Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Identify car crash editorial anti-patterns using NLP

visionzeroreporting.com

Post reply on HN
Print view View original

41–50 of 213 posts

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#41

Post by ihaveajob »

dash2 wrote: post #35

I'm afraid I find this somewhere between creepy and ridiculous. On the ridiculous side: the idea that calling a car accident an "accident" is somehow wrong - as if, because I understand that what happened was an accident, I am somehow incapable of thinking "hey, maybe we need better road safety legislation". On the creepy side: the general approach which is, instead of making public arguments about why cars are bad,…

The problem with using "accident" as the default term is that it preempts any evidence-finding, implying that it was nobody's fault. If someone shoots their gun randomly and someone gets hurt by just being there, nobody will call it an "accident", but somehow if the weapon is a car, the driver gets the benefit of the doubt regardless of the facts.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#42

Post by zip1234 »

JackFr wrote: post #30
whimsicalism wrote: post #20

Earlier quoted context omitted.

Yes, I do not think you bear responsibility if you collapse and someone strikes you with their car and then flees the scene.

Don’t be obtuse. A person collapses into the street and is struck by oncoming traffic - no one is to blame. It is an accident.

Perhaps if she is struck then the drivers were going too fast? Maybe the speed limit should be lowered. Perhaps the street could be designed in a safer way to make this less likely to happen. Calling everything an accident is ignoring the actual causes.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#43

Post by kwhitefoot »

bluGill wrote: post #26
dymk wrote: post #9

This is neat, but something bugs me about the framing of the intended goal versus how you propose to get there. If the intent is to raise public awareness and to put pressure on leadership to make roads generally safer, why do almost all the corrections follow a pattern of shifting blame from the individual pedestrian onto the individual driver? Many of the suggested fixes are just their own form of counterfactual -…

One way to make roads safer is to make better drivers. There is a lot of training and continuing education that can be done to make better drivers. It is far too easy to drive, and once you get your license it is very hard to lose it even as your body degrades to the point you cannot see. People think of bad driving tickets (mostly speeding) as just a price you pay to drive. I've never heard of someone getting a tail…

> very hard to lose it even as your body degrades to the point you cannot see.

People bring this up often. As far as I can tell such people are not heavily represented in car crash statistics.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#44

Post by chiefofgxbxl »

dymk wrote: post #9

This is neat, but something bugs me about the framing of the intended goal versus how you propose to get there. If the intent is to raise public awareness and to put pressure on leadership to make roads generally safer, why do almost all the corrections follow a pattern of shifting blame from the individual pedestrian onto the individual driver? Many of the suggested fixes are just their own form of counterfactual -…

This is a great question and I agree with you entirely.

Ultimately, Framing (thematic vs. episodic) is the most important consideration for these articles. It's important for articles to discuss road design, safety initiatives, etc. in order to equip readers with the knowledge necessary to get governments to act.

However, Framing is also the most difficult pattern to detect simply because there are so many different thematic elements you can discuss, and many different ways to say each one. Here are a few examples:

- This is the 3rd pedestrian fatality this year.

- Residents have long complained about people racing down their street.

- Washington St. has had a long lasting problem with weather-related crashes.

- According to the NHTSA, there are X crashes per year related to prescription drug-induced drowsy driving.

- Street width directly correlates with driving speed, which in turn increases the time needed to react and brake.

I'm relying on manually annotated articles (700+ so far) and a custom-trained spaCy model to detect these elements, and the accuracy of that model depends on having enough training data. Since thematic elements are so rare in real-life articles, I'm having a difficult time providing enough training data. Hence why the tool emphasizes the other editorial anti-patterns, for now, but this will change once the Framing reporter gets better.

Anyone have any suggestions for augmenting my training set for thematic elements?

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#45

Post by markstos »

Here’s why the reframing matters:

Accidents are “oops” that are to accepted as fact of life. The no reason to change the system. Accidents happen.

Crashes are not accidents. They have a cause and the cause can addressed. Crashes can be prevented.

So the reframing shifts from talking about a system we should accept how it is to one that can be improved.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#46

Post by karaterobot »

bluGill wrote: post #26
dymk wrote: post #9

This is neat, but something bugs me about the framing of the intended goal versus how you propose to get there. If the intent is to raise public awareness and to put pressure on leadership to make roads generally safer, why do almost all the corrections follow a pattern of shifting blame from the individual pedestrian onto the individual driver? Many of the suggested fixes are just their own form of counterfactual -…

One way to make roads safer is to make better drivers. There is a lot of training and continuing education that can be done to make better drivers. It is far too easy to drive, and once you get your license it is very hard to lose it even as your body degrades to the point you cannot see. People think of bad driving tickets (mostly speeding) as just a price you pay to drive. I've never heard of someone getting a tail…

Let's say all that is true. How does using a computer program to reword all traffic accident reports as being caused by drivers solve this problem? Doesn't it just make people trust accident reports less?

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#48

Post by chiefofgxbxl »

Spivak wrote: post #40
ecshafer wrote: post #31

Earlier quoted context omitted.

If someone collapses into the street, the driver should be able to stop in time, unless they are going to fast. Who I can think to blame here: * The auto manufacturer for not having sensors to detect this. * The city for making car dependent areas * The city for making roads that have cars drive so fast by pedestrians, that the pedestrian can die if hit * The driver for not driving more reasonably * The city for not…

Yeah, this isn't an act of god that no one could have possibly foreseen. This is system designed by humans that we have 100% control over.

This is the key point of the research and the tool I built, and clearly I still have some work to do on making this clear on my site:

The objective is not to shift blame from one party to the other, even if its shifting from VRUs to drivers. The important thing is that we emphasize that all car crashes have a cause and known solutions! Nearly all news articles miss this point, and tend to place blame (even inadvertently) on the parties involved rather than discuss how to prevent future incidents from happening.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#49

Post by karaterobot »

markstos wrote: post #45

Here’s why the reframing matters: Accidents are “oops” that are to accepted as fact of life. The no reason to change the system. Accidents happen. Crashes are not accidents. They have a cause and the cause can addressed. Crashes can be prevented. So the reframing shifts from talking about a system we should accept how it is to one that can be improved.

It would be inaccurate, unethical, and potentially libelous for any newspaper to construe the events of a potential criminal act (or any event, really) without doing any investigation to support that conclusion, and primarily in order to further an editorial agenda.

Re: Show HN: Identify car crash editorial anti-patterns using NLP

#50

Post by dml2135 »

PeterisP wrote: post #19
fermigier wrote: post #18

I like the general idea, and the use of NLP to implement it. But I am dubious about some of the principles behind it. - For example, I find it inaccurate to say "the driver hit the pedestrian", which to me suggests a collision between two people, not between a person and a vehicle. (Of course this does not apply to a phrase like "the vehicle fled the scene" - it is clear that it was the driver who fled the scene). Wh…

I find "the vehicle fled the scene" as a useful phrase that's used to distinguish this situation from the very different "the driver fled the scene" which is commonly used in the (not that rare) cases where the driver abandons the wrecked car after an accident, in some cases to hide that they were intoxicated.

Easy to have both -- "the driver fled the scene in their vehicle"
Post reply on HN