Show HN - A tweetmeme of its own for every city | Hacker News Reader

Show HN - A tweetmeme of its own for every city

frrole.com

25 pointsgayakwad20 comments

20 cities is hardly "every city". I tried 'frrole.com/mycity', expecting a page to be generated on the fly, but no such luck. Should be possible though, and then you could legitimately state that you offer results for every city.

It would also be a nice touch to use geolocation to automagically generate a city (or region) page. Right now, it presented me with the page for Mumbai, while I'm in Europe. Out of the available options on the site, London is closest to me.

Agree, should have said "Building a tweetmeme of its own for every city". This is the first release, we are taking the no. of cities to 50+ with the next release in 6-8 weeks.

The focus till now has been on building algorithms that are kickass enough to filter junk out and keep only useful stuff in. Still WIP I know, but what do you think about that. Take http://frrole.com/san+francisco for example.

The suggestion is that the user should be able to enter any city, not just use the preset ones.

Thanks, that is how it is intended to be in the longer term. But going to hit Twitter's rate limits very soon and access to full firehose is way too costly for a bootstrapped startup.

Hope to scale to 50+ cities by ourselves, validate traction with users from those 50 cities and then look at really big scaling, if it is getting some love:)

And yes, about geo location, that is built in. But if you match neither the city nor the country, then it takes you to a global default.

just a mashup like Twits in London: http://tlondon.linkstore.ru

Also, relevant tweets for New York and San Francisco are already pretty easy to guess for anyone following technology news or national news. Chicago, Atlanta and Miami had much more diverse results and organic feeling results - provided tweets that felt characteristic of those cities.

There is a value in providing a "social news paper for a city" but that's not what OP is doing now. "Swarm" intelligence is superior to an individual. Filtering news articles by number of re-tweets sets higher threshold for signal-to-noise ratio, however, 140 characters hardly constitute a "news article". Below, my suggestion for a data mining web application that is focused on delivering a news paper customized for a city with high signal-to-noise ratio news articles.

Input: X = country, Y = city, tweets from twitter

For each X scan last N million tweets that have a link to a web-site, output M most linked domains (Facebook.com, techcrunch.com, cnn.com, wsj.com, etc) - this gives you a base for news feeds customized for every country

For each Y in X, scan last n thousand tweets that have a link to a web-page, output m most linked web-pages that are hosted on M domains for Y country - this gives you the "social news paper for Y city in X country"

Parameters M and N prevent spamming and shape long term trends. Parameters n and m give you short term trends, which in essence are the "swarm" intelligence about which articles are worth reading for a given city in a given country. The value here is a time saver of what is worth reading to keep up with your community.

Note to self: could be a data mining web app

Really great inputs @zeratul, we will certainly analyze your technique :)

I agree some and disagree some. Regarding M and N, I don't think they can control spam any more than m and n, given that M and N are going to be super-sets of root domains of m MULTIPLIEDBY Y{1 to j} and no of tweets of n MULTIPLIEDBY Y{1 to j} respectively, where j is the no of Y in each X.

Where I totally agree is the "Swarm" intelligence is superior to an individual philosophy. The same thing is dictating our thought process as well. Here: http://frrole.com/about-us

Where I again think differently is the 140 chars vs. the linked article line of thought. Grant that maybe we should call this "social headlines for your city" instead of "social newspaper..." ( 140 chars make for perfectly good headlines) but I believe that the information overload today is increasing drastically, attention span therefore (and otherwise) is decreasing significantly; so much so that more content, specially textual, is not the answer. Less content in fact is the answer, but chosen intelligently enough so that it does not leave things out.

Where I definitely agree with gayakwad is that these are fantastic inputs, the kind of inputs one definitely expects on HN. See from your profile that your areas of interest too are those that we are working on (Frrole is a lot, LOT more than no of retweets btw), maybe we should talk offline?

This is interesting, but looking at the NYC one, your "Xpert Tweeters" really distort the idea of "local".

I'm seeing tons of national news from the WSJ, the NBA, and a score of other media outlets that are based in New York but not covering New York (aside from incidentals, e.g., that Wall Street is a street in Manhattan).

Also, as you expand, it might be worth while to drill down NYC by borough. Having a Brooklyn, Manhattan, Queens, and even Hoboken NJ page would probably be worthwhile both for your users, and for you.

Yup, have heard that one a couple of times already, we are going to take care of that one VERY soon. Plan to aggregate tweets at country/global level too going forward, so have to do it in a way that we remove them from the city but not lose them totally.

Check out some of the deals, events etc, though. Good "democratized" info (I hope:)), including things that would never make it through standard aggregators.

Yup, some Xpert tweeter they talk about whole world (generic) but keep there location set to some particular city, which mess up data a bit, we have already found a solution for this problem, changes will be rolled out soon :)

looks great.. way to go.. ranking tweets based on re-tweets is the best way to filter out twitter noise.. a suggestion would be to keep it local to begin with.. india specific.. else it'll again somewhat become noise at international level..

I created something on same lines but much much smaller (and minimal) : www.lessvu.com

primarily to track certain active communities in india.. (presently works for journalists only)

thanks. Retweets is just one rule. Pretty much like google uses back links as an indicator, but has 200-300 other rules, we use retweets as an indicator but have 100+ rules probably. or around. These rules are spread across metadata analysis (language based, response type, location based filtering etc), statistical analysis(no of retweets, recency based scoring etc) and linguistic/textual analysis (syntactical construction, use of spammy words, characters etc)

Please expand to multiple cities, more content and Mobile. As it is first version, Still looking good.

Best wish

Looks awesome !, really interesting data in my city Bangalore Are you ppl looking only at no of Retweets ?

Good to know that you liked it, Certainly we consider tweet count but that's not the only thing, we have put lots of logic for analyzing tweet's nature (textual/NLP), relevance (by city and category) and most importantly we have worked on cutting noise.

Vancouver before Montreal? Does not make any sense :-)

Montreal is up in the next release.

Reply on news.ycombinator.com