nxb

Nice, and has a built-in parser and annotation system too. Very nice.

See also: http://brat.nlplab.org/examples.html

Any others like this?

I've always liked the Elance / Freelancer escrow systems, in principal, and I use them when buying services. But those sites have way too much of a low-end feel for me to ask my own clients to pay me via those, for 5 or 6-figure jobs.

If this can be a high-end equivalent of those low-end escrow services, that will be nice.

That forking behavior is exactly why I hate the idea of companies expecting people to use Github to showcase their side projects.

And I really support the idea of side projects, preferably on something other than a default Github page though.

Shockingly, signature is often not actually required in the UK for changes to employment contracts. Sorry I can't find better sources at the moment, but this was recently ruled on again by the UK courts. No signature is required, simply remaining at the job is enough to constitute acceptance in the UK for employment contracts. The employers don't even to show any evidence that you were aware that the contract has been changed.

> Following on from this, a contract of employment does not need to be signed for it to be a binding agreement between the parties as acceptance of its terms can be either verbal or implied through conduct.

http://www.thisismoney.co.uk/money/experts/article-2715470/I...

Add to this, the UK courts are often upholding extremely unilateral employment contract clauses that give the employer the "right to change the contract without the employee's permission at any time." Extremely unilateral power. The UK courts see no problem with it at all. Quite insane.

Is a new signature required from the worker, in order to make the newly added non-compete binding, or not? This part seems confusing from that summary.

If the workers signed a new non-compete and then continued to work for the company, this ruling seems pretty obvious.

If this is something like adding in a non-compete to an existing employment agreement without needing to obtain a new signature (as the completely insane UK laws allow,) then this is very bad.

I, personally, would love to hear more about the hundreds of lower tier accelerators, and insight as to why they're doing so terribly. In many of the cases I've seen, the people running the accelerators just lay 100% of the blame on the entrepreneurs.

It's really sad, tens of thousands of founders wasting years of their career on the low-tier net-value-negative accelerators.

How bad could they be? Far worse than you'd imagine, as I've personally seen.

I made of these back in the first month that Omegle was launched. I think they started automatically detecting it and quarantining these types of bots pretty quickly, though.

One solution I can see now, is to use something like word2vec, or better yet, skip-thought vectors, to replace each sentence with synonymous sentences and avoid tripping their detector.

The particular case I'm thinking of, is not a one-off type of service at all. It's a long-tail ecommerce shopping advice app that generates affiliate & advertising commissions that are much higher than the cost per user acquisition. The same customers are coming back repeatedly, and buying different things each time.

To clarify, the problem is that most people are only talking about "organic retention" when they say "retention", and ignore any "paid retention".

So in my example, I can get a very high percentage of people to return if I continue to spend money. But if I stop spending money to acquire users, they mostly stop coming back.

You think it's rare? Consider where you actually spend most of your money online. Ecommerce, Uber, Airbnb, Fintech services, etc. Most of these you don't use every single day, as you would with Facebook or Google or a dating website. When weighted by money spent, the picture is very different.

It's not a constructed example. It's my company's exact situation.

Here are my problems with retention being the star metric:

What if you have a customer lifetime value far higher than your acquisition costs per customer, and a very wide addressable market, but low retention rates? E.g. I acquire users for $5 each and same-day convert them to $15 profit each, and they rarely come back unless I return some of the profits back into the funnel to acquire them again.

You may ask - Shouldn't we assume that that growth channel will quickly collapse? -- Well, what if it doesn't? Something is very wrong with the trend of assuming that retention is everything, particularly in the case where there's a high immediate return on investment for each acquired user.

My second problem with retention is the time-span. Consider an ecommerce site for basic household necessities that may be accessed 20 times per year by a 100% retention user that always uses your service to his problem that your service solves. That return frequency is so low though, that if you're growing at 40%+ month over month, your retention may appear to be near zero, even though your retention is near 100%! Most of your users are new users and only a tiny percent returning, all due to your high growth rate. The compounding growth combined with the long expected time between visits just makes 100% retention look like 0%.

- Burnout

- Founder <-> founder resentment

- Founder <-> investor resentment

Those 3 causes are far above everything else, in cases I've seen. Even higher than running out of money.

Next, try the taggers on a more realistic setting than the standard corpuses -- e.g. a product review that compares several products, and you'll instantly see how incredibly poor the current state of the art NER is.

Technology is really going to advance once we have anything that comes close to human level on NER and relation extraction. Kind of like self driving cars, the basic ideas have been around for decades, but performance in realistic adverse conditions remains awful for almost everywhere that it could theoretically be used.

I can't begin to guess as to why you'd feel that way.

Facebook runs one of the top machine learning research labs in the world. They're among the most clever in this domain, by every possible metric. They've written extensively about how they employ machine learning across their entire product line, and are adding more constantly. They've acquired numerous ML startups and world-class researchers.

https://research.facebook.com/

https://research.facebook.com/researchers/1543934539189348

Respectfully, you're absolutely wrong, and I'm suspicious as to why you'd be intentionally spreading misinformation that they don't use ML/AI to fill in the gaps in their information about their users.

- Deducing based on my median friends' ages.

- Easily deducing my approximate age from my Linkedin profile.

- Any of the numerous other private people data sources they have access to or have bought over the years. I'm sure my birthday and basic details are in some of those. It's impossible to get those all deleted once the information is out.

- Day and month are easy, because everyone wishes me happy birthday in the messages and wall posts. All that's remaining is the year, which could be easily deduced from Linkedin and other third party data sources. - Which I've mostly cleaned up now, until Facebook exposed it all again.

- I had a profile for a few months in 2005 before deleting it. Used a different computer, different location, different email and everything for this new one a decade later, but probably they've connected the identities. Looks like if you ever tell Facebook some private information, it's non-revokably theirs to expose to the world forever.

Facebook invests very heavily in machine learning / AI, and uses it heavily through out their platform. They collect and buy tons of data both on the site and from third parties. This deduction is trivial for them.

The bad part is not so much that they deduced these details about me though, it's that I specifically made it clear that I didn't want any age public, let alone the real age, and they exposed me anyway.

Facebook does the same with birthdays. I input my age as 90 when creating my account. They changed it to my real birthday somehow.

AND then they expose it publicly to anyone who looks up my profile, even when they aren't my friends.

AND this is all after me having disabled the "show birthday on timeline" and enabling every other privacy setting there was.

If there's one company using AI and data collection for evil, it's Facebook. Not to mention the years of silently disabling previously set privacy settings without getting the user's permission - e.g. the "show profile in search".

Insane. Facebook should be slammed hard by the courts.