Google+

Monday, December 02, 2013

Do more with your tweets - Social reporting at ICT4Ag

In my previous post about tracking the online buzz generated around the ICT4Ag conference I’ve mentioned a couple of the tools that we used in the social reporting of the event. At the ICT4Ag conference, blogs, workspaces, Twitter and the other ‘usual suspects’ formed the backbone of our technology infrastructure.

However, besides these tested and tried solutions, this time around I wanted to experiment with some new applications around the edges of the process.

In this and the following posts I’m going to zoom in and focus on technology, presenting some new (to me at least) applications that I’ve used and what I’ve learnt - and what I would do different next time to further develop my own practice.

Specifically, this post looks at different tools to do (much) more with Twitter.

Conference structure, KM and Twitter #tags

The conference programme was organized around 3 main conference streams, a Plug and Play day and a Hackathon running in parallel to the main event. While it was a no brainer to decide the conference tag and the tags for Plug and Play and Hackathon, the design of the conference in streams and sessions presented some challenges in terms of effective use of #tags in Twitter.

On the one hand, we were expecting (as it indeed happened) quite some buzz on Twitter - and we needed to have ways to aggregate and disaggregate tweets around different conversations. On the other hand, CTA KM team was also interested in finding ways to archive and search the different tweets - around the conference #tag and specifically according to the different conference streams. Finally, we wanted to have the possibility to display specific twitter feeds on each tabs of our social dashboard.

So once we had decided a basic #tag vocabulary, with specific #tags for each session, I was faced with the challenge to find ways to:
  • Extract information from Twitter a Search - possibly a feed search results for each session #tag; 
  • Aggregate this info to display twitter feeds around each of the three conference streams; 
  • Find a way to archive and search these Tweets.

Display and aggregate Twitter search RSS feeds

This sounded complicated to achieve when I first looked at it. With the recent changes in Twitter API policy, users are no longer able to obtain any of the Twitter streams – search results, timelines of users, users’ favorites etc – in an RSS feed.

After spending some time researching and testing online what tools I could use to get a Twitter RSS feed, I found that some smart folks out there had the answer I was looking for and I was able to use some simple Google Scripts to set up a Twitter RSS feed for each session #tag search result.

Once I had created these 25 odd Twitter search feeds (one for each session #tag), I was then able to aggregate them into consolidated feeds for each of the 3 conference streams, using Yahoo Pipes. So for example, out of the Twitter search feeds for each of the 12 sessions in the Emerging Innovation stream, I was able to produce a single, aggregated Twitter feed.

Never miss a Tweet!

Once I’d solved this challenge, I was still presented with the need to find ways to archive tweets - as after a while they disappear from the search results. Plus, this Twitter archive had to be searchable, to meet the needs of the CTA team to browse and use this content in the future.

The solution here came from a brilliant post by Martin Hawksey, where he presents TAGS5.0 - a way to archive AND visualize tweets, automatically pulling results from a Twitter Search into a Google Spreadsheet.

This was indeed a great discovery and using it has proved to be of great interest and value

The full Twitter archive for the conference hashtag #ict4ag13 is available as a public Spreadsheet on Google Drive. You can also explore the interactive visualization of the conversations on Twitter, mapping replies, retweets and mentions, and the relations between different users. Finally, it gives you a complete Twitter archive that can be easily browsed and searched by keyword or by username.

Together with one main Spreadsheet for the conference hashtag, I’ve also repeated the process and created different Spreadsheets for:
For each of them, you can visualize the interaction and search the online archive of tweets (see on Summary tab, then Public web views).

Back up if you can

While the whole setup was quite easy, I did encounter one main technical problem. The script failed to archive several tweets for two days of the conference - maybe due to the very high volume of tweets that were generated.

Luckily I had also set up a backup for the #tag search in Hootsuite and I was the able to reconstruct the full database in the days after the event. But of course this is not ideal, as it require quite some time in moving data from one spreadsheet to the other.

Despite this, the end result is quite useful I believe, especially as it enables the preservation of the tweets for future reference and research, and to compare different events in terms of online conversation and engagement on Twitter.

I’d also like to see if the visualization of the Twitter conversation can be improved and provide a clear picture of the connections.

Any expert out there that would like to give it a try and improve the visualizations from the complete dataset?

 -