Show HN: Who's Hiring? (postings mined from Twitter)
jobstractor.com
jobstractor.com
Some suggestions:
I'd like to see how old a tweet is. I don't have any idea how fresh some of this information is (without clicking on the user and searching through their history).
Related to that, I'd also like a link to the original tweet, so that I can potentially follow any conversation on that tweet.
It'd also be nice if any hyperlinks in the tweets were active links, rather than just text that I need to cut/paste.
Something that's potentially a little harder would be to let the user determine the radius that they want to see tweets for. I'm in Minneapolis, and I'm seeing some tweets for Iowa and Wisconsin. Those could be useful to me, but I'd more likely want to filter those to a smaller radius.
I think modifying the radius should be possible, I'm using mongodb for storage and I think it has some nice geospatial capabilities.
Thanks for the feedback.
I do feel that you have built something quite useful.
Nothing on there should be more than 6 days old which was a guess by me as to how much history made sense to show. Possibly that needs to be more recent even, or maybe ordering by date will just make it less of a surprise when you find one which is gone already.
Thanks.
A common complaint is that it's hard to find out about new openings because they're often pushed through your social networks.
Directly mining the social networks is a great man-I-should've-thought-about-that idea. I'm infinitely likelier to tweet about a job opening than to post it somewhere.
Is this useful? Or could it be with a tweak or two?
Thanks, Robin
Edit: I see that you can already change location. Well done!
Better filtering is high on the priorities. As for the dates on the tweets. I'm only holding 6 days worth at the moment, I think I'll break them up by Today/Yesterday/Older in future. Otherwise it'll be too easy to miss new tweets maybe not at your exact location. Better linking or a summary pane is on the list as well.
Thanks for taking the time to look and comment.
I recently had to do a similar thing for a site I threw together to get my sporting news in one location (http://sportszealots.com). It's very hacky code written very quickly (in ruby), but if you'd like it send me an email and I'll send you back what I have.
EDIT: Well damn, I didn't know about the handy utility mentioned below...that's definitely better solution!
Nice, simple idea - solid MVP. Good use case for text processing too because the domain of candidate phrases, patterns you are looking for is relatively small :)
Here are UX some suggestions:
-It will require some effort, but you should try to surface companies as tags if you can. Will be a popular feature...We did this at my last startup Tracked.com and it was a really popular. You can start by getting a list of companies via Crunchbase API and then use that set to do the entity extraction.
-Same as above, but for programming language keywords
-Show timestamps of Tweets (you've heard this a lot)
-Make links hypertext (you've heard this a lot)
Thanks,
@mileslennon
good stuff. you should always be tweaking.
best of luck
One suggestion - a hide and/or mark as spam button would be useful. I noticed a "work from home" spam job appear in the listings which stuck out compared to the other tweets. A hide button would be useful to make jobs that I've applied for/are not interested in.
The second tweet that comes up for me (Providence, RI) is one labeled "North America and Europe." I think local listings should be prioritized over broad ones.
Filters based on language/expertise. I see a lot of these tweets go into specifics already so this could probably be implemented easily.
As someone else mentioned, a link to the tweet would be good, and preservation of the hyperlinks in them so I can click directly on someone's job listing etc.
Perhaps you could also normalize the format of the locations, and group them under one heading. For example, most of the listings for Providence are Boston, and it's a list that looks like, "Boston", "Boston, MA", "Boston", "Boston", "Boston, MA", etc. A consistent format would be good for readability and having a new <h2> for each tweet is redundant.
Basically, you have a good crawler (I assume you're using one) and now you need to alter the design of your page to make it easy to explore.
thanks again.
Code runs in the browser then:
if (google.loader.ClientLocation != null) { loc = google.loader.ClientLocation.address.city; lat = google.loader.ClientLocation.latitude lng = google.loader.ClientLocation.longitude }
hope that helps.
(forgive the mention of this fact but:)
StyleOwner (SF) is hiring (today we're at the UC Berkeley Startup Fair) and we'll be at the Github party on Monday. email matt@styleowner.com for more details. Hiring for both frontend and backend positions. Ideal candidate will want to do 20% time toward open source work.
I'd consider something like this, but I'm not sure how it'd work. Would you just loook for people saying 'I'm looking for a job' or would it be something more complex based on what expertise you can infer from there tweets?
Thanks for the feedback.
Here's my $0.02:
- Typo in the about page : "As weel..."
- I'd change the 1st 2 paragraphs in the about section to "Jobs Tractor looks for people trying to hire developers on Twitter, but filters out tweets that are from recruitment agencies or jobs boards."
- Love the map view
- Does the search field allow you to filter by location, keyword, or both?
- Group by location maybe?
- I'd take the word "jobs" out of the logo - to me it looks cluttered
- Do you save the posts or is it just what the Twitter search API pulls up? Jobs might not get tweeted about more than once, so a posting could still be legitimate after a week or so
I think he stores them; the About page mentions the site uses MongoDB.
re the search field. It's just to change locaiton at the moment.
re saving the posts, I'm saving them but only for 6 days at the moment to keep the data pretty fresh.
Thanks for the feedback.
Also, The locations could be refined. I see a lot of them being listed under the header "Boston" when they really happen "King of Prussia, PA", Burlington MA, or "West Newburry, MA". Those locations are not Boston. They're close, but not quite (and in the case of the PA one not even close). There's some value to the grouping things under Boston but I would prefer granularity in location headers.
Definitely see this as being very helpful though.
e.g.:
"Tacoma cannery property to be repositioned: Pinnacle Foods Group is looking for a real estate developer who can ... http://t.co/UobgHb39
and
"Online entrepreneur pitches for funding for app ... - TasteBand.com: An world wide web entrepreneur is looking t... http://t.co/JKAnZdBk
Suggestions 1. People who dont look at the search bar at the top would have no idea a. why there isn't any job postings for their location b. How to search for jobs in a specific location.
2. Could be a good idea to get the browser location and search for that, to make the site instantly relevant.
Also, after changing location, nothing seems to happen for a couple of seconds. Some visual feedback that it's loading the location info would be nice.
Congrats I smiled, bookmarked, and did a search. I think you're onto a winner :)
Best of luck with the recruitment for those jobs.
Seems broken to me. Am I doing it wrong?
Thanks for letting me know. There's some improvements to be made as to how it fails wrt this.
hyperlink any links in the tweets.
Do you have plans to expand the data source beyond Twitter? Maybe with filters for users to select source(s)?
The map is pretty minimal/useless at the moment. There's a lot I can do to improve it.
Thanks
The map would be for location search, the list for details. It'd be cool if you could filter the list view. Of course, it'd be great if you didn't have to switch modes from list to map.
Even how it is right now, this is pretty awesome. I'm looking for jobs and I plan to use this alot.
Secondly, I can't sign up for the weekly email. It says: Cannot POST /new
Thanks.