HNHacker News
TopNewBestAskShowJobs

miket

1,595 karma · joined August 13, 2009

Stanford AI Lab. Founder, Diffbot. http://miketung.com @mikektung
submissionscomments
miket··on Ask HN: Who is hiring? (August 2014)
Diffbot is hiring at Stanford/Palo Alto, CA

We're looking for experts in computer vision, distributed systems, natural language processing, and rendering (Webkit internals).

Come help us build the web of structured data.

http://diffbot.com

miket··on Deep Learning Image Classifier
This was pre-trained on ImageNet classes. You can find more information here: http://www.image-net.org
miket··on Show HN: DashNotes – Gather Notes directly to your new tab page
DashNotes looks great, if you'd like to use Diffbot to generate summaries of pinned links, I'd be happy to hook you up with some free calls. Hit me up at mike@diffbot.com
miket··on Show HN: SwiftHN, a Hacker News reader in Swift
Looks great, Thomas. If you'd like to use Diffbot [1] to generate clean content summaries of the links for mobile, ala premii, I'd be happy to hook you up with some calls. [1] http://diffbot.com
miket··on Google Alerts is now with Material design
What particular aspects, interactions in this app make you feel like it was rushed? How could it be improved?
miket··on Show HN: Fresh baked bread, delivered to your door via SMS
..and this service won't be available in your area any time soon, either.
miket··on Ask HN: Who is hiring? (February 2014)
Diffbot - Palo Alto, one block from the Stanford campus. We're applying computer vision and NLP to web pages to structure data for developers. Read more here: http://diffbot.com/company/

We're looking for full-time and interns for the summer. Write jobs@diffbot.com

miket··on Ted Nelson: It All Went Wrong at Xerox PARC
Is he sitting on a church pew?
miket··on Kippt (YC S12) Takes Its Enterprise Tool For Sharing Useful Content Out Of Beta
These guys having an amazing sense for design and product. Early paid user of Kippt and looking forward to this pivot.
miket··on Deeply Moving: Deep Learning for Sentiment Analysis
The sentence trees are parsed with the Stanford NLP Parser, available here: http://nlp.stanford.edu/software/lex-parser.shtml
miket··on Starbucks’ WiFi goes Google
Hope it's less flaky than the Google-provided wifi in Mountain View
miket··on Upton: A Web Scraping Framework
Instead of (or in addition to) manually creating scrapers, you can use Diffbot to automatically extract this type of information from news articles using computer vision: http://diffbot.com/products/automatic/article. It also allows you to create rules with a WYSIWYG editor: http://diffbot.com/products/custom/
miket··on Do_good.js
Before embarking on this project, you should check out the company that was founded 6 years ago that does the thing you described: http://www.pluraprocessing.com/
miket··on German Girl Turned Away at Border Due to Private Facebook Messages
Not at all unlikely, but routine. TechCrunch previously wrote about what these Facebook comprehensive account reports to these "lawful" requests look like: http://techcrunch.com/2012/04/06/heres-what-a-facebook-respo...
miket··on Unofficial documentation of the Tesla Model S REST API
This is great! Can't wait until they add steering, gas, and brake to the API.
miket··on Show HN: Built a Hacker News radio web app I can listen to
Keep up the great work longzheng! You guys rock.
miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
yes, its on the roadmap. a full photo api is next.
miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
Thanks. Yes, the page classifier still does miss, which is why it's still in beta status and not fully in production as with our other APIs. Would love to hear about your use case.
miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
The next API we'll train an API for is likely image/photo pages. Event pages are definitely on our roadmap, but we've still got work to do there. Could definitely use your help with training data and learning more about your use case.
miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
Glad you liked that! That infographic was made by the multi-talented Dan Mayer, who we highly recommend. Check out his portfolio at http://danmayer.com
miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
The most pressing problem that the page classifier solves is for the current community of developers using Diffbot. We only offer APIs at the moment for extracting information out of frontpages and article pages (see http://www.diffbot.com/our-apis/). While we'll be releasing more soon for the other page types that you see in the infographic chart, this limitation is a problem if, for example, you're passing in a photo page or recipe into the article API--you're not going to get an article back since its not appropriate. Having the technology to visually understand what the types of pages are allow us to route those requests.

Beyond that, it's useful for companies that perform analytics or use the information to do better search or auto-categorization.

miket··on Diffbot launches a web page classifier API: analyzes a day of Twitter
Wanted to thank the HN community for all your encouragement. I first released the Diffbot API as a "Show HN:" post last year (http://news.ycombinator.com/item?id=2310852). $2M+ and lots of hard work later, we're powering some of the largest destination sites out there like Stumbleupon and the new Digg.
miket··on Show HN: Readability-like API Using Machine Learning
I tried a couple wikipedia pages and it seemed to work ok. Can you email me an example?
miket··on Show HN: Readability-like API Using Machine Learning
I just timed that Huffington Post example on my computer and it took around 80ms to extract at 100% CPU. The rub though is that, if you type in a totally new URL on our demo page, most of the time is actually spent in downloading the page from the original source and its associated CSS files (which we use to do visual rendering).
miket··on Show HN: Readability-like API Using Machine Learning
In short, yes. The key innovation is that we've come up with a lossy, fixed-length representation of the visual features that we can use to do classification upon. I'll try to do a more detailed writeup on our blog when I find some time.

btw, our blog has no rss feed, but you can just use our RSS API :-) http://www.diffbot.com/api/rss/http:/www.diffbot.com/blog

miket··on Show HN: Readability-like API Using Machine Learning
It's fairly CPU intensive. Like many classification-based techniques, much of the computation is in constructing the features. For our case, this means we had to implement most of CSS to get the visual features of every element on the page.
miket··on Show HN: Readability-like API Using Machine Learning
Right now, the API just returns back the raw text for simplicity's sake, but it would be possible to make an option for returning a bit of HTML structure, which would address the problem of sections, inline images, tables, etc.

The combination of the two APIs is a great idea.

miket··on Show HN: Readability-like API Using Machine Learning
We have a separate API for non-article pages (It's called the Follow API). It's not well documented yet, but you can get an idea for it from this demo: http://www.diffbot.com/mobilizer which will turn any webpage into a mobile version. Try putting in http://techcrunch.com or your forum thread page.
miket··on Show HN: Readability-like API Using Machine Learning
I'd be happy to talk about the visual-based statistical classification technology in a follow-up blog post if there's interest.
miket··on Show HN: Readability-like API Using Machine Learning
Great suggestion, you can use the article API already to do this (providing the URL of the post and the 'tag' parameter), but maybe you are thinking of analyzing other types of content? Or perhaps the ability to POST your own text data instead of just the URL?
← PreviousPage 3 of 4Next →