4 ms·

You could theoretically use the Streaming API to analyze tweets en masse from different sources/individuals. Coding in a bit of text analysis (length, blacklist

You could theoretically use the Streaming API to analyze tweets en masse from different sources/individuals. Coding in a bit of text analysis (length, blacklisting/whitelisting keywords, sentiment, etc.) would discriminate between superfluous tweets and the ones you care about. That'd would be a great way to automate the process of scrolling through crap every morning, but you'd need to build a web app to house the results and scroll through that instead. Of course, that's how I would do it as a third-party developer. Will Twitter ever support this? Doubt it. I don't think it's relevant to enough users to implement that kind of feature. Most people who use Twitter evidently enjoy reading what Justin Anniston and Jennifer Bieber eat for breakfast.

jelled 13y ago
This is essentially what we do with sports news at fancloud.com. By the time the tweets make it through our filters the end result is much more newsworthy then a raw twitter stream.
- kafkaesque 13y ago
  It seems like there'd be a higher demand for these type of specialised content filters. I'm a big NHL fan, so I'll check out fancloud.com.
danudey 13y ago
Not sure if it's what you're suggesting specifically, but if you analyzed common links within a span of time and analyzed them for keywords, you could associate other tweets from certain individuals. Tim Bray or Zed Shaw, as random examples, could tweet about an issue without providing a link to it (to infer context), but you could infer context by looking for relevant links from similar people. If the system analyzed the links you clicked on (via the t.co URL shortener Twitter operates), the links you posted, and the topics you tweeted about, they could build a profile of interest for you and highlight those tweets. Their current 'you should check this out' system on twitter.com and in the twitter client seems to be 'hey, here's some stuff everyone else likes', which (the last time I looked at it) was complete tripe about idiotic topics that 'the masses' care about, with nothing about the sort of tech news or political/economic interest that I tend to prefer, but I feel like a personalized feed is something that would be far more useful to me, and probably many other people.

jelled 13y ago

This is essentially what we do with sports news at fancloud.com. By the time the tweets make it through our filters the end result is much more newsworthy then a raw twitter stream.

kafkaesque 13y ago
It seems like there'd be a higher demand for these type of specialised content filters. I'm a big NHL fan, so I'll check out fancloud.com.

danudey 13y ago

Not sure if it's what you're suggesting specifically, but if you analyzed common links within a span of time and analyzed them for keywords, you could associate other tweets from certain individuals. Tim Bray or Zed Shaw, as random examples, could tweet about an issue without providing a link to it (to infer context), but you could infer context by looking for relevant links from similar people. If the system analyzed the links you clicked on (via the t.co URL shortener Twitter operates), the links you posted, and the topics you tweeted about, they could build a profile of interest for you and highlight those tweets. Their current 'you should check this out' system on twitter.com and in the twitter client seems to be 'hey, here's some stuff everyone else likes', which (the last time I looked at it) was complete tripe about idiotic topics that 'the masses' care about, with nothing about the sort of tech news or political/economic interest that I tend to prefer, but I feel like a personalized feed is something that would be far more useful to me, and probably many other people.