places, organizations, and countries. You can watch our intro video on youtube here: https://www.youtube.com/watch?v=Wpwgat54KXg&t=3s and signup for the beta here: https://newsphere.org or keep reading to learn more about the Newsphere.
Basically, our model analyzes thousands of news articles every day, and uses natural language processing to extract the different entities in them, as well as identifying relationships between them. Over time we construct narratives, which the user can see illustrated on a graphical timeline. The goal here is to give context to current events, enabling users to view recent developments in light of past patterns of behavior.
If you’re interested in learning more, you can email us directly at [email protected] and we’ll be happy to get back to you. Again, you can sign up for our beta here: https://newsphere.org
I really like your timeline concept. I'd almost maybe double down on that, and move away from the graph model (where the geometric layout is not really grounded to an easily accessible mental model). I'm not sure what the ideal looks like, but my hunch is the timeline view might be at the center of it.
Agreed with this. If the premise of the service is that the news is clear, concise, and comprehensive, show me that with a clear and concise explanation of the product on the front page (and perhaps a comprehensive drill-down in the rest of the site). I care far more about the UX of the tool than the ML behind the scenes.
Sooner you can have a workable current affair as part of your front page, the better. You could even have subdomains or special features on major events, even a dedicated domain (China, impeachment, etc). Looks like one of those things best understood by clicking around. Best of luck.
Not sure if he'd have useful info or time to spare, but I think Austen from Lambda School had a news-related venture (Grasswire or something) in the past and could have feedback for you.
Thanks for the feedback! That's something we've been thinking of implementing, perhaps with a search feature as well. And we'll definitely try to reach out to Austen, especially with his success in the startup world. Are there any other people/ companies/ startups you'd suggest reaching out to for feedback? We've mainly talked to professors/ business leaders from Brandeis University and a few professionals in the news industry.
Looks like a cool project. I'll sign up for the beta.
How many news sites are you pulling content from? I'm curious about how, or if, you're getting content from news sites that have subscriptions - e.g. The New York Times. Do you store archives of information?
I know news articles can get updated or lost. It would be interesting to keep track. I can imagine using this or something like it when writing a book or an article and the author may want to understand relationships over time.
Thanks! Currently for our MVP just AP News, which we're getting from here: https://newsapi.org/. We're working on perfecting the site with one or two news sources first before expanding to more options. Definitely a good point about articles changing/ getting updated, something we'll be looking into how to address going forward.
I'm working on something very similar. I do also have built a feature similar to the timeline. (your version is much cooler.)
I had a question.
Mostly the title are clickbait and doesn't represent what actually news is and many news site are biased and distort the meaning by carefully selecting the words.
It's great to know other people are also trying to tackle this problem! I agree, the titles of news stories are often to grab one's attention rather than convey the true contents of a story, and we're still trying to solve this problem. Perhaps some feature that visualizes the mood/ topic of a story based on the full text, but it's a difficult problem to solve. We're open to suggestions though!
I wanted to do something very similar. This is amazing! May I ask, where do you get your news feed from? The sources I enquired at the time were quite pricey.
For sure, our beta is pretty bare bones but we'll refine it over the next few months, especially after we get feedback from our planned beta launch at the end of the month
Curious to see how you handle weighting/filtering of the sources, the whole "fake news" issue.
Not your primary concern, I'm sure, but I don't (normally) use Google, nor allow Google-tools into my browser. I am a small-but-vocal minority. There are good alternatives to Google's recaptcha, Google-docs registration, etc. Just sayin'.
Great point, we'll have to work on that going forward. We've been using AP News as our main source for the moment as it is fairly neutral, but we're looking for ways to handle different sources going forward. We went with google but are open to other suggestions for future forms
We are still working on our pricing model but right now we are looking at a doing freemium model. Free version would have some limitations (ex: a limited number of news sources that stories can be pulled from, so choosing 3 sources from list of options) but the average user would still get a ton out of the site. Premium version would be subscription based, giving users full range of features (ex: pulling from as many news sources as you wish). We would love feedback on any suggestions for pricing though, we'll be tinkering and tailoring it based on how the beta goes/ what changes we make to the site + idea
[+] [-] bmusto19|6 years ago|reply
Basically, our model analyzes thousands of news articles every day, and uses natural language processing to extract the different entities in them, as well as identifying relationships between them. Over time we construct narratives, which the user can see illustrated on a graphical timeline. The goal here is to give context to current events, enabling users to view recent developments in light of past patterns of behavior.
If you’re interested in learning more, you can email us directly at [email protected] and we’ll be happy to get back to you. Again, you can sign up for our beta here: https://newsphere.org
Thanks again!
[+] [-] chris_st|6 years ago|reply
[+] [-] breck|6 years ago|reply
I really like your timeline concept. I'd almost maybe double down on that, and move away from the graph model (where the geometric layout is not really grounded to an easily accessible mental model). I'm not sure what the ideal looks like, but my hunch is the timeline view might be at the center of it.
[+] [-] anigbrowl|6 years ago|reply
[+] [-] rijoja|6 years ago|reply
[+] [-] burnte|6 years ago|reply
[+] [-] charliepark|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
[+] [-] prawn|6 years ago|reply
Not sure if he'd have useful info or time to spare, but I think Austen from Lambda School had a news-related venture (Grasswire or something) in the past and could have feedback for you.
[+] [-] bmusto19|6 years ago|reply
[+] [-] ken|6 years ago|reply
All I get is the slowly-appearing stock clip art lower on the page, which doesn't really explain it at all.
[+] [-] bmusto19|6 years ago|reply
Additionally we have a one page PDF summary here: https://docs.google.com/document/d/1-D7MmRph8mRC9MUL55Vh5hNy...
[+] [-] MattPalmer1086|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
[+] [-] si1entstill|6 years ago|reply
[+] [-] joegahona|6 years ago|reply
[+] [-] juancampa|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
[+] [-] ALittleLight|6 years ago|reply
How many news sites are you pulling content from? I'm curious about how, or if, you're getting content from news sites that have subscriptions - e.g. The New York Times. Do you store archives of information?
I know news articles can get updated or lost. It would be interesting to keep track. I can imagine using this or something like it when writing a book or an article and the author may want to understand relationships over time.
[+] [-] bmusto19|6 years ago|reply
[+] [-] kburman|6 years ago|reply
I had a question. Mostly the title are clickbait and doesn't represent what actually news is and many news site are biased and distort the meaning by carefully selecting the words.
How do you handle it or have plan to do it?
[+] [-] bmusto19|6 years ago|reply
[+] [-] alixaxel|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
[+] [-] fouc|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
[+] [-] prepend|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply
Additionally you can check out our one page user guide PDF here: https://docs.google.com/document/d/1-D7MmRph8mRC9MUL55Vh5hNy...
The PDF also includes a link to a more in depth explanation of Newshpere
[+] [-] spiraldancing|6 years ago|reply
Curious to see how you handle weighting/filtering of the sources, the whole "fake news" issue.
Not your primary concern, I'm sure, but I don't (normally) use Google, nor allow Google-tools into my browser. I am a small-but-vocal minority. There are good alternatives to Google's recaptcha, Google-docs registration, etc. Just sayin'.
[+] [-] bmusto19|6 years ago|reply
[+] [-] JaimeThompson|6 years ago|reply
[+] [-] bmusto19|6 years ago|reply